admin/LithosAnanake
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: reorganise -- current reference gathered, finished work archived, superseded and dead ideas marked as such

Browse Source 
docs/CLAUDE.md Phase 2, 2026-10-14.  All moves by git mv; every archived
file carries a header saying what it is, why, and where to look instead.

- docs/reference/ (new): roadmap and versioning policy, M7.1, word-ACL
  design, kernel args, hosted acceptance test, dictionary, primitives,
  ontology, L8 table -- previously spread over docs/, docs/lithosananke/
  and docs/03-architecture/.
- docs/working/archive/historical/: finished work (session plan, test-gap
  audit, birthing plan, fix write-ups, the January architecture draft).
- docs/working/archive/superseded/: the Hermes-VM Tripod docs and July
  Hermes designs (-> FABRIC-3.5/3.6), the StarForth roadmap, every working
  doc already flagged superseded or obsolete.
- docs/working/archive/dead/: the hal_* target contract (never adopted),
  L4Re, metrics-and-knobs.
- Papers and proposals into docs/working/papers/.
- References rewritten across the repo (not logs/, not docs/patent/);
  CLAUDE.md's documentation section rewritten.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
rajames
2026-10-14 16:03:56 -04:00
co-authored by Claude Opus 5.5
parent 1a4b163861
commit 7aed588f78
116 changed files with 305 additions and 175 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/working/archive/superseded/2025-12-14-starforth-roadmap.md
+666
View File
@@ -0,0 +1,666 @@
> **SUPERSEDED (docs triage 2026-10-14).** The StarForth-era VM -> kernel -> OS -> FPGA plan of 2025-12-14, flagged obsolete 2026-08-26.
> Instead: docs/reference/ROADMAP.md (the LithosAnanke roadmap and versioning policy). Moved here from `docs/ROADMAP.md`.
# StarForth Roadmap: VM → Kernel → OS → FPGA
> **OBSOLETE (flagged 2026-08-26, Captain Bob's call — this is more than "stale," the
> architecture, branch topology, and terminology it describes no longer exist).** This is a
> StarForth-era plan dated 2025-12-14, written before the repo split, before
> Tripod/Stadium/word-level ACL existed, and before "Phase 1 Starting (HAL)" had any of the
> meaning it has now. Kept for historical record only. For current status, see
> `docs/reference/ROADMAP.md` (repo-specific, LithosAnanke roadmap) and
> `FABRIC-0.md`/`FABRIC-1.md`/`FABRIC-2.md` (design history and current work) — not this
> document.
**Version**: 1.0
**Date**: 2025-12-14
**Timeline**: 2025-2028 (3.5 years)
**Status**: Phase 0 Complete (VM), Phase 1 Starting (HAL)
---
## VISION
**From**: Adaptive FORTH VM running on hosted OS (Linux)
**To**: Self-hosting operating system running on custom FPGA hardware
```
2025 (NOW) 2026 2027 2028
│ │ │ │
V V V V
┌─────────┐ ┌──────────┐ ┌────────────┐ ┌──────────┐
│StarForth│ → │StarKernel│ → │ StarshipOS │ → │ FPGA │
│ VM │ │(Bare Metal) │(Self-hosting) │ Hardware │
└─────────┘ └──────────┘ └────────────┘ └──────────┘
COMPLETE 6-8 months 12-18 months 12-18 months
```
**Key Principle**: Each phase builds on the previous. No skipping steps.
---
## CURRENT STATE (Phase 0: COMPLETE ✅)
### What We Have
**StarForth VM** - Production-ready adaptive FORTH interpreter:
- ✅ FORTH-79 compliant (780+ tests passing)
- ✅ Adaptive runtime (7 feedback loops, 0% algorithmic variance)
- ✅ Platforms: Linux (HISTORICAL: also L4Re/Fiasco.OC through mid-2026; removed as an active target)
- ✅ Performance: 100M iterations/sec, profile-guided optimization
- ✅ Quality: 9.2/10 codebase rating, strict ANSI C99
- ✅ Validation: Peer-reviewed, 90-run statistical proof
- ✅ Patent pending: Adaptive runtime mechanisms
**Documentation**:
- ✅ 121-page academic paper
- ✅ Scientific defense documents (9 files)
- ✅ Ontology & formal definitions
- ✅ HAL architecture (documented, not implemented)
**Infrastructure**:
- ✅ Make-based build system (40+ targets)
- ✅ DoE (Design of Experiments) framework
- ✅ Reproducibility guarantees
- ✅ CI-ready test suite
### What's Missing
**See**: [docs/GAP_ANALYSIS.md](docs/GAP_ANALYSIS.md) for complete list (37 gaps identified)
**Critical Blockers**:
1. HAL abstraction layer (8 components) - **Blocks everything**
2. StarKernel implementation (13 components)
3. StarshipOS subsystems (12 components)
4. FPGA feasibility & design (19 components)
---
## PHASE 1: HAL IMPLEMENTATION (Months 1-2)
**Goal**: Make StarForth platform-agnostic via Hardware Abstraction Layer
**Why**: HAL is the foundation for all future work. Without HAL, cannot build kernel or port to FPGA.
### Deliverables
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 1.1 | Define HAL interfaces (6 headers) | 500 | 1 week | 📋 Planned |
| 1.2 | Implement Linux HAL | 1500 | 2 weeks | 📋 Planned |
| 1.3 | Migrate VM core to HAL | 800 | 2 weeks | 📋 Planned |
| 1.4 | Migrate physics subsystems | 300 | 1 week | 📋 Planned |
| 1.5 | Validate determinism (DoE) | Testing | 2 days | 📋 Planned |
**Total**: ~3,100 LOC, 6-7 weeks
### Success Criteria
- ✅ All HAL interface headers exist (`hal_time.h`, `hal_interrupt.h`, etc.)
- ✅ Linux HAL implementation complete
- ✅ VM core has zero direct platform dependencies
- ✅ All 780+ tests pass on HAL
- ✅ 0% algorithmic variance maintained
- ✅ No measurable performance regression
### Milestones
**Week 2**: HAL headers defined, Linux implementation started
**Week 4**: Linux HAL complete, VM migration started
**Week 6**: All tests passing, determinism validated
**Blocker Risk**: Low (well-understood refactoring)
---
## PHASE 2: STARKERNEL (Months 3-5)
**Goal**: Boot StarForth on bare metal (UEFI → kernel → "ok" prompt)
**Why**: Kernel enables true OS development and demonstrates platform independence.
### Architecture
```
┌─────────────────────────────────────────────┐
│ UEFI Firmware │
│ • Memory map, ACPI, framebuffer │
└────────────────┬────────────────────────────┘
│ ExitBootServices()
▼
┌─────────────────────────────────────────────┐
│ StarKernel Boot (boot/uefi_loader.c) │
│ • Collect BootInfo │
│ • Jump to kernel_main() │
└────────────────┬────────────────────────────┘
▼
┌─────────────────────────────────────────────┐
│ StarKernel HAL (platform/kernel/) │
│ • CPU init (GDT, IDT, interrupts) │
│ • Memory (PMM, VMM, heap) │
│ • Time (TSC, HPET, APIC timer) │
│ • Console (UART + framebuffer) │
└────────────────┬────────────────────────────┘
▼
┌─────────────────────────────────────────────┐
│ StarForth VM │
│ • vm_create() │
│ • Physics subsystems │
│ • REPL → "ok" prompt │
└─────────────────────────────────────────────┘
```
### Deliverables
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 2.1 | UEFI boot loader | 800 | 2 weeks | 📋 Planned |
| 2.2 | Physical memory manager (PMM) | 1200 | 2 weeks | 📋 Planned |
| 2.3 | Virtual memory manager (VMM) | 1500 | 2 weeks | 📋 Planned |
| 2.4 | Kernel heap (kmalloc) | 800 | 1 week | 📋 Planned |
| 2.5 | CPU initialization (GDT/IDT) | 1000 | 1 week | 📋 Planned |
| 2.6 | UART driver (serial console) | 600 | 1 week | 📋 Planned |
| 2.7 | Time subsystem (TSC/HPET) | 1000 | 2 weeks | 📋 Planned |
| 2.8 | APIC interrupt controller | 1200 | 2 weeks | 📋 Planned |
| 2.9 | Framebuffer driver | 800 | 1 week | 📋 Planned |
| 2.10 | ACPI parsing | 600 | 1 week | 📋 Planned |
| 2.11 | HAL implementation (kernel) | 1000 | 2 weeks | 📋 Planned |
| 2.12 | Build system integration | 500 | 1 week | 📋 Planned |
**Total**: ~11,000 LOC, 12-14 weeks
### Success Criteria
- ✅ UEFI boots on QEMU + OVMF
- ✅ Serial console prints "StarKernel booting..."
- ✅ PMM/VMM allocate memory correctly
- ✅ Kernel heap works (kmalloc/kfree)
- ✅ Interrupts and timers functional
- ✅ VM starts and prints "ok" prompt
- ✅ Can execute simple FORTH code (`1 2 + .` → `3`)
- ✅ Physics subsystems operational on bare metal
### Milestones
**Month 3**: UEFI boot + serial output working
**Month 4**: Memory managers complete, VM starts
**Month 5**: Full HAL working, "ok" prompt on QEMU
**Blocker Risk**: Medium (complex low-level code, but well-documented)
### Testing Strategy
**Development**: QEMU with OVMF (UEFI firmware)
**Validation**: Physical hardware (Intel NUC or similar)
**Automation**: QEMU headless testing in CI
---
## PHASE 3: STARSHIPOS (Months 6-18)
**Goal**: Self-hosting operating system (compiles itself, runs applications)
**Why**: Demonstrates complete independence from Linux/POSIX, enables FORTH-native ecosystem.
### Architecture
```
┌─────────────────────────────────────────────────────────┐
│ StarshipOS Layers │
├─────────────────────────────────────────────────────────┤
│ Userland │
│ • Shell, editors, compilers, tools │
├─────────────────────────────────────────────────────────┤
│ System Services │
│ • Filesystem (FAT32 or custom) │
│ • Networking (TCP/IP stack) │
│ • Process manager, scheduler │
├─────────────────────────────────────────────────────────┤
│ Drivers │
│ • Storage (AHCI, NVMe) │
│ • Network (VirtIO, E1000) │
│ • Block/char device abstraction │
├─────────────────────────────────────────────────────────┤
│ StarKernel (from Phase 2) │
│ • CPU, memory, time, interrupts │
└─────────────────────────────────────────────────────────┘
```
### Deliverables (Staged)
#### Stage 3A: Storage (Months 6-9)
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 3.1 | AHCI driver | 2500 | 1 month | 📋 Planned |
| 3.2 | NVMe driver | 2000 | 1 month | 📋 Planned |
| 3.3 | Block device abstraction | 1500 | 2 weeks | 📋 Planned |
| 3.4 | FAT32 filesystem | 3000 | 1 month | 📋 Planned |
| 3.5 | VFS layer | 2000 | 2 weeks | 📋 Planned |
**Subtotal**: ~11,000 LOC, 3-4 months
**Success Criteria**:
- ✅ Read/write files to disk
- ✅ Boot from hard drive (not just RAM)
- ✅ Persistent FORTH definitions
---
#### Stage 3B: Networking (Months 10-12)
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 3.6 | VirtIO-net driver | 2000 | 1 month | 📋 Planned |
| 3.7 | E1000 driver | 1500 | 2 weeks | 📋 Planned |
| 3.8 | lwIP TCP/IP stack (port) | 8000 | 2 months | 📋 Planned |
| 3.9 | Socket API (FORTH-native) | 1500 | 2 weeks | 📋 Planned |
**Subtotal**: ~13,000 LOC, 3-4 months
**Success Criteria**:
- ✅ Send/receive packets
- ✅ TCP connections work
- ✅ Simple HTTP server in FORTH
---
#### Stage 3C: Process Model (Months 13-15)
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 3.10 | FORTH task abstraction | 2000 | 1 month | 📋 Planned |
| 3.11 | Scheduler (round-robin) | 1500 | 2 weeks | 📋 Planned |
| 3.12 | IPC mechanism | 2500 | 1 month | 📋 Planned |
| 3.13 | System call interface | 1000 | 2 weeks | 📋 Planned |
**Subtotal**: ~7,000 LOC, 2-3 months
**Success Criteria**:
- ✅ Multiple concurrent tasks
- ✅ Task switching works
- ✅ Tasks can communicate (send/receive)
---
#### Stage 3D: Self-Hosting (Months 16-18)
| # | Component | LOC | Duration | Status |
|---|-----------|-----|----------|--------|
| 3.14 | FORTH compiler (meta-compilation) | 3000 | 1 month | 📋 Planned |
| 3.15 | Assembler (FORTH-hosted) | 2000 | 2 weeks | 📋 Planned |
| 3.16 | Linker/loader | 1500 | 2 weeks | 📋 Planned |
| 3.17 | Shell & core utilities | 2500 | 1 month | 📋 Planned |
| 3.18 | Build system (FORTH-based) | 1000 | 2 weeks | 📋 Planned |
**Subtotal**: ~10,000 LOC, 2-3 months
**Success Criteria**:
- ✅ StarshipOS compiles itself
- ✅ No dependency on Linux toolchain
- ✅ Can build applications entirely in FORTH
---
**Phase 3 Total**: ~41,000 LOC, 12-15 months
### Milestones
**Month 9**: Persistent storage working, boot from disk
**Month 12**: Network stack operational, HTTP demo
**Month 15**: Multitasking working, IPC functional
**Month 18**: Self-hosting achieved, OS compiles itself
**Blocker Risk**: High (large scope, complex subsystems)
**De-risking Strategy**: Port existing code where possible (lwIP for TCP/IP, FAT32 reference)
---
## PHASE 4: FPGA FEASIBILITY (Months 6-7)
**Goal**: Determine if FPGA implementation is worth pursuing
**Why**: FPGA is a massive investment (12-18 months). Need go/no-go decision before committing.
### Feasibility Study
| # | Task | Duration | Deliverable |
|---|------|----------|-------------|
| 4.1 | Survey existing FORTH FPGA cores | 1 week | Comparison table |
| 4.2 | Estimate resource usage (LUTs, BRAMs) | 3 days | Resource budget |
| 4.3 | Select target FPGA | 3 days | FPGA selection rationale |
| 4.4 | Prototype stack machine (Verilog) | 2 weeks | Minimal stack machine |
| 4.5 | Synthesize & test on FPGA board | 1 week | Synthesis report |
| 4.6 | Performance analysis | 3 days | Speed vs. software |
| 4.7 | Cost/benefit analysis | 2 days | Go/no-go recommendation |
**Total**: 5-6 weeks
### Decision Criteria
**GO if**:
- ✅ Fits in affordable FPGA (~$200-500 board)
- ✅ Performance ≥ 50 MHz clock (faster than software?)
- ✅ Adaptive runtime feasible in hardware (critical!)
- ✅ Development time justified by benefits
**NO-GO if**:
- ❌ Too large for affordable FPGAs
- ❌ Slower than software implementation
- ❌ Adaptive runtime not synthesizable
- ❌ >18 months to implement (opportunity cost too high)
### Prototype Spec
**Minimal Stack Machine** (proof of concept):
- Data stack (16 entries × 64-bit)
- Return stack (16 entries × 64-bit)
- Basic operations: +, -, *, /, DUP, DROP, SWAP
- Dictionary ROM (256 entries)
- Simple UART I/O
**Target**: Prove concept works in hardware, get resource usage numbers
---
## PHASE 5: FPGA IMPLEMENTATION (Months 8-24)
**Status**: CONDITIONAL (only if Phase 4 says GO)
**Goal**: Hardware FORTH processor on FPGA
### Architecture
```
┌────────────────────────────────────────────────┐
│ FPGA Top Module │
├────────────────────────────────────────────────┤
│ ┌──────────────┐ ┌──────────────┐ │
│ │ Stack Machine│ │ Memory │ │
│ │ • Data stack│ │ Controller │ │
│ │ • Return stk│ │ • Dict ROM │ │
│ │ • ALU │ │ • RAM │ │
│ └──────────────┘ └──────────────┘ │
├────────────────────────────────────────────────┤
│ ┌──────────────┐ ┌──────────────┐ │
│ │ UART │ │ Timers │ │
│ │ IP Core │ │ Counters │ │
│ └──────────────┘ └──────────────┘ │
├────────────────────────────────────────────────┤
│ ┌──────────────────────────────────────────┐ │
│ │ Adaptive Runtime (if feasible) │ │
│ │ • Execution counters │ │
│ │ • Hot-word cache │ │
│ │ • Hardware decay (?) │ │
│ └──────────────────────────────────────────┘ │
└────────────────────────────────────────────────┘
```
### Deliverables
#### Stage 5A: Core Implementation (Months 8-13)
| # | Component | LOC (HDL) | Duration | Status |
|---|-----------|-----------|----------|--------|
| 5.1 | Data stack (16 deep) | 500 | 2 weeks | 📋 Conditional |
| 5.2 | Return stack (16 deep) | 500 | 2 weeks | 📋 Conditional |
| 5.3 | ALU (arithmetic/logic ops) | 2000 | 1 month | 📋 Conditional |
| 5.4 | Dictionary memory (ROM) | 1000 | 2 weeks | 📋 Conditional |
| 5.5 | RAM controller | 1500 | 3 weeks | 📋 Conditional |
| 5.6 | Instruction decoder | 2000 | 1 month | 📋 Conditional |
| 5.7 | Control FSM | 1500 | 3 weeks | 📋 Conditional |
**Subtotal**: ~9,000 lines Verilog, 4-5 months
---
#### Stage 5B: I/O & Peripherals (Months 14-16)
| # | Component | LOC (HDL) | Duration | Status |
|---|-----------|-----------|----------|--------|
| 5.8 | UART transmitter/receiver | 800 | 2 weeks | 📋 Conditional |
| 5.9 | Timer/counter IP cores | 600 | 1 week | 📋 Conditional |
| 5.10 | Interrupt controller | 1000 | 2 weeks | 📋 Conditional |
| 5.11 | Memory-mapped I/O bus | 1200 | 3 weeks | 📋 Conditional |
**Subtotal**: ~3,600 lines Verilog, 2-3 months
---
#### Stage 5C: Adaptive Runtime (Months 17-20)
**CRITICAL UNCERTAINTY**: Can we synthesize this?
| # | Component | LOC (HDL) | Duration | Status |
|---|-----------|-----------|----------|--------|
| 5.12 | Execution frequency counters | 1500 | 1 month | 📋 Conditional |
| 5.13 | Hot-word cache (top-K) | 2000 | 1 month | 📋 Conditional |
| 5.14 | Hardware decay logic | 1000 | 2 weeks | 📋 Conditional |
| 5.15 | Transition tracking | 1500 | 3 weeks | 📋 Conditional |
**Subtotal**: ~6,000 lines Verilog, 3-4 months
**Fallback**: If adaptive runtime doesn't synthesize well, implement in software on soft-core CPU (MicroBlaze/NIOS II) alongside FORTH core.
---
#### Stage 5D: Validation & Optimization (Months 21-24)
| # | Task | Duration | Deliverable |
|---|------|----------|-------------|
| 5.16 | Testbench (all modules) | 2 months | 100% coverage |
| 5.17 | Formal verification (critical paths) | 1 month | Proofs |
| 5.18 | Synthesis optimization | 1 month | Timing closure |
| 5.19 | Hardware-in-loop testing | 2 weeks | Physical validation |
| 5.20 | Performance tuning | 2 weeks | Final optimizations |
**Subtotal**: 4-5 months
---
**Phase 5 Total**: ~18,600 lines Verilog + testbenches, 15-18 months
### Target FPGAs (Options)
| FPGA | Cost | Resources | Speed | Notes |
|------|------|-----------|-------|-------|
| **Xilinx Zynq-7000** | $300 | 85K LUTs, ARM core | 200 MHz | **Recommended**: Hybrid ARM+FPGA |
| Intel Cyclone V | $250 | 77K LEs | 150 MHz | Good alternative |
| Lattice iCE40 UP5K | $50 | 5K LUTs | 50 MHz | Budget option, slower |
| Xilinx Artix-7 | $200 | 33K LUTs | 200 MHz | Pure FPGA, no ARM |
**Recommendation**: Zynq-7000 (ARM + FPGA fabric)
- Run StarshipOS on ARM
- Offload FORTH core to FPGA fabric
- Best of both worlds
### Success Criteria
- ✅ FPGA boots, prints "ok" via UART
- ✅ Can execute FORTH code
- ✅ Performance ≥ software on same workload
- ✅ Adaptive runtime functional (or gracefully degraded)
- ✅ 780+ tests pass on hardware
- ✅ 0% algorithmic variance maintained (if adaptive works)
### Milestones
**Month 13**: Core stack machine working in simulation
**Month 16**: I/O functional, boots on real FPGA
**Month 20**: Adaptive runtime implemented (or fallback decided)
**Month 24**: Full validation, production-ready FPGA core
**Blocker Risk**: VERY HIGH
- HDL expertise required
- Timing closure can be unpredictable
- Adaptive runtime may not be synthesizable
- Resource constraints on affordable FPGAs
---
## INTEGRATION TIMELINE
### Parallel Paths
**Months 1-5**: Sequential (HAL → Kernel)
**Months 6-18**: **Parallel** (OS + FPGA feasibility)
**Months 8-24**: **Conditional** (FPGA impl if feasible)
```
2025 2026 2027 2028
│ │ │ │
V V V V
Phase 1 Phase 2 Phase 3 Phase 5
HAL Kernel OS FPGA
(2m) (3m) ┌──────────────────(12m)────────────┐ (16m)
│ │ │ │
│ Phase 4: Feasibility (2m) │ │
│ └─────┬─────────────────────────────┘ │
│ │ │
└────────────────┴──GO/NO-GO Decision───────────────┘
│
If NO-GO: Focus on OS only
If GO: Parallel OS + FPGA work
```
### Resource Allocation (Solo Development)
**Months 1-5** (HAL + Kernel): 100% focus
**Months 6-7** (Feasibility): 80% OS, 20% FPGA study
**Months 8-18** (Parallel): 70% OS, 30% FPGA (if GO)
**Months 19-24** (FPGA finish): 50% OS, 50% FPGA
**Community Contributions**: Welcomed at all stages (see CONTRIBUTING.md)
---
## RISK MITIGATION
### High-Risk Items
| Risk | Probability | Impact | Mitigation |
|------|-------------|--------|------------|
| **HAL breaks determinism** | Low | Critical | Extensive testing, DoE validation |
| **Kernel never boots** | Medium | Critical | Incremental dev, QEMU testing |
| **FPGA too slow** | Medium | High | Feasibility study gates investment |
| **Adaptive runtime unsynthesizable** | High | High | Software fallback on ARM core |
| **Project too large (solo)** | High | Medium | Community outreach, phased delivery |
| **FPGA cost too high** | Low | Medium | Target affordable boards ($200-500) |
### Fallback Plans
**If HAL migration fails**: Continue hosted-only, defer kernel
**If OS scope too large**: Deliver minimal components (storage + network only)
**If FPGA infeasible**: Software-only path is still valuable
---
## DELIVERABLES SUMMARY
### By End of 2025 (Months 1-2)
- ✅ HAL implementation complete
- ✅ StarForth platform-agnostic
- ✅ All tests passing on HAL
### By Mid-2026 (Months 3-5)
- ✅ StarKernel boots to "ok" on QEMU
- ✅ Physics subsystems work on bare metal
### By End of 2026 (Months 6-12)
- ✅ Storage subsystem operational
- ✅ Network stack functional
- ✅ FPGA feasibility determined (GO/NO-GO)
### By Mid-2027 (Months 13-18)
- ✅ Process model working
- ✅ StarshipOS self-hosting
- ✅ (If FPGA GO): Prototype FPGA core functional
### By End of 2027 (Months 19-24)
- ✅ StarshipOS production-ready
- ✅ (If FPGA GO): FPGA core validated
### By 2028 (Future)
- 🎯 Ecosystem growth (applications, libraries)
- 🎯 FPGA optimizations & variants
- 🎯 Community-driven features
---
## SUCCESS METRICS
### Technical Metrics
| Metric | Target | Measurement |
|--------|--------|-------------|
| **HAL overhead** | <5% | Benchmark comparison |
| **Kernel boot time** | <1 second | QEMU timing |
| **OS compile speed** | <10 min | Self-hosting build |
| **FPGA clock speed** | ≥50 MHz | Synthesis report |
| **Test pass rate** | 100% (780+) | All platforms |
| **Determinism** | 0% CV | DoE validation |
### Project Health Metrics
| Metric | Target | Current |
|--------|--------|---------|
| **Code quality** | ≥9.0/10 | 9.2/10 ✅ |
| **Test coverage** | ≥80% | ~90% ✅ |
| **Documentation** | Complete | Strong ✅ |
| **Community** | ≥10 contributors | 1 (solo) |
| **GitHub stars** | ≥500 | TBD |
---
## COMMUNITY & ECOSYSTEM
### How to Contribute
See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
**High-value contributions**:
1. HAL implementations (new platforms)
2. Kernel drivers (AHCI, NVMe, network)
3. FPGA expertise (Verilog, synthesis)
4. Testing & validation
5. Documentation improvements
### Seeking Collaborators
**Needed expertise**:
- Kernel development (UEFI, memory management)
- FPGA design (Verilog/VHDL)
- OS design (filesystems, networking)
- Formal verification (HDL proofs)
**Contact**: rajames440@gmail.com (Robert A. James)
---
## CONCLUSION
**This roadmap is ambitious but achievable.**
**Key Success Factors**:
1. ✅ Strong foundation (Phase 0 complete)
2. ✅ Clear milestones (phased delivery)
3. ✅ Risk mitigation (fallback plans)
4. ⚠️ Community growth (critical for timeline)
5. ⚠️ FPGA decision (feasibility gates investment)
**Realistic Timeline**:
- Solo: 3.5 years (42 months)
- With 3-5 contributors: 2 years (24 months)
- With 10+ contributors: 1.5 years (18 months)
**The vision is clear. The path is defined. Now we build.**
---
**Next Steps**:
1. Read [docs/GAP_ANALYSIS.md](docs/GAP_ANALYSIS.md) for detailed gaps
2. Start Phase 1: HAL implementation
3. Join the project (see CONTRIBUTING.md)
**License**: See ./LICENSE
docs/working/archive/superseded/2026-07-08-hermes-message-block-storage-design.md
+397
View File
@@ -0,0 +1,397 @@
> **SUPERSEDED (docs triage 2026-10-14).** Storage design for the Hermes VM's messages and channels; the Hermes VM no longer exists -- messages are kernel-Hermes's.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/working/architecture/HERMES-MESSAGE-BLOCK-STORAGE-DESIGN-20260708.md`.
# Hermes Message/Channel Block-Backed Storage — Design Doc
**Date:** 2026-07-08 (rev 2026-07-09d)
**Branch:** `lithosananke`
**Status:** Design only. Not implemented. This is the dependency flagged
in `HERMES-MESSAGE-CHANNEL-PHYSICS-DESIGN-20260708.md`'s Status header —
that doc's heat-window sizing is provisional until this one lands.
**Author:** Captain Bob / Claude Code
**Revision history** (kept honest per this repo's convention — see
`VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md` for the pattern):
- **rev a** (original): proposed Hermes duplicating Artemis's free-map
allocator shape verbatim, in isolation, as a standalone Hermes-only
change. Wrong framing — see rev b.
- **rev b**: corrected rev a toward a two-VM "shared Physical BAM /
Logical BAM effort" — Hermes and Artemis jointly building toward a
common logical layer. **Also wrong.** Captain Bob's direct correction:
*"no direct coupling between Hermes and Artemis... Hermes knows NOTHING
about Artemis or a bam of any kind."* Hermes's only inter-VM capability
is messaging (send/receive on the common channel, request → ACK/NACK)
— never reaching into another VM's storage, and never a BAM concept
shared across VMs.
- **rev c** (this revision): the actual confirmed architecture, reached
after grounding in how Artemis concretely lays out its own disk today.
**Every VM manages its own storage independently** — own RAM blocks,
then own ramdrive blocks, then its own dedicated appended virtio-blk
disk, exactly mirroring Artemis's already-proven pattern but with zero
knowledge of or coupling to Artemis (or to any other VM). No shared
allocator, no coordinator VM. "Physical BAM" and "Logical BAM" are
per-VM concepts — each VM eventually has its own of each — not a
cross-VM split. Confirmed verbatim: *"each VM uses its ram blocks and
ramdisk blocks then appends the virtual drive"* / *"yes, that's
correct"* / *"it's going to look like PBAM and LBAM are the same right
now. this will change when we start figuring out how to use and shape
the tripod."*
- **rev d** (this revision, 2026-07-09): rev c is corrected. A full
Tripod-docs audit (reading `.claude/TRIPOD.md`/`HERMES.md`/`ARTEMIS.md`
in full, which should have happened before rev a was ever written)
found that rev c's "every VM independently, no coordinator" picture is
not how the Tripod is actually specified. `TRIPOD.md`'s dispatch table
is explicit: `Block I/O → Artemis`, reached via `VM-EXEC` naming the
VM by capability, not heat or convenience. `ARTEMIS.md` is direct:
*"Artemis is memory. She owns the external attached virtual disk image
and everything that lives on it."* Hermes does not grow its own disk.
**Corrected architecture: Artemis is sole storage authority for the
fleet.** Hermes's message-block persistence need is satisfied by
*asking* Artemis, not by building parallel storage machinery — this
supersedes rev c's "own dedicated appended virtio-blk disk" tier for
Hermes entirely (see "Storage tiering," corrected below).
This does **not** reopen rev b's rejection of direct coupling. *"There
is to be no tight coupling between ANY VM"* — reaffirmed by Captain Bob
in the same breath as naming Artemis storage authority — and it isn't
in tension with that at all: `TRIPOD.md`'s capability-based dispatch
(`VM-EXEC` by name, request/ACK-NACK) is already the Tripod's normal,
loose, message-mediated interaction model. Hermes asking Artemis to
store a message's payload is the same shape as any other cross-VM
request — it was only ever a contradiction under the mistaken
assumption that "Artemis as authority" meant Hermes reaching into
Artemis's BAM or address space directly. It doesn't. Hermes never
learns a physical LBN, never touches `FM-*`/`BLK-ALLOC`, never knows a
zone exists. See "The logical-reference gap, resolved" below for the
concrete mechanism.
- **rev e** (this revision, 2026-07-09): rev d overcorrected. Captain
Bob's direct correction: *"ALL VMs regardless have ram backed blocks
zero to 2047, ramdrive from 2048 to its end. that's a VMs state and
identity hard requirement."* Rev c's tiers 1–2 (own RAM 0–2047, own
ramdrive 2048+, per VM) were **right** — rev d was wrong to fold them
into "Artemis owns everything." They're not storage Hermes borrows or
requests; they're part of what makes a VM a VM at all, same category as
its own dictionary and stacks. Corroborating evidence, not invented:
`capsule_birth_baby()`'s own comment — *"PERSONALITY: per-VM block
storage is M9 scope; no-op until then"* — names this as a real, planned
milestone, just not built yet.
**What actually changes under Artemis-as-storage-authority is narrower
than rev d claimed: only the tier beyond a VM's own ramdrive.** Rev c's
tier 3 ("own dedicated appended virtio-blk disk") is the one thing that
doesn't survive — no VM other than Artemis appends a persistent disk.
Once Hermes's own ramdrive is exhausted, *that's* when it asks Artemis,
not before. Sections below are corrected to reflect three tiers again:
own RAM, own ramdrive, then (not own disk, but) ask Artemis.
---
## Execution rules
Binding for all implementation work under this doc, matching the
discipline `VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md` established and
Captain Bob's explicit intent for this specific piece of work — this is
as much a Tripod discovery exercise as it is a storage redesign:
1. **Small units.** One new word, one call-site migration, one allocator
primitive at a time — not the whole redesign in one pass.
2. **Clean + make at checkpoints.** `make -f Makefile.starkernel clean`
then build after each unit that touches buildable code.
3. **QEMU when applicable**, one ISA at a time (amd64 first, fastest under
TCG), reserving the full three-arch pass for milestones — new
primitives wired end-to-end, deletions landed, before declaring a
phase implemented. This redesign is expected to surface real Tripod
gaps the same way the `LOAD-DOE` bug did during the `doe-campaign.4th`
migration — that's a feature of doing this incrementally with real
boots, not a risk to design around.
4. **Never modify `FM-TEST`/`FM-SET`/`FM-CLR`/`FM-FIND-FREE`/`BLK-ALLOC`/
`BLK-FREE` in `artemis/init.4th`.** They're registered, tested,
HARD LOCKED (blocks 4110–4113, `capsules/MANIFEST.md`) — Artemis's own
disk-tier allocator. Per rev e, Hermes never touches these, and never
builds an equivalent for Artemis's disk; Hermes's own RAM/ramdrive
tiers (own M9 allocator, once that lands) are a separate, independent
implementation over Hermes's own blocks, not a modification, extension,
or generalization of Artemis's disk-tier primitives.
---
## Problem
`hermes/init.4th`'s message and channel storage is a fixed in-memory pool:
`CREATE MSG-ARENA MSG-MAX MSG-CELLS * CELLS ALLOT` (Block 4101), a
32-slot, 9-cell-per-message (`MSG-CELLS`) free-list array; `CH-ARENA` is
the same shape at 16 slots. A fixed compile-time array can't grow without
recompiling, and Hermes's own comments already flag real exhaustion risk
(`FLEET-DOE-WORKLOAD`'s block 2058 comment: `BROADCAST-TEST`... would
exhaust Hermes's fixed member/message arenas after ~20 runs").
Separately, the heat-physics sibling doc's window-sizing already depends
on knowing the real capacity messages will be observed at (see that doc's
Status header) — an arena bound that can silently change relative to
`MSG-MAX`/`CH-MAX` compile-time constants makes that reasoning fragile.
## Goal
Give Hermes's messages and channels a way to outgrow the fixed 32/16
in-memory arenas — first into Hermes's own per-VM RAM/ramdrive blocks
(M9 scope), and only past that, **by asking Artemis to hold what Hermes's
own storage can't** — matching `TRIPOD.md`'s dispatch table exactly:
*"Block I/O → Artemis... `VM-EXEC` takes a VM name. The caller names the
correct VM for the work type. There is no routing algorithm."* Hermes
does grow its own block-addressed storage (own RAM, own ramdrive — that's
its per-VM identity, not optional); what it never grows is a *disk* of
its own. Past its own ramdrive, it sends a message; Artemis does the
storing.
**Explicitly not the goal:** Hermes appending its own persistent disk
(rev c's tier 3 — superseded, see revision history). Also not the goal:
designing Hermes's own per-VM RAM/ramdrive block layout (M9 scope,
separate work) or Artemis's own storage internals (zones, BAM cell
layout, migration policy) — the latter already specified, in more detail
than this doc would add, in `.claude/ARTEMIS.md`'s Future Material
section, and explicitly not build-authorized yet ("Do not implement any
of this without Captain Bob explicitly reopening it").
## The C/FORTH boundary question
Tiers 1–2 (Hermes's own RAM/ramdrive) are M9 scope — genuinely deferred,
not this doc's to design, so this question doesn't apply to them here.
For tier 3 (asking Artemis), the question doesn't arise for Hermes at
all — Hermes only ever calls `VM-EXEC` (or the equivalent message-send
word) naming Artemis, the same capability-based dispatch every other
Tripod interaction uses. Artemis's own block-storage subsystem is already
proven pure FORTH (`artemis/init.4th`'s free-map bitmap, allocator,
`LE32!`/`LE64!` arch-neutral I/O, all composed over `BLOCK`/`BUFFER`/
`C@`/`C!`, zero new C) — that precedent matters for whoever eventually
builds out Artemis's Future Material, not for anything in this doc.
## Storage model: three tiers — own RAM, own ramdrive, then ask Artemis
Every VM — Hermes included — has its own RAM-backed blocks 0–2047 and its
own ramdrive from 2048 to its end, as a hard requirement of what a VM is
(`capsule_birth_baby()`'s own "PERSONALITY: per-VM block storage is M9
scope" comment names this as planned, not yet built). `ARTEMIS.md`'s
*"Artemis is memory. She owns the external attached virtual disk image
and everything that lives on it"* is about the **external persistent
disk** specifically — it doesn't reach into a VM's own RAM or ramdrive,
which are that VM's own identity, not storage Artemis administers.
1. **Own RAM blocks first (0–2047).** Hermes's own private RAM-backed
block range, part of its M9 per-VM identity, once that milestone
lands. Today's `MSG-ARENA`/`CH-ARENA` (`CREATE...ALLOT`) are the
pre-M9 stand-in for this tier — an in-VM-memory working set, not yet
block-addressed. Fast, synchronously touchable, no round trip — the
right home for a message actively being delivered, ACKed, or NACKed.
2. **Own ramdrive next (2048 to its end)**, once tier 1 isn't enough.
Also part of Hermes's own per-VM identity under M9 — Hermes's own
ramdrive, not a shared or Artemis-administered one. This is where
`MSG-ARENA`/`CH-ARENA` graduate to once they outgrow the fixed
32/16-slot compile-time arrays, block-addressed within Hermes's own
range, still entirely within Hermes.
3. **Ask Artemis, only once tier 2 is exhausted.** This is the one place
rev c was wrong and rev d's correction actually applies: Hermes does
not append its own persistent disk. Once its own ramdrive runs out (or
a message needs to survive past Hermes's own reap), it sends Artemis a
store request and gets back a handle — see "The logical-reference gap,
resolved," below.
**Escalation between tiers is compudynamics-driven, not a size
threshold.** `ARTEMIS.md`'s own Migration Policy already establishes the
pattern for exactly this kind of question: *"No zone is special... when a
zone has pressure... the coldest block migrates first because it is
coldest... no magic constants... everything else is self-organizing."*
That's not specific to Artemis's disk zones — it's the same heat/cool/
reap model already governing word heat, VM heat, and (once
`msg_decay_slope_q48` lands) message heat. Tier 1→2 pressure and tier
2→3 pressure work the same way: the coldest message under RAM-tier
pressure is what migrates to Hermes's own ramdrive, and the coldest
ramdrive-tier message under pressure is what gets sent to Artemis — not
an arbitrary byte-count cutoff invented for this doc. The concrete
numbers (window depth, pressure threshold) are still M9/Future-Material
scope, unspecified — but the *mechanism* isn't an open question, it's
already-established compudynamics, applied one level up.
`capsules/MANIFEST.md`'s existing Hermes ranges (4100–4153) stay what
they've always been — capsule source organization for `hermes/init.4th`'s
word definitions, not a claim on tiers 1–2's future per-VM block ranges
(those are a distinct, not-yet-designed M9 concern, out of scope here —
see Open questions).
## The logical-reference gap, resolved
The open question flagged when rev c was superseded: `ARTEMIS.md`'s
already-specified Logical BAM (`ARTEMIS.md`, "Two BAMs" — 4 cells:
physical LBN, heat, XXHash64 content identity, flags) is keyed by
content. That's the right key for what it's explicitly scoped for —
*"Content-addressing, ACL records, and cold capsule storage live here...
Identity is derived from block content — same content, same identity...
Mutation is detectable: hash changes if content changes."* A live Hermes
message is exactly the case that breaks that key: its content can change
during its lifetime (composed, delivered, ACKed), so a content hash
cannot be the stable handle Hermes holds onto across a request/fetch
round trip — the hash itself would move whenever the content did.
The resolution doesn't require inventing a new mechanism — `HERMES.md`
already has the right shape for a stable, VM-minted reference, used today
for exactly this purpose: *"`channel_id` = owner VM identity + monotonic
sequence number. Unforgeable by non-owners. Hermes mints it. No VM
constructs its own channel_id."* A message or channel that needs
Artemis-backed storage gets stored under that same kind of handle —
Hermes's own minted ID, opaque to Artemis, stable for the object's
lifetime regardless of what its content is doing or where Artemis
physically keeps it. Artemis's storage-request protocol accepts that
opaque ID as the key for this class of entry — necessarily a *different*
internal bookkeeping path from the content-addressed Logical BAM entries
`ARTEMIS.md` already specifies for cold/immutable data, since the two
have genuinely different identity semantics (stable-opaque-handle vs.
content-hash). Designing that internal path is Artemis's own future
work, not this doc's — flagged here only so the gap has a named, grounded
answer instead of sitting open.
## Physical BAM / Logical BAM — per-VM, both Hermes's and Artemis's own
Rev c had this right and rev d wrongly discarded it: Physical BAM
(free-map + block-heat allocator) and Logical BAM (normalizing that same
VM's own heat for fleet-conservation participation) are **per-VM
concepts**, not something only Artemis has. Under M9, Hermes eventually
needs its own Physical/Logical BAM over its own RAM/ramdrive blocks
(tiers 1–2) — a new, independent implementation, out of scope for this
doc (M9 is separate work), but real and not something this doc should
keep implying doesn't exist.
What rev d got right and stays true under rev e: Hermes's own BAM (once
built) has nothing to do with, and never touches, Artemis's disk-tier
BAM. `ARTEMIS.md`'s Future Material section already specifies Artemis's
own version at a level of detail this doc doesn't need to restate:
Physical BAM (2 cells — physical LBN, free flag; "the raw substrate... no
heat, no identity"), Logical BAM (4 cells — physical LBN, heat, content
identity, flags), three thermal zones with a self-organizing migration
policy. That whole section is explicitly marked Future Material in
`ARTEMIS.md`, not build-authorized without Captain Bob reopening it —
unrelated to whatever shape Hermes's own future per-VM BAM eventually
takes, and `artemis/init.4th`'s `FM-*`/`BLK-ALLOC`/`BLK-FREE` (blocks
4110–4113, ★ HARD LOCKED) remain untouched by Hermes either way.
## The mechanism (sketch — tier 3 protocol shape only)
Tiers 1–2 (Hermes's own RAM/ramdrive) are M9 scope, not sketched here —
that's per-VM block storage work shared by every Tripod VM, not specific
to Hermes's messages, and belongs in its own doc when M9 is reopened.
What follows is tier 3 only: what happens once Hermes's own ramdrive is
exhausted and it has to ask Artemis.
### 1. A store/fetch request, dispatched the same way any Tripod work is
`TRIPOD.md`'s dispatch model already covers this — Hermes names Artemis
by capability, same as any Block I/O work: something like `ART-STORE
( addr len logical-id -- ack/nack )` and `ART-FETCH ( logical-id --
addr len )`, sent via `VM-EXEC` (or the message-passing equivalent once
Hermes's async model, `HERMES.md`'s G1/G8 items, supports a return
value). `logical-id` is Hermes's own minted ID — see "The
logical-reference gap, resolved," above.
### 2. What Hermes sends is a payload, not a block
Hermes doesn't decide block geometry, packing, or free-map granularity —
that's Artemis's Future Material to design (`ARTEMIS.md`'s "Physical BAM
Block Geometry" section already sketches a 3-data+1-metadata packing
scheme for variable-length records generally, which a Hermes message
would be one instance of, once that chapter is reopened). This doc stops
at "Hermes hands Artemis a payload and an ID" — how Artemis packs,
allocates, and migrates it is out of scope here by design, not by gap.
### 3. Synchronous limitation, already named in `HERMES.md`
`HERMES.md`'s G8 gap applies directly: *"Wiring [a cross-VM value] into
`K-FLEET` in Hera requires a cross-VM return value — something the
current synchronous `VM-EXEC` model cannot deliver."* `ART-FETCH`
returning a payload has the same shape problem. Until the async
inter-VM model lands, a Hermes→Artemis fetch request is bounded by
whatever `VM-EXEC`'s synchronous call/return can actually carry — not
redesigned here, just inherited as a precondition.
## Consumer code migration
Deferred to implementation, for the same reason rev c deferred it: the
concrete request/response word names, and everything on Artemis's side
of the boundary, depend on Future Material that isn't build-authorized
yet (`ARTEMIS.md`: *"Do not begin implementation without explicit
instruction from Captain Bob"*). A follow-up pass once Artemis's flat-pool
Immediate Goal is done and this chapter is explicitly reopened should add
this section for real.
## Verification approach
- Three-arch acceptance as usual once any unit reaches boot-behavior-
affecting completeness, per the Execution Rules above.
- Real message-storm test: repeat `BROADCAST-TEST`-style traffic well
past the ~20-run point that exhausts today's 32-slot arena (per the
comment already in `hermes/init.4th`), confirm messages that overflow
to Artemis are still retrievable and don't silently drop — the
concrete, observable proof this redesign did something, not just moved
the same limit elsewhere.
- `TRIPOD-TEST`'s existing `PASS: broadcast reached all 3` /
`PASS: msg queued` / `PASS: E2E msg flow` checks should continue
passing unchanged throughout — this redesign changes what happens on
arena overflow, not Hermes's external message-delivery behavior.
## Explicitly out of scope
- **The heat-decay mechanism** (`msg_decay_slope_q48`) — sibling doc,
`HERMES-MESSAGE-CHANNEL-PHYSICS-DESIGN-20260708.md`, independent of
this doc's rev d/e corrections.
- **Hermes's own per-VM RAM/ramdrive BAM (tiers 1–2, M9 scope).** Real,
confirmed (`capsule_birth_baby()`'s own comment names it), and not
designed in this doc — a separate, larger piece of work shared in
shape by every Tripod VM, not specific to Hermes's messages.
- **Channel storage** (`CH-ARENA`) — this doc focuses on messages, the
higher-volume and more clearly arena-exhaustion-prone of the two per
the existing `hermes/init.4th` comment. Whether channels need the same
overflow-to-Artemis treatment is a question for after messages are
done, not decided here.
- **Artemis's own storage internals** (zones, BAM cell layout, packing,
migration policy). Fully specified as Future Material in `ARTEMIS.md`
already; explicitly not build-authorized without Captain Bob reopening
that chapter. This doc's protocol sketch (section 1-3 above) doesn't
depend on those internals being resolved to be a correct *shape* — it
just can't be implemented before they are.
- **Finishing Artemis's own `K-FLEET` wiring gap.** Tracked in
`ARTEMIS-BAM-ACCEPTANCE-20260703.md`'s Deferred/Next table and
`ARTEMIS-BLOCK-PHYSICS-DESIGN-20260708.md`. Not this doc's to close.
- **Artemis block heat generally** — sibling doc,
`ARTEMIS-BLOCK-PHYSICS-DESIGN-20260708.md`.
## Open questions (flagged, not guessed at)
1. **The concrete word names and wire shape** for the store/fetch
protocol — sketched at the level of "a request with a logical ID,"
not specified further, since it depends on Artemis's own
not-yet-authorized storage internals.
2. **Target message capacity** — "bigger than 32" is not a number.
Needs a real target, ideally informed by the message-storm test's
observed exhaustion behavior at various loads.
3. **How Artemis's internal store distinguishes opaque-handle entries
(Hermes messages) from content-addressed entries (cold capsules, ACL
certs)** — flagged in "The logical-reference gap, resolved," above,
as Artemis's own design work.
4. **The synchronous-call-return limitation** (section 3 above) — whether
this doc's protocol waits for `HERMES.md`'s async model, or an interim
shape is worth designing sooner.
5. **Whether anything here genuinely needs a new C primitive** — no
evidence either way yet, since Artemis's side of this isn't designed.
Precedent (`artemis/init.4th`'s existing block subsystem is pure
FORTH) suggests no, but that's Artemis's question to answer when this
chapter reopens, not this doc's to assume.
6. **The concrete pressure threshold for tier escalation** — the
*mechanism* is settled (compudynamics heat-pressure, matching
`ARTEMIS.md`'s Migration Policy — see "Storage model," above); the
*numbers* (how much ramdrive pressure triggers a tier-2→3 request)
depend on M9's own design, which doesn't exist yet.
7. **`ARTEMIS-FREEMAP-CAPSULE-RESERVATION-GAP-20260710.md` applies
here too.** Artemis's live `BLK-ALLOC` doesn't yet reserve the LBN
range her own capsule-loaded content occupies — a gap in Artemis's
current allocator, not something this doc introduces. But a future
`ART-STORE` request from Hermes (section 1 above) would hand out
blocks from that same unprotected pool, so this doc's tier-3
implementation inherits that gap and shouldn't be built before (or
without accounting for) it being closed.
docs/working/archive/superseded/2026-07-08-hermes-message-channel-physics-design.md
+324
View File
@@ -0,0 +1,324 @@
> **SUPERSEDED (docs triage 2026-10-14).** Heat physics for the Hermes VM's messages and channels; the Hermes VM no longer exists.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/working/architecture/HERMES-MESSAGE-CHANNEL-PHYSICS-DESIGN-20260708.md`.
# Hermes Message/Channel Heat Physics — Design Doc
**Date:** 2026-07-08 (rev 2026-07-08a)
**Branch:** `lithosananke`
**Status:** Design only. Not implemented. **Depends on a sibling
architectural doc**: `HERMES-MESSAGE-BLOCK-STORAGE-DESIGN-20260708.md`
(rev c), which plans to move Hermes's message area off its current
in-memory `CREATE...ALLOT` arena (`MSG-ARENA`/`CH-ARENA`, `MSG-MAX`=32/
`CH-MAX`=16) onto a dedicated, LBN-backed block range within the kernel
ramdrive — a Hermes-owned physical BAM over Hermes's own blocks, built
independently of and with zero coupling to Artemis (Artemis's own
disk-block model is precedent for the allocator *shape*, not something
Hermes's storage shares or depends on). That redesign is out of scope for
this doc — this doc's window-sizing reasoning (section 2) is provisional
against the *current* in-memory scale and will need revisiting once the
block-backed redesign lands, since block-backed storage will likely
change both the arena's real capacity and its churn characteristics.
Sibling doc to
`VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md`'s "Fold-in question" section,
which settled that messages and blocks get their own independent decay
mechanisms — not merged into `VMPhysics`, not derived from its slope —
each fit from its own rolling window, in separate follow-on docs. This is
that doc for Hermes (messages + channels). See
`ARTEMIS-BLOCK-PHYSICS-DESIGN-20260708.md` for the block-heat sibling.
**Author:** Captain Bob / Claude Code
**Correction (2026-07-09):** this doc's original Goal section (rev a)
claimed message/channel heat has no conservation invariant, citing a
"settled fold-in decision" from `VM-PHYSICS-DYNAMIC-FLEET-DESIGN-
20260705.md`. That claim was wrong — it was derived without first reading
`.claude/HERMES.md`, which is authoritative and states plainly: *"Every
object Hermes manages — messages and channels — participates in K≡1.0.
Fleet K includes the thermal contribution of in-flight messages and open
channels."* A full Tripod-docs audit (`TRIPOD.md`/`HERMES.md`/
`ARTEMIS.md`) confirmed this directly contradicts what's below. The Goal
section is corrected in place. The rest of this doc's mechanism — an
independently-fit decay *rate* for messages and channels, its own
`HermesHeatWindow` struct, its own rolling window, separate from
word-level and VM-fleet physics — is unaffected: `HERMES.md` requires
message/channel heat to *participate in* K≡1.0, not that it share
VM-fleet's inference mechanism. Those are separable, and only the
conservation claim was wrong.
---
## Problem
`capsules/hermes/init.4th` hardcodes `65208 CONSTANT Q-DECAY` (Block 4100)
and applies it identically to two different entities via straight
multiplicative shrink:
- `MSG-COOL-ONE`/`MSG-COOL-ALL` (Block 4108): `heat = heat * Q-DECAY` for
every live message in the 32-slot `MSG-ARENA`.
- `CH-COOL-ALL` (Block 4114): same formula, walked over the live channel
linked list (16-slot `CH-ARENA`).
This is the exact "before" picture `VM-PHYSICS-DYNAMIC-FLEET-DESIGN`
already diagnosed for VM heat: `65208` was independently re-declared as
an identical hand-picked constant in three separate files
(`compudynamics.4th`, `hermes/init.4th`, `artemis/init.4th`) — copy-pasted,
uncoordinated, with no regression anywhere. The VM-heat instance of this
was fixed (`capsule_vm_physics.c`). Hermes's and Artemis's instances
weren't — this doc is Hermes's fix.
No evidence exists that `65208` (≈0.9950 per cool-call, i.e. a message or
channel's heat halves after roughly 138 `MSG-COOL-ALL` calls) was ever
tuned against real message/channel churn. It's the same number VM heat
used, which was itself never tuned against VM touch frequency either.
## Goal
Replace the hardcoded `Q-DECAY` with a `msg_decay_slope_q48` inferred from
observed cooling behavior, mirroring the *shape* of the word-level and
VM-fleet physics engines — same mechanics (rolling window, inferred rate,
skip-rather-than-substitute-a-default when unwarmed) applied to a third
kind of entity — without merging into either existing engine's *state* or
*inference mechanism*.
**Message and channel heat DOES participate in K≡1.0** — per
`.claude/HERMES.md`'s Compudynamic Invariant: *"Fleet K includes the
thermal contribution of in-flight messages and open channels... Hermes
reaping a message or channel must rebalance K correctly."* This doc does
not design that redistribution mechanism — `HERMES.md` already establishes
it conceptually (ACK → clean death, K redistributed; reap → K rebalanced),
and `HERMES.md`'s own G8 gap tracks the one known incompleteness (`HERMES-
K` exists and totals live message/channel heat, but wiring it into Hera's
`K-FLEET` is deferred pending an async inter-VM return-value model). What
this doc *does* design is orthogonal to that: **the rate** at which a
live message or channel's heat decays between creation and reap. Whether
that decay is later credited back to the fleet correctly is `HERMES.md`'s
concern and G8's tracked gap, not this doc's.
## Why this can't just copy either existing mechanism verbatim
**Not word-level's shape (rolling window of `word_id`s + trajectory
replay against current heat):** words are permanent dictionary entries —
`extract_heat_trajectory` (`inference_engine.c`) replays a window of
`word_id`s and looks up each one's *current* heat, exploiting that a word
never stops existing. Messages and channels are transient: `MSG-REAP`
frees a message's arena slot back to the free list once its heat cools
past whatever threshold or it's explicitly consumed, and a freed slot's
"current heat" is meaningless — it may already belong to an unrelated
message by the time anyone would replay a window containing its old ID.
Per-entity trajectory replay does not work here.
**Not VM-fleet's shape either, but structurally the closest analog:**
`capsule_vm_physics.c` (rev o) doesn't curve-fit a trajectory — it
recovers the transfer rate directly, because the transfer law is known
exactly (`amount = elapsed_us * slope_q48 >> 16`) and any unclamped touch
can be inverted for an exact per-sample rate; the fleet-wide
`fleet_transfer_slope_q48` is the *median* of recovered per-sample rates
over a 64-deep window of aggregate touch samples, not per-VM. The same
"known law, recover the rate, aggregate not per-entity" shape applies
here: `heat_new = heat_old * Q-DECAY` is *also* a fully known multiplicative
law, so there is nothing to curve-fit — what's actually being inferred is
not "what is the decay law" (known) but "is the current decay *rate*
tuned right for observed churn," which is a different question than either
existing engine answers, addressed below.
## The mechanism
### 1. What "inferring the rate" actually means here
Because the decay law itself is known and fixed-shape (`heat *= rate`),
there's no unknown functional relationship to recover the way VM-fleet
recovers `fleet_transfer_slope_q48` from touch samples, or the way Loop #6
recovers `decay_slope_q48` from a heat trajectory. What *can* legitimately
vary and be worth observing: **how fast messages/channels are actually
being reaped relative to how fast they're being created.** If churn is
high (many `MSG-ALLOC`/`MSG-REAP` cycles per cool-sweep), a slow decay
rate lets dead-weight messages linger, consuming arena slots that new
messages need. If churn is low, an aggressive decay rate discards heat
signal (e.g. NACK-retry priority, per `hermes/init.4th`'s own
`MSG-REDELIVER-NACKED`) before it's useful.
`msg_decay_slope_q48` should therefore be a function of **observed reap
throughput relative to arena occupancy**, not a re-derivation of the
`Q-DECAY` law itself. Concretely: a rolling window of samples taken once
per `MSG-COOL-ALL` call, each recording `(live_count, reaped_since_last)`;
periodically re-fit `msg_decay_slope_q48` so that, at the observed churn
rate, an average message crosses the reap threshold within a target
window (e.g. "most messages should be reaped within N cool-sweeps of
going idle" — the specific target is an open question, see Resolved
Items).
This is a genuinely different inference problem than either existing
engine solves, which is why it doesn't reduce to copying `infer_decay_slope_q48`
or `vm_physics_tick`'s median-of-rates — it's closer in spirit to Loop #5's
window-width inference (which also asks "is the current parameter well-
matched to observed dynamics," not "what is the true decay curve").
### 2. `HermesHeatWindow` — new struct, two instances (message, channel)
```c
typedef struct {
uint32_t live_count; /* live entities at this sample */
uint32_t reaped_since_last; /* entities freed since the prior sample */
} HermesHeatSample;
typedef struct {
HermesHeatSample samples[HERMES_HEAT_WINDOW_DEPTH];
uint32_t head;
uint32_t count; /* saturates at HERMES_HEAT_WINDOW_DEPTH */
int is_warm; /* count >= HERMES_HEAT_WINDOW_DEPTH */
} HermesHeatWindow;
```
Two instances — one for messages, one for channels — since they have
independent arenas, independent churn rates, and are proposed to have
independent decay slopes as this doc's own conservative default (see
Open questions). Not one shared window; `MSG-MAX`=32
and `CH-MAX`=16 are different enough scales that conflating them risks
exactly the kind of unjustified shared-constant problem this doc exists
to fix.
**`HERMES_HEAT_WINDOW_DEPTH`: proposed 16, not yet settled, and provisional
against a scale that's about to change.** `MSG-MAX`=32 and `CH-MAX`=16 are
today's *in-memory arena* limits — far smaller than word-level's millions
of executions or even VM-fleet's `VM_FLEET_WINDOW_DEPTH`=64, so a window
deeper than the arena itself provides no additional information at
*current* scale. But per this doc's Status header, the message area is
planned to move onto a dedicated block range, which will likely raise
that capacity substantially (Artemis's equivalent, `ART-DATA-BLKS`, is
22,998 — three orders of magnitude larger than `MSG-MAX`=32). 16 is a
starting guess against *today's* scale, not derived from anything, and
should be treated as provisional until the block-backed redesign's real
capacity is known (see Resolved Items).
### 3. Recording — passive, hooked at the existing cool-sweep call sites
`MSG-COOL-ALL` and `CH-COOL-ALL` already sweep their entire arena once per
call; a new `hermes_heat_sample(HermesHeatWindow*, uint32_t live_count,
uint32_t reaped_since_last)` call at the end of each sweep is the only new
call site needed — no new dispatch points, matching both existing engines'
"passive observer, hooked at what's already there" discipline.
**`reaped_since_last` bookkeeping is the one real implementation question
this doc leaves open.** `MSG-REAP` and channel reaping aren't currently
instrumented to report a count to the cool-sweep — that's new plumbing,
not free. Whether it's a shared counter incremented at each reap site and
read-and-reset by the cool-sweep, or something else, is unresolved (see
Resolved Items) — flagged rather than guessed at.
### 4. Inference — heartbeat-gated, same skip-don't-substitute discipline
`hermes_heat_tick()`, called from `HERMES-TICK`'s existing heartbeat path
(mirroring `vm_physics_heartbeat_tick`'s hook into each VM's own tick):
when a window is warm, re-fits that window's decay slope; when unwarmed,
leaves the current slope untouched rather than substituting a default —
same philosophy both existing engines already use.
The actual fit function (how `(live_count, reaped_since_last)` samples
become a new `msg_decay_slope_q48`) is intentionally not specified in
concrete formula form here — see Resolved Items. Writing a specific
formula now, without the real observed-churn data either existing engine's
design had before it settled on its final shape (VM-fleet went through
revisions b through o before landing on median-of-rates), would be
inventing content rather than designing it.
### 5. Application — replaces `Q-DECAY Q.*` at both cool-sweep sites
`MSG-COOL-ONE` and `CH-COOL-ALL`'s `Q-DECAY Q.*` becomes
`msg_decay_slope_q48 Q.*` / `ch_decay_slope_q48 Q.*` (or a single
`msg_decay_slope_q48` if the message/channel split above turns out
unwarranted after real observation — see Resolved Items). No change to
the multiplicative decay *shape* itself, only to where the rate comes
from.
## Consumer code migration
- **`hermes/init.4th` Block 4100**: `65208 CONSTANT Q-DECAY` deleted.
- **`hermes/init.4th` Block 4108** (`MSG-COOL-ONE`/`MSG-COOL-ALL`):
`Q-DECAY` reference replaced with a call to fetch the current
`msg_decay_slope_q48` (new C primitive, e.g. `MSG-DECAY-SLOPE@`);
`MSG-COOL-ALL` gains the sample-recording call described in section 3.
- **`hermes/init.4th` Block 4114** (`CH-COOL-ALL`): same pattern for
channel heat.
- New kernel-only file pair, matching `capsule_vm_physics.c`'s own scope
(Hermes-specific state, no hosted-build equivalent since the hosted
build has no Hermes):
- `src/starkernel/capsule/hermes_heat_physics.c`
- `include/starkernel/hermes_heat_physics.h`
- New FORTH-visible primitives (backed by the new file): `MSG-DECAY-SLOPE@`,
`CH-DECAY-SLOPE@` (or one shared accessor if sections above's split is
resolved differently), plus a status word mirroring `VM-PHYSICS-STATUS`'s
shape for diagnostics.
## What gets deleted
- `hermes/init.4th` Block 4100: `65208 CONSTANT Q-DECAY`.
Nothing else — `MSG-COOL-ONE`/`MSG-COOL-ALL`/`CH-COOL-ALL` keep their
existing structure, only the decay-rate source changes.
## Verification approach
- Three-arch acceptance as usual (kernel-only code,
`#ifdef __STARKERNEL__` gates per `.claude/CLAUDE.md`).
- A targeted test: drive real message/channel churn (repeated
`SEND-BROADCAST-TEST`-style traffic) at two different rates across
separate boot runs, confirm `msg_decay_slope_q48` converges to
different values reflecting the different observed churn — the direct
analog of confirming VM-fleet's `fleet_transfer_slope_q48` actually
responds to real touch frequency rather than sitting at its seed value.
- Confirm `HermesHeatWindow`'s dead-entity handling (a message reaped
mid-window) doesn't fault or corrupt the sample count — same class of
test as VM-fleet's "kill-during-warm-up" test
(`VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md`).
## Explicitly out of scope
- **Moving Hermes's message/channel storage onto a dedicated LBN-backed
block range** (mirroring Artemis's `ART-DATA-BLKS`). Flagged in this
doc's Status header as a real dependency, but the redesign itself —
which blocks, what the on-disk layout looks like, how it interacts with
`MSG-ARENA`/`CH-ARENA`'s current in-memory shape — is a separate,
not-yet-written architectural doc, not part of this heat-decay design.
- **Artemis block heat** — sibling doc, `ARTEMIS-BLOCK-PHYSICS-DESIGN-20260708.md`.
- **Designing how K is credited/debited at message and channel alloc/reap**
— already governed conceptually by `HERMES.md`'s Compudynamic Invariant
section; the one known incompleteness (wiring `HERMES-K` into Hera's
`K-FLEET`) is `HERMES.md`'s own tracked G8 gap, not something this doc
redesigns. This doc's decay-rate inference operates on top of that
redistribution, not in place of it.
- **The multi-level DoE rewrite** (word/VM-fleet/message/block as four
independent metric spaces) — this doc gives message/channel heat its
mechanism; observing it as part of a real experiment is separate,
larger work already on the punch list.
- **Implementation** — this is a design doc. Writing
`hermes_heat_physics.c` is follow-on work, not part of this pass.
## Resolved items
None yet — this is a first-pass design doc, not an iterated one. Unlike
`VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md`'s extensive revision history
(rev a through t, each resolving a question found during real
implementation and three-arch verification), this doc has had no
implementation pass yet to surface and resolve open questions against.
## Open questions (explicitly not resolved — flagged, not guessed at)
1. **`HERMES_HEAT_WINDOW_DEPTH` value.** Proposed 16 (matching `CH-MAX`),
not derived from real observed churn data. Needs a real implementation
pass with instrumented boot logs (same discovery method VM-fleet used:
`doe_log.c`'s per-VM heat CSV columns exposed the dead-flat-trajectory
bug that led to VM-fleet's rev-f seed-value fix) before treating this
as settled.
2. **The concrete fit function** for turning `(live_count,
reaped_since_last)` samples into `msg_decay_slope_q48`. Section 4
deliberately stops short of a formula. A first candidate worth
prototyping: target reap-latency-in-cool-sweeps, adjust the slope
proportionally to the ratio of observed to target latency — but this
is a starting hypothesis to implement and test against, not a
specified design.
3. **`reaped_since_last` bookkeeping mechanism** — where the reap-count
plumbing lives and how it's reset per sample window.
4. **Whether messages and channels genuinely need independent slopes**,
or whether observed data shows one shared `msg_decay_slope_q48`
suffices — section 2 assumes independence as this doc's own
conservative default (matching the pattern word-level and VM-fleet
physics each use their own independently-fit rate), but this should
be revisited once real data exists.
docs/working/archive/superseded/2026-08-02-artemis-hermes-compudynamics-plan.md
+379
View File
@@ -0,0 +1,379 @@
> **SUPERSEDED (docs triage 2026-10-14).** An implementation plan for Artemis and the Hermes VM; the Hermes half is gone with the VM, the Artemis half was carried by the FABRIC work.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/working/architecture/ARTEMIS-COMPUDYNAMICS-IMPLEMENTATION-PLAN-20260802.md`.
# Artemis (and Hermes) Compudynamics — Implementation Plan
**Date:** 2026-08-02
**Branch:** `master`
**Status:** Plan approved in shape, **not final and not started.** Captain Bob
has more to discuss on the design before implementation begins — this document
is expected to change as a result. Do not treat it as settled, and do not
start coding from it as written. The six Open Questions are the known gaps;
the pending discussion may add or reframe others.
**Author:** Captain Bob / Claude Code
**Siblings:** `ARTEMIS-BLOCK-PHYSICS-DESIGN-20260708.md` (the decay-rate design
this implements), `HERMES-MESSAGE-CHANNEL-PHYSICS-DESIGN-20260708.md`,
`ARTEMIS-BAM-ACCEPTANCE-20260703.md` (which tracks the `ART-TICK heartbeat`
gap this closes), `VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md` (pattern source).
---
## Preamble: how this plan came to be written
This plan exists because of a wasted day, and the failure is worth recording
so it isn't repeated.
On 2026-08-02 a 30-replicate Artemis surface-stress campaign was run across
all three architectures to confirm the `block_words.c` stale-pointer
cache-aliasing fix held at scale. It did: **4500 trials, zero failures,
amd64/aarch64/riscv64**. That result is real and stands.
While reviewing the campaign's heartbeat CSV, the assistant observed the
`hera_heat_q48` / `hermes_heat_q48` / `artemis_heat_q48` columns sitting
perfectly flat (`65536, 0, 0`) across all 51,092 ticks, concluded this was a
defect, and rebuilt the campaign driver to route every stress-test phase
through Hera via `VM-EXEC` so that fleet heat would move.
That was wrong three separate ways:
1. **Wrong metric.** Those columns are *VM-fleet* heat
(`capsule_vm_physics.c`, `vm_physics_heat_of()`). The quantity of interest
for Artemis is *block* heat (`BLK-HEAT` / `ART-K-TOTAL`). Different
mechanisms, no connection between them.
2. **Wrong method.** Manufacturing cross-VM calls to force a physics reading
is explicitly prohibited: `TRIPOD.md` — *"No VM manages another VM's
internal heartbeat"*, *"There is no 'send to the hottest VM.' That model is
wrong and must not be implemented"*; `capsule_vm_physics.h:35-37` — *"a
passive observer of VM activity, never a driver of it."*
3. **Wrong premise.** Block heat cannot move today under any workload,
because nothing ages it (see below). No amount of driving would have
produced a signal.
All three were avoidable by reading `.claude/TRIPOD.md` and `.claude/ARTEMIS.md`
completely before touching Tripod code — which both files instruct in their own
headers, and which `.claude/CLAUDE.md` also instructs. They were skimmed by
keyword instead. Keyword search cannot surface a prohibition, because
prohibitions do not contain the term you are searching for.
The capsule changes from that detour have been reverted. **Standing rule
going forward: read the governing documents in full before starting any major
piece of work in this repo.**
---
## The actual defect
Block heat and message heat do not decay. Not "decay incorrectly" — **do not
decay at all**, unless something manually calls a sweep word.
- `ART-TICK` (`ART-COOL ART-REAP`, `capsules/artemis/init.4th` block 4140) has
**zero call sites** anywhere in the tree — no C, no FORTH, no capsule.
- `HERMES-TICK` (`capsules/hermes/init.4th` block 4116) has **zero C call
sites**. Every caller is Hera poking it manually via `VM-EXEC`
(`capsules/init.4th` lines 31, 47, 66, 83, 130).
`BLK-HEAT@` and `MSG-HEAT@` read a stored number that nothing ages. Heat is
frozen between manual pokes — neither lazy nor scheduled, simply absent.
Consequences:
- Blocks never reap by cooling. `ART-REAP` frees blocks at heat exactly 0;
nothing ever brings a block to 0 except an explicit `BLK-FREE`.
- Message TTL never expires on its own, contradicting `HERMES.md`'s stated
model (*"A message that nobody answers simply cools to death"*).
- `Q-DECAY` (65208) is hardcoded and hand-copied into three separate capsules
(`artemis/init.4th`, `hermes/init.4th`, and the now-deleted
`compudynamics.4th`) with no shared derivation — flagged in the block
physics design doc as the "before" picture.
`ARTEMIS-BAM-ACCEPTANCE-20260703.md` already tracks this as a named deferred
item ("ART-TICK heartbeat — wire into compudynamic tick loop"). This plan
closes it.
---
## The governing model: mirror how words work
Per Captain Bob's direction: **word-level execution physics is the reference
implementation.** Messages decay on **TTL**; blocks decay on **access
frequency**. Two distinct behaviours, one shared mechanism shape.
The word model, verified in source, has four parts:
**1. Storage.** `DictPhysics` (`include/vm.h:257-268`) carries
`last_decay_tick`, commented in-source as *"the only clock Loop #3 decay
reads."* The two `*_ns` fields beside it are explicitly marked **diagnostics
only, never read for decay** — the residue of the wall-clock bug fixed in
`VM-FLEET-ATTRACTOR-DESIGN-20260705.md` rev t.
**2. Lazy decay at every access point.** `physics_pre_execute()`
(`src/physics_execution_hooks.c:85-89`), `physics_on_lookup()` (`:203-215`),
and the kernel's inline duplicates (`src/starkernel/vm/vm_core.c:676, 870,
880`) all do exactly this:
```c
uint64_t elapsed_ticks = vm->heartbeat.tick_count - word->physics.last_decay_tick;
physics_metadata_apply_linear_decay(word, elapsed_ticks, vm);
word->physics.last_decay_tick = vm->heartbeat.tick_count;
/* then accumulate: */
physics_execution_heat_increment(word);
```
**3. Bounded background sweep with a resumable cursor.**
`vm_tick_apply_background_decay()` (`src/starkernel/vm/vm_runtime.c:357-388`)
processes at most `HEARTBEAT_DECAY_BATCH` dictionary entries per tick and
stores `vm->heartbeat_decay_cursor_id` to resume on the next tick.
**This is the answer to "sweeping 22,998 blocks per tick is too expensive."**
An earlier fleet-DoE attempt stalled outright on exactly that cost. The word
engine never sweeps its whole space in one tick — it amortizes. Artemis must
do the same.
**4. Decay math.** `physics_metadata_apply_linear_decay()`
(`src/physics_metadata.c:320-386`): `decay = (elapsed_ticks * slope_q48) >> 16`,
subtractive, clamped at zero. The slope is adaptive (`vm->decay_slope_q48`,
tuned by Loop #6), and the whole path is gated by
`ssm_config->L3_linear_decay` so L8 can switch it off.
### Two model differences to resolve, not assume
| | Words today | Artemis today | Consequence |
|---|---|---|---|
| Decay shape | **linear** subtract | **multiplicative** (`heat * 65208/65536`) | Multiplicative approaches zero asymptotically but may never reach exactly 0 in Q48.16; `ART-REAP` fires only at exactly 0. |
| On access | **increments** heat | `BLK-FETCH` **resets** to `Q.1` | Reset is recency-only; increment is frequency-sensitive. Bob named *frequency* as the block signal. |
Recommendation: match the word model on both counts. Confirm first (O2, O4) —
this changes when blocks reap.
---
## Design decisions
### D1. Block heat becomes C-side state
Today `BLK-HEAT` is a FORTH `CREATE ... ALLOT` array (block 4137, marked
immutable in `MANIFEST.md`). Word heat by contrast lives in C
(`DictEntry.execution_heat` + `DictPhysics`), with FORTH words reading it.
Mirror that. Block heat plus `last_decay_tick` move into
`src/starkernel/capsule/artemis_heat_physics.c`, with FORTH primitives for the
observational surface.
This is the decision that makes everything else fall out: the background sweep
becomes plain C called from `vm_tick()`, and **no C→FORTH bridge is required.**
That matters more than it might appear. **A mechanism for C to invoke a named
FORTH word on a tick does not exist in this codebase** — there is no
`vm_execute_word_by_name`, and `capsule_vm_hooks.c` carries only birth-time
hooks (exec, dict-hash, vm-alloc). Building one would be a substantial new
mechanism with its own risks. This design avoids needing it.
Per the ruling recorded in `VM-FLEET-ATTRACTOR-DESIGN-20260705.md` rev s —
*"TRIPOD.md's/ARTEMIS.md's 'StarForth dialect ONLY' applies to a thin
administrative/observational word surface; the underlying mechanics are meant
to be C99"* — this is in bounds. **Confirm the ruling still holds (O5);** D1
depends entirely on it.
### D2. Each VM's own heartbeat tick is the clock
`HeartbeatState` is embedded per-VM (`include/vm.h:280-298`). `tick_count`
advances only from `vm_tick()`, driven by that VM's own word executions
(`src/physics_execution_hooks.c:172-177`). Artemis's blocks therefore age on
Artemis's own tick. No cross-VM coupling, nothing injected, nothing to drive.
One clock, as the standing rule requires.
**Known consequence:** an idle VM does not tick, so its blocks and messages do
not age. For words this is explicitly intentional and documented as such. For
**messages it is a genuine problem** — a TTL that stops while Hermes is idle
is not a TTL. See O3.
### D3. The insertion point already exists
`vm_tick()` (`src/starkernel/vm/vm_runtime.c:145-149`) carries two
commented-out "future plugin" slots. That is the natural home for
`artemis_heat_tick()` and later `hermes_heat_tick()`.
---
## Phase 1 — Artemis block heat
**Goal:** block heat ages against Artemis's heartbeat, accumulates on access,
and reaps at zero, with an inferred slope replacing the hardcoded constant.
1. **New kernel-only files** (`#ifdef __STARKERNEL__`), following
`capsule_vm_physics.c`'s established shape:
- `include/starkernel/artemis_heat_physics.h`
- `src/starkernel/capsule/artemis_heat_physics.c`
State: per-block `{ heat_q48, last_decay_tick }` across `ART-DATA-BLKS`
(22,998 entries), plus the `ArtemisHeatWindow` the design doc specifies —
`ArtemisHeatSample { uint32_t live_count; uint32_t reaped_since_last; }`,
`ARTEMIS_HEAT_WINDOW_DEPTH` proposed 64.
2. **Lazy decay on access** — resolve heat at `BLK-FETCH` / `BLK-HEAT@` with
the exact `elapsed_ticks` pattern above, then apply the accumulation term.
3. **Bounded sweep** — `artemis_heat_tick()`, called from `vm_tick()`, gated
to Artemis's VM, processing a fixed batch per tick with a resumable cursor.
Mirror `vm_tick_apply_background_decay()` directly. This is what allows
reap to happen at all without an O(n) stall.
4. **Adaptive slope** — `blk_decay_slope_q48` fit from the window; when the
window is unwarmed, **skip the update rather than substituting a default**
(the discipline both existing engines use, and the specific trap that
froze VM-fleet physics at zero twice — see rev l and rev m of the fleet
doc).
5. **New FORTH primitives** — `BLK-DECAY-SLOPE@`, plus a status word
mirroring `VM-PHYSICS-STATUS`. Registration pattern is
`void (*)(VM *vm)` + `register_word(vm, "NAME", fn)`; see
`src/word_source/q48_words.c:76,204` for the minimal example and
`mama_forth_words.c:1091` for the kernel/MAMA variant (note it registers
into both the FORTH and MAMA vocabularies — follow that duplication).
6. **Capsule changes** — `ART-COOL` (block 4139) stops being a full sweep;
`ART-TICK` (block 4140) is retained for manual diagnostics. Both blocks are
editable — the hard-locked set is 4110–4113 plus 4128 and 4137. Any new
blocks belong in **4200–4299** (genuinely unclaimed; 4175–4199 is nominally
Hermes extension space).
### Available Q48.16 vocabulary
`Q.+ Q.- Q.* Q./ Q.ABS Q.NEG Q.LOG Q.EXP Q.SQRT Q.FROM-INT Q.TO-INT Q.1 Q.0
Q.SCALE Q.= Q.< Q.> Q.0= Q.MAX Q.MIN Q.PRINT` — all registered from
`src/word_source/q48_words.c:200-223`. There is **no** `Q.MOD` or remainder
word; if decay must be exactly conservative, that arithmetic has to come from
C. (Note `src/word_source/q48_16_words.c` exists but has no register function
and is not referenced by `word_registry.c` — dead code, do not follow it.)
---
## Phase 2 — Hermes message and channel heat
Same mechanism, after Artemis is proven. Per
`HERMES-MESSAGE-CHANNEL-PHYSICS-DESIGN-20260708.md`: `hermes_heat_physics.c`,
primitives `MSG-DECAY-SLOPE@` / `CH-DECAY-SLOPE@`, sampling hooked at
`MSG-COOL-ALL` (block 4108) and `CH-COOL-ALL` (block 4114).
**Behavioural difference from blocks:** messages are pure TTL — born hot,
decay only, never refreshed on access. No accumulation term. That is the whole
distinction Bob drew between the two.
**Correct that design doc before implementing it.** It proposes calling
`hermes_heat_tick()` from *"`HERMES-TICK`'s existing heartbeat path"* and
contrasts this against Artemis as having the harder problem for lacking one.
**That premise is false** — verified 2026-08-02: `HERMES-TICK` has no
heartbeat path either. Both VMs are in identical positions, and the doc's
sizing of the two efforts is wrong as written.
Message nodes are 8 cells (`HERMES.md:305-313`) with heat at cell 5. A
`last_decay_tick` needs either a 9th cell or C-side storage — C-side, per D1.
---
## Phase 3 — K participation (deferred; needs design, not wiring)
`.claude/ARTEMIS.md` requires that blocks participate in K≡1.0: *"Every
logical block's heat contributes to Artemis's K total. Reap must credit K
back. Alloc must charge K correctly."* Today nothing does, and closing it is
not a small job:
- **The current model is not conservative by construction.** `BLK-ALLOC` and
`BLK-FETCH` mint `Q.1` from nothing, `ART-COOL` destroys heat, `BLK-FREE`
zeroes it. 22,998 blocks at `Q.1` sums to 22,998.0, not 1.0.
- **`K-FLEET`, `K-LOCAL@`, and `K-CONSERVED?` no longer exist** — deleted in
commit `9323f776` along with `fleet-k.4th` and `compudynamics.4th`. The BAM
acceptance doc's "wire `ART-K-TOTAL` into `K-FLEET`" item refers to words
that are gone. Fleet conservation today is C-side only
(`vm_physics_conserved()`, 5% tolerance, `capsule_vm_physics.c:456`).
- Proper normalization is the **Logical BAM**, which is `FUTURE MATERIAL` in
`ARTEMIS.md` and gated behind Captain Bob explicitly reopening it.
Contrast with VM-fleet physics, which *is* conservative by construction via a
single balanced primitive `vm_physics_transfer(from, to, amount)` — the sum
invariant holds by induction rather than by enforcement. Blocks have no
equivalent primitive. Designing one is real work.
**Stale text to fix:** `VM-PHYSICS-DYNAMIC-FLEET-DESIGN-20260705.md`'s
"Fold-in question" section states that conservation does *not* generalize to
messages or blocks. The 2026-07-09 corrections on both sibling docs overturn
this, citing `.claude/ARTEMIS.md` and `.claude/HERMES.md` as authoritative.
The corrections win; the stale passage should be amended so it stops
contradicting them.
---
## Open questions — answer before writing code
**O1.** `Q-DECAY` lives in block **4110, marked ★ HARD LOCKED** in
`MANIFEST.md`. The block physics design doc says delete it. The manifest says
don't touch that block. Which wins — delete, or leave it and shadow it?
**O2.** Linear-subtractive or multiplicative decay for blocks? The word model
is linear; Artemis is currently multiplicative. This determines whether reap
ever fires.
**O3.** Idle-VM TTL. If Hermes executes no words it accrues no ticks, so
messages never expire. Accept this (TTL means "ticks of that VM's own
execution"), or does message aging need a different clock source? Note that
any answer involving a second clock conflicts with the one-clock rule.
**O4.** `BLK-FETCH` currently *resets* heat to `Q.1`. Change to increment
(frequency-sensitive, matches words, matches "how frequently" as the stated
signal) or keep reset (recency-only)?
**O5.** Confirm the C99-mechanics / FORTH-surface split still holds. D1 —
and therefore the whole no-bridge-needed argument — depends on it.
**O6.** `ARTEMIS_HEAT_WINDOW_DEPTH` (proposed 64) and the concrete slope-fit
function are both explicitly unresolved in the design docs. Start with the
proposed values and instrument, or settle them up front?
---
## Known risk
`ARTEMIS-BAM-ACCEPTANCE-20260703.md` records an **unfixed riscv64
heartbeat/compile race**: identical binaries alternately pass and fail
`TRIPOD-TEST` with `execute_colon_word: NULL cell in 'K-PUSH' after
'(start)'`, discriminated by boot-time heartbeat tick alignment (DoE row 59
reading `...,107,10,33,...` passing versus `...,107,9,32,...` failing).
This plan adds work inside `vm_tick()`, which shifts tick alignment. **Expect
to encounter this race.** It was reported and never fixed — "awaiting Captain
Bob's direction."
---
## Verification
1. `build/tools/mkcapsule --lint capsules` → 26 files, 0 violations.
2. Hosted build clean, zero warnings under `-Wall -Werror`.
3. Three-arch QEMU acceptance — `make -f Makefile.starkernel ARCH=<arch> clean
qemu` for amd64, aarch64, riscv64, one at a time, foreground. This is the
only acceptance authority for kernel changes.
4. **Behavioural proof that decay is live:** allocate blocks, execute
unrelated Artemis words to advance Artemis's own tick count *without ever
calling `ART-TICK` manually*, then confirm heat has dropped and cold blocks
have reaped. This is impossible today — nothing ages — so it is a genuine
discriminator.
5. **Slope responds to real churn** (the design doc's own test): drive block
churn at two different rates across separate boots and confirm
`blk_decay_slope_q48` converges to different values rather than sitting at
its seed.
6. **`dict_hash` must stay byte-identical across all three architectures.**
Established baseline: Hera `0x83c2c109100e2ed6`, Artemis
`0x5284ea5cd0f9983c`, Hermes `0x4159dcb326d79759`, Mama
`0xc88c3c1db6ef601b`. Any divergence means a wall-clock dependency has
crept in — precisely the bug class rev t of the fleet doc fixed.
## Not in scope
- Logical BAM / Physical BAM split, thermal zones (Phase 3 blocker,
`FUTURE MATERIAL`, requires explicit reopening)
- ACL Phase 8 / PKI — standing instruction is **STOP before starting**
- Any commit without Captain Bob asking for one
---
*Governing documents for this work, to be read completely before starting:*
`.claude/ARTEMIS.md`, `.claude/TRIPOD.md`, `.claude/HERMES.md`,
`.claude/CLAUDE.md`, and the four sibling design docs named in the header.
docs/working/archive/superseded/2026-08-tripod/03-architecture-README.md
+22
View File
@@ -0,0 +1,22 @@
> **SUPERSEDED (docs triage 2026-10-14).** The index of the old docs/03-architecture/ folder; its parts moved (word-acl to docs/reference/, tripod here).
> Instead: docs/README.md. Moved here from `docs/03-architecture/README.md`.
# docs/03-architecture/
Architecture-constraint documents for LithosAnanke subsystems that are still
being designed or built. These are binding constraints for anyone (human or
Claude Code) touching the affected code, not general background reading.
- **[tripod/](tripod/)** — Architectural contracts for the three Tripod VMs
(Hera, Hermes, Artemis): roles, dispatch rules, the compudynamic K≡1.0
invariant, and explicit prohibitions.
- **[word-acl/](word-acl/)** — Design doc for the word-level ACL system
(`capsules/ACL.4th`, `capsules/zuse.4th`, `src/word_source/acl_words.c`).
## See also
- [`docs/working/archive/historical/2026-07-birthing/`](../birthing/) — VM birthing plan/status that implements
the Tripod contract described in `tripod/TRIPOD.md`.
- [`docs/working/archive/historical/2026-10-lithosananke-orientation.md`](../lithosananke/README.md) — kernel-branch
overview these designs build on.
- [Project root](../../README.md)
docs/working/archive/superseded/2026-08-tripod/ARTEMIS.md
+190
View File
@@ -0,0 +1,190 @@
> **SUPERSEDED (docs triage 2026-10-14).** The Tripod as Hera/Hermes/Artemis; the Hermes VM was removed (messaging moved into the kernel as kernel-Hermes) and the Tripod is Hera/Artemis/Hestia.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/03-architecture/tripod/ARTEMIS.md`.
# ARTEMIS.md — Artemis VM Architecture
# StarshipOS / StarForth — Captain Bob (Robert Allan James)
# This document constrains Claude Code behavior. Read it completely before touching Artemis.
---
## Language Constraint — Non-Negotiable
All Artemis implementation is in StarForth dialect ONLY.
This includes:
- Data structures
- Constants
- Variables
- Control flow
- Everything
C99 is forbidden unless explicitly blocked in StarForth AND explicit written permission
is given by Captain Bob for that specific construct. "I could not figure out how to do
this in StarForth" is not permission. Ask first. Wait for the answer.
If you are about to write a C struct, a #define, or a C variable — stop.
Implement it in StarForth or ask.
---
## When Stuck — Non-Negotiable
If you are stuck, STOP.
Do not confabulate.
Do not invent dictionary words that do not exist.
Do not fabricate primitives.
Do not guess at behavior.
Do not work around the problem silently.
STOP. Report exactly what is blocking you. Wait for instruction.
Silent confabulation has caused git resets on a project with a hard patent deadline.
It is not acceptable under any circumstances.
---
## What Artemis Is
Artemis is memory. She owns the external attached virtual disk image and everything
that lives on it. She manages block storage compudynamically.
Artemis does **not** move messages — that is Hermes.
Artemis does **not** make VM lifecycle decisions — that is Hera.
Artemis does **not** own the internal ramdisk — that is kernel territory.
---
## Storage Topology
### Internal Ramdisk — Not Artemis Scope
- Baked into each VM at build time by mkcapsule.c
- LBN 0–3071
- Kernel territory
- Every VM carries its own ramdisk
- Artemis does not touch it, manage it, or know about it
### External Attached Virtual Disk Image — Artemis Territory
- Mounted virtual disk image, attached to the QEMU instance
- Artemis owns this exclusively
- LBN boundaries to be determined as Artemis is built
- Nothing persists between runs yet — that is a future chapter
### Nothing Persists Between Runs Yet
Persistence across runs is a future chapter. Do not implement it yet.
Do not design around it. The current model is: VM starts, Artemis initializes
her free map, VM stops, everything is gone. That is correct for now.
---
## What Artemis Owns
- The free map over her LBN range on the external disk image
- Block fetch and persist operations on the external disk image
- ACL lists — structured as Artemis-managed blocks (see ACL section)
- Cold capsule store — evicted from Hermes cache (future chapter)
- Block metadata
---
## Compudynamic Block Lifecycle
Artemis manages blocks compudynamically. Same model as everything else in StarshipOS:
- Block enters active use — born hot
- Cools over time with each heartbeat tick
- Frequently accessed blocks stay warm (heat refreshed on access)
- Cold blocks are candidates for eviction or reuse
- Block reap must rebalance K correctly
Artemis is the hardest VM to design because she has three interleaved
compudynamic concerns:
1. The free map — which LBNs are available
2. Block thermal state — how hot is each active block
3. K≡1.0 — fleet conservation across all block operations
These must be designed together. Do not implement one without the others.
Do not begin implementation without explicit instruction from Captain Bob.
---
## ACL — Access Control Lists
### Current State
ACL is a toggle. It is currently OFF.
A `\` comment disables ACL enforcement. When commented out, zuse is the
implicit and only user. No enforcement occurs.
```forth
\ ACL-ENFORCE ( disabled — zuse only )
```
### Structure
ACL lists must be structured as Artemis-managed blocks from day one.
Even before Artemis is fully built, ACL data structures must assume
they will live in Artemis-managed storage. No shortcuts that require
redesign later.
### When ACL Turns On
ACL enforcement becomes mandatory when VMs outside the Tripod arrive.
Until then the toggle stays off.
When enforcement is enabled:
- Every VM has an identity
- Every operation is checked against the ACL
- Artemis is the authority — she holds the lists
### Security Model
Security grows with the threat surface. Not before it.
The architecture is always structurally correct — enforcement is simply
not active yet. No retrofit needed when the time comes.
Toggling ACL on must require zero structural changes to the codebase.
Only the `\` comment is removed. If toggling ACL requires anything else,
the implementation is wrong.
---
## What Is Not Artemis
- **Internal ramdisk** — kernel territory, not Artemis scope
- **Message routing** — that is Hermes
- **VM lifecycle** — that is Hera
- **Persistence across runs** — future chapter, not now
- **Cold capsule eviction from Hermes** — future chapter, not now
---
## Compudynamic Invariant
Every block Artemis manages participates in K≡1.0.
Block fetch, persist, and reap operations must rebalance K correctly.
Artemis's thermal contribution to the fleet is the sum of her active block heat.
Do not write code that breaks K≡1.0 and add a comment explaining why it's okay.
---
## Current Scope
Artemis is not yet started. Before any implementation begins:
1. Captain Bob must give explicit instruction to begin
2. The three compudynamic concerns (free map, block thermal state, K≡1.0)
must be designed together before any code is written
3. ACL data structures must be Artemis-block-shaped from first line of code
4. Storage boundaries (LBN ranges on external disk image) must be confirmed
**Do not begin implementation without explicit instruction.**
Future chapters (do not implement):
- Persistence across runs
- Cold capsule store from Hermes
- Full ACL enforcement
- LBN range expansion
---
*This document is authoritative. If it conflicts with something in the codebase,
the codebase is wrong.*
docs/working/archive/superseded/2026-08-tripod/HERMES.md
+253
View File
@@ -0,0 +1,253 @@
> **SUPERSEDED (docs triage 2026-10-14).** The Tripod as Hera/Hermes/Artemis; the Hermes VM was removed (messaging moved into the kernel as kernel-Hermes) and the Tripod is Hera/Artemis/Hestia.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/03-architecture/tripod/HERMES.md`.
# HERMES.md — Hermes VM Architecture
# StarshipOS / StarForth — Captain Bob (Robert Allan James)
# This document constrains Claude Code behavior. Read it completely before touching Hermes.
---
## Language Constraint — Non-Negotiable
All Hermes implementation is in StarForth dialect ONLY.
This includes:
- Data structures
- Constants
- Variables
- Control flow
- Everything
C99 is forbidden unless explicitly blocked in StarForth AND explicit written permission
is given by Captain Bob for that specific construct. "I could not figure out how to do
this in StarForth" is not permission. Ask first. Wait for the answer.
If you are about to write a C struct, a #define, or a C variable — stop.
Implement it in StarForth or ask.
---
## What Hermes Is
Hermes is the messenger. He moves messages between VMs and manages channels.
That is his complete contract.
Hermes does **not** store persistent data — that is Artemis.
Hermes does **not** make lifecycle decisions — that is Hera.
Hermes does **not** route by load or heat — dispatch is capability-based, always.
Capsules are baked into the OS blob by mkcapsule.c at build time.
Hermes has no capsule cache, no fetch logic, no repository concern.
That is a future chapter.
---
## The Event Loop
Hermes runs a compudynamic event loop. Every message is a compudynamic object:
- Born hot (full heat at creation)
- Cools each heartbeat tick
- ACK received → clean death, K redistributed to fleet
- NACK received → requeue with reduced heat, or reaped if heat floor reached
- TTL expires (heat → 0) → Hermes reaps the message, K rebalanced
There is no special timeout logic. The physics handles it.
A message that nobody answers simply cools to death.
---
## Message Structure
Every message carries:
```
type — work type tag (determines recipient)
sender — originating VM identity
recipient — destination VM identity
payload — message body
heat — current thermal value (compudynamic)
sequence — monotonic sequence number
```
Point-to-point only. Sender addresses recipient directly.
Hermes delivers and tracks thermal state. He does not inspect payload.
---
## ACK/NACK Protocol
- ACK → message dies cleanly, K redistributed
- NACK → message requeued with reduced heat
- No response → message cools naturally to death via TTL
The binary receipt of a response IS the ACK where applicable.
No separate acknowledgement frame needed in those cases.
---
## PubSub — Channels
Hermes implements a publish/subscribe model. Topics are called **channels**.
### The Common Channel
There is one permanent channel: `COMMON`.
- Always exists, always hot
- Any VM can publish to it
- Used for negotiation and channel establishment only
- Not for business transactions — those happen on ephemeral channels
### Ephemeral Channels
Ephemeral channels are compudynamic objects — born hot, reaped when cold or closed.
**Channel structure:**
```
channel_id — owner VM identity + monotonic sequence number
owner — VM that created the channel (always the responder)
members[] — participant list
heat — compudynamic, born hot at creation
state — NEGOTIATING | OPEN | CLOSING
```
### The Negotiation Protocol
```
1. X publishes request on COMMON — born hot, TTL ticking
2. Y sees request, responds yes/no on COMMON
— No response → message cools to death → implicit NACK to X
3. Y accepts → Y creates ephemeral channel
→ Y publishes channel_id to X over COMMON
4. X and Y conduct business on ephemeral channel
5. Business concluded → owner Y reaps channel → K rebalanced
```
Y owns channel creation and destruction. Always the responder, never the requester.
A channel dying of cold (abandoned transaction) is not an error — the physics handles it.
### Multi-Party Channels
Channels are bilateral by default. The owner may invite additional VMs:
```
1. Owner publishes invite to candidate VM on COMMON
2. Candidate accepts or declines on COMMON — same TTL/cool protocol
3. Accept → candidate added to members[]
4. Decline or timeout → candidate never joins, channel continues bilateral
```
Multi-party is designed for but not implemented yet.
Do not implement invite logic until explicitly instructed.
The members[] structure must exist from the start to avoid future redesign.
### Channel Identity
`channel_id` = owner VM identity + monotonic sequence number.
Unforgeable by non-owners. Hermes mints it. No VM constructs its own channel_id.
### Compudynamic Invariant for Channels
Channels participate in K≡1.0.
An open channel with active messages holds heat.
Channel reap must redistribute K correctly to the fleet.
A channel cannot be reaped while messages on it are still in flight —
drain first, then reap.
---
## What Is Not Hermes
- **Capsule cache** — future chapter, not now
- **Remote fetch** — future chapter, not now
- **UDP server / client** — future chapter, not now
- **Instance authentication** — future chapter, not now
- **Persistent block storage** — that is Artemis
- **VM lifecycle decisions** — that is Hera
- **Heat-based routing** — wrong model, never implement
---
## Compudynamic Invariant
Every object Hermes manages — messages and channels — participates in K≡1.0.
Fleet K includes the thermal contribution of in-flight messages and open channels.
Hermes reaping a message or channel must rebalance K correctly.
Do not write code that breaks K≡1.0 and add a comment explaining why it's okay.
---
## Current Scope
1. Compudynamic event loop — message emit, drain, TTL, ACK/NACK, reap
2. COMMON channel — permanent, always hot, negotiation only
3. Ephemeral channels — compudynamic lifecycle, owner=responder, bilateral default
4. Channel negotiation protocol — request on COMMON, Y creates, Y reaps
5. members[] structure — present from the start, multi-party invite not yet implemented
Capsule management is a future chapter.
Remote repository is a future chapter.
Multi-party channel invite is a future chapter.
---
## v1 Block Map — LOCKED
Blocks 4110–4113 are Artemis. NEVER touch them.
```
4100 Constants: event codes, channel states, node sizes, arena sizes
4101 Arena CREATE: MSG-ARENA CH-ARENA MBR-ARENA; free-list vars; MSG-SEQ CH-ACTIVE
4102 MSG-INIT-FREE + CH-INIT-FREE
4103 MBR-INIT-FREE + MSG-ALLOC + MSG-FREE-NODE
4104 CH-ALLOC + CH-FREE-NODE + MBR-ALLOC + MBR-FREE-NODE
4105 Message field accessors: MSG-TYPE@/! MSG-FROM@/! MSG-TO@/! MSG-PADDR@/! MSG-PLEN@/! MSG-HEAT@/! MSG-SEQ@/!
4106 Channel+member accessors: CH-ID@/! CH-OWNER@/! CH-STATE@/! CH-HEAT@/! CH-MBRS@/! CH-NEXT@/! MBR-NEXT@ MBR-VM@
4107 IDX>NAME + MSG-DELIVER + MSG-SEND
4108 VARIABLE MSG-SCAN + MSG-COOL-ONE + MSG-COOL-ALL
4109 MSG-REAP
---- 4110–4113: ARTEMIS — DO NOT TOUCH ----
4114 VARIABLE CH-SCAN + CH-COOL-ALL
4115 CH-REAP-SAFE
4116 VARIABLE COMMON-CH + COMMON-INIT + HERMES-TICK
4117 EVENT-EMIT + EVENT-WAIT + EVENT-DRAIN (backward compat)
4118 HERA-NOTIFY-SPAWN + HERA-NOTIFY-KILL + HERA-DISPATCH-ONE + HERA-DISPATCH
4119 CH-REQUEST + CH-ACCEPT + CH-CLOSE
4120 WELCOME + CD-INIT
```
### Node layouts (cells)
**Message node — 7 cells:**
- 0: type (in-use) / next-free ptr (free)
- 1: sender VM index
- 2: recipient VM index
- 3: payload addr (FORTH string addr)
- 4: payload len
- 5: heat (Q48.16)
- 6: seq
**Channel node — 6 cells:**
- 0: channel-id (in-use) / next-free ptr (free)
- 1: owner VM index
- 2: state (0=NEGOTIATING 1=OPEN 2=CLOSING)
- 3: heat (Q48.16)
- 4: members-head (→ member list)
- 5: next-active (active channel list link)
**Member node — 2 cells:**
- 0: next-ptr
- 1: vm-id
### Physics
Cooling constant: Q-DECAY (65208) from compudynamics.4th — same as VM-DECAY-ONE.
Messages and VMs cool at identical rates. Fleet is thermodynamically consistent.
---
*This document is authoritative. If it conflicts with something in the codebase,
the codebase is wrong.*
docs/working/archive/superseded/2026-08-tripod/README.md
+31
View File
@@ -0,0 +1,31 @@
> **SUPERSEDED (docs triage 2026-10-14).** The Tripod as Hera/Hermes/Artemis; the Hermes VM was removed (messaging moved into the kernel as kernel-Hermes) and the Tripod is Hera/Artemis/Hestia.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/03-architecture/tripod/README.md`.
# docs/working/archive/superseded/2026-08-tripod/
Architectural constraints for the three StarForth VMs that make up the
Tripod. These documents are binding — read the relevant one in full before
touching any Tripod code. Violating the mythological contract they describe
is treated as an architectural error, not a style preference.
- **[TRIPOD.md](TRIPOD.md)** — The Tripod as a whole: Hera (VM-HERA=0,
governor — spawns/manages/reaps VMs), Hermes (VM-HERMES=1, messenger —
moves events), Artemis (VM-ARTEMIS=2, memory — owns block storage).
Covers capability-based dispatch (no heat-based routing), the CD-TICK
heartbeat, the K≡1.0 compudynamic conservation invariant, the DoE test
fixture's role, and explicit prohibitions (no heat-based VM routing, no
primitives added without instruction, no `BLOCK` calls outside the
0–3071 ramdrive range).
- **[HERMES.md](HERMES.md)** — Hermes VM architecture: the messenger role.
- **[ARTEMIS.md](ARTEMIS.md)** — Artemis VM architecture: the memory/block-
storage role.
All three documents share a non-negotiable constraint: Tripod implementation
is StarForth dialect only — no C99 without Captain Bob's explicit written
permission for that specific construct.
## See also
- [`docs/working/archive/historical/2026-07-birthing/`](../../birthing/) — birthing plan and QEMU acceptance
logs for the Hera-spawns-Hermes/Artemis constellation described here.
- [Project root](../../../README.md)
docs/working/archive/superseded/2026-08-tripod/TRIPOD.md
+191
View File
@@ -0,0 +1,191 @@
> **SUPERSEDED (docs triage 2026-10-14).** The Tripod as Hera/Hermes/Artemis; the Hermes VM was removed (messaging moved into the kernel as kernel-Hermes) and the Tripod is Hera/Artemis/Hestia.
> Instead: docs/fabric/FABRIC-3.5.md and FABRIC-3.6.md. Moved here from `docs/03-architecture/tripod/TRIPOD.md`.
# TRIPOD.md — Architectural Constraints for the Tripod VMs
# StarshipOS / StarForth — Captain Bob (Robert Allan James)
# This document constrains Claude Code behavior. Read it completely before touching any Tripod code.
---
## The Tripod: What It Is
Three StarForth VMs constitute the Tripod. They are not generic workers. Each holds a
specific architectural contract expressed by its mythological name. Violating the
mythological contract is an architectural error, not a style preference.
| VM | ID | Contract |
|------------|-------------|-----------------------------------------------|
| Hera | VM-HERA=0 | Governor. Spawns, manages, and reaps VMs. |
| Hermes | VM-HERMES=1 | Messenger. Moves events between VMs. |
| Artemis | VM-ARTEMIS=2| Memory. Owns block storage and persistence. |
**Hera is queen. She does not do work. She governs who exists and who doesn't.**
**Hermes moves messages. He does not store anything.**
**Artemis holds things. She does not route anything.**
If you find yourself putting event dispatch logic in Artemis, or storage logic in Hermes,
or work execution logic in Hera — stop. You have the wrong VM.
---
## The Immediate Goal
Hera must spawn Hermes and Artemis at boot, and all three must cooperate
compudynamically under K≡1.0 conservation.
That is the complete scope. Nothing else.
Acceptance criteria:
1. Hera boots and initializes her own compudynamic state.
2. Hera spawns Hermes. Hermes initializes his own compudynamic state.
3. Hera spawns Artemis. Artemis initializes her own compudynamic state.
4. All three VMs are running. Fleet K = sum of individual VM K values. K≡1.0 holds.
5. Hera can reap (cleanly terminate) a child VM.
6. The system is stable across a long-running test workload.
---
## Compudynamics: Dual Role — Read This Carefully
Compudynamics serves TWO roles in this system. Do not conflate them.
### Role 1: Intra-VM Physics Engine
Inside each VM, Compudynamics drives the adaptive heartbeat:
- 7-loop rolling window
- K conservation arithmetic
- Phase operators on the heartbeat signal
Every VM — Hera, Hermes, Artemis — runs its own compudynamic engine internally.
This is autonomous. No VM manages another VM's internal heartbeat.
### Role 2: Hera's VM Lifecycle Driver
Hera observes the fleet K picture and uses it to make **lifecycle decisions**:
- Is a child VM healthy?
- Is a child VM starved?
- Does a child VM need to be spawned, suspended, or reaped?
Hera does NOT use fleet K to route work. She uses it to govern existence.
**The fleet K view is lifecycle telemetry. It is not a dispatch mechanism.**
---
## Dispatch: Capability-Based, Always
Work routing is determined by work type. It is not determined by heat, K value,
VM load, or any compudynamic signal.
| Work type | Destination |
|-----------------|-------------|
| Block I/O | Artemis |
| Event emit/drain| Hermes |
| VM lifecycle | Hera |
`VM-EXEC` takes a VM name. The caller names the correct VM for the work type.
There is no routing algorithm. There is no dispatch table keyed on heat.
There is no "send to the hottest VM." That model is wrong and must not be implemented.
---
## CD-TICK: What It Is and Is Not
CD-TICK advances the compudynamic heartbeat. That is all it does.
CD-TICK is **not**:
- A scheduler
- A timeslicer
- A run-queue manager
- A work dispatcher
- A router
If you are calling CD-TICK to decide where work goes, you are using it wrong.
---
## The DoE (Design of Experiments) Test Fixture
The DoE campaign (PHASE6-TEST, K-conservation CSV logging, 38,400-run Latin Square)
is a **stress and measurement fixture**. It exists to:
- Validate K≡1.0 conservation under long-running diverse workloads
- Generate telemetry for the SSRN paper
- Provide a long-runtime acceptance test for the compudynamic physics
The DoE is **not**:
- A description of what the system does in production
- A feature
- A model for how work gets dispatched
`DOE-WORK` must never appear in the production dispatch path.
If you see `DOE-WORK` being dispatched by anything other than the test fixture, remove it.
The DoE is a good test precisely because it is long-running and diverse.
It exercises the physics. It does not define the system.
---
## What Not To Do — Explicit Prohibitions
1. **Do not implement heat-based VM routing.** It causes starvation. It is wrong.
2. **Do not pin Hera's heat floor.** That is a band-aid on a wrong model.
3. **Do not put DOE-WORK in the production dispatch path.**
4. **Do not call CD-TICK a scheduler** in comments, docs, or variable names.
5. **Do not call the fleet K view a router** in comments, docs, or variable names.
6. **Do not add primitives to StarForth without explicit instruction.**
NIP is not registered. TUCK is not registered. Assume nothing is available
that is not in the verified primitive table. Use SWAP DROP instead of NIP.
7. **Do not call BLOCK on LBNs outside the ramdrive range (0–3071).**
LBN 4501–4599 are Artemis metadata space. They are not in the kernel ramdrive.
Use ART-STATUS to touch the free map. Do not call ART-SELF-TEST in acceptance tests.
8. **Do not exceed 64 bytes per line or 16 lines per block in capsule files.**
mkcapsule.c enforces this at build time. Violations will fail the build.
---
## The Invariant
K≡1.0 is conserved across the fleet at all times.
Fleet K = sum of all VM K values.
This is a physics constraint, not a software convention.
Do not write code that breaks it and then add a comment explaining why it's okay.
---
*This document is authoritative. If it conflicts with something in the codebase,
the codebase is wrong.*
---
## Language Constraint — Non-Negotiable
All Tripod implementation is in StarForth dialect ONLY.
This includes:
- Data structures
- Constants
- Variables
- Control flow
- Everything
C99 is forbidden unless explicitly blocked in StarForth AND explicit written permission
is given by Captain Bob for that specific construct. "I could not figure out how to do
this in StarForth" is not permission. Ask first. Wait for the answer.
If you are about to write a C struct, a #define, or a C variable — stop.
Implement it in StarForth or ask.
---
## When Stuck — Non-Negotiable
If you are stuck, STOP.
Do not confabulate.
Do not invent dictionary words that do not exist.
Do not fabricate primitives.
Do not guess at behavior.
Do not work around the problem silently.
STOP. Report exactly what is blocking you. Wait for instruction.
Silent confabulation has caused git resets on a project with a hard patent deadline.
It is not acceptable under any circumstances.
docs/working/archive/superseded/L8_WIRING_PLAN.md
+358
View File
@@ -0,0 +1,358 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09: the plan for wiring L8, since built.
> Instead: docs/working/experiments/02-experiments/L8_WIRING_IMPLEMENTATION_SUMMARY.md. Moved here from `docs/working/experiments/02-experiments/L8_WIRING_PLAN.md`.
<!-- Moved from docs/02-experiments/L8_WIRING_PLAN.md to docs/working/archive/superseded/L8_WIRING_PLAN.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** The wiring plan was executed; superseded by
> `L8_WIRING_IMPLEMENTATION_SUMMARY.md` and `L8_WIRING_CODE_REVIEW_GUIDE.md`. Kept as a
> record of the plan.
# L8 Control Wiring Implementation Plan
**Status**: Infrastructure exists but execution paths are not gated
**Date**: 2025-12-05
**Issue**: L8 Jacquard mode selector computes 4-bit control signals (L2/L3/L5/L6) but feedback loops ignore them
---
## Executive Summary
The L8 Steady State Machine (SSM) infrastructure is **90% complete** but the final 10% is critical:
✅ **Complete**:
- L8 state machine (`ssm_jacquard.c`) computes 4-bit mode (C0-C15) based on entropy/CV/temporal metrics
- `ssm_apply_mode()` sets L2/L3/L5/L6 flags in `vm->ssm_config`
- `vm_heartbeat_update_l8()` updates L8 state every heartbeat cycle (lines 993-1045 in `vm.c`)
- Mode transitions logged with debug info
❌ **Missing**:
- Feedback loop execution paths don't check `ssm_config` flags before running
- L2/L3/L5/L6 run unconditionally despite L8 setting mode bits
- Result: L8 is a "supervisor without authority"
---
## Current Architecture (From DoE Analysis)
From `Makefile:112-127` and `ssm_jacquard.h:32-51`:
| Loop | Name | Status | Control |
|------|------|--------|---------|
| L1 | Heat Tracking | ALWAYS OFF | Disabled (harmful in 86% of configs) |
| L2 | Rolling Window | **NEEDS GATING** | L8 bit 3 (entropy-driven) |
| L3 | Linear Decay | **NEEDS GATING** | L8 bit 2 (temporal locality) |
| L4 | Pipelining Metrics | ALWAYS OFF | Disabled (harmful in 100% of configs) |
| L5 | Window Inference | **NEEDS GATING** | L8 bit 1 (CV-driven) |
| L6 | Decay Inference | **NEEDS GATING** | L8 bit 0 (CV + temporal) |
| L7 | Adaptive Heartrate | ALWAYS ON | Enabled (beneficial in 71% of configs) |
| L8 | Jacquard Selector | **IMPLEMENTED** | 4-bit mode controller (C0-C15) |
**Key Insight**: L1 and L4 are compile-time disabled. L7 is always-on. **L2/L3/L5/L6 need runtime gating** based on L8 mode bits.
---
## Required Changes (4 Locations)
### 1. L2: Rolling Window Recording (src/rolling_window_of_truth.c)
**Current Code** (`rolling_window_record_execution()` line 181):
```c
int rolling_window_record_execution(RollingWindowOfTruth* window, uint32_t word_id)
{
if (!window || !window->execution_history)
return -1;
window->execution_history[window->window_pos] = word_id;
// ... always executes
}
```
**Required Change**:
Add check at function entry or call site. Two options:
**Option A** (Preferred - Clean separation):
Modify call site in `src/vm.c:1652`:
```c
// Before (line 1652):
rolling_window_record_execution(&vm->rolling_window, word_id);
// After:
if (vm->ssm_config && ((ssm_config_t*)vm->ssm_config)->L2_rolling_window) {
rolling_window_record_execution(&vm->rolling_window, word_id);
}
```
**Option B** (Alternative - Gate inside function):
Modify `rolling_window_record_execution()` to accept VM pointer and check internally.
**Files to modify**:
- `src/vm.c:1652` - Call site in `execute_word()`
- `src/rolling_window_of_truth.c:181` - Consider adding VM* parameter for internal check
- `include/rolling_window_of_truth.h` - Update function signature if Option B chosen
---
### 2. L3: Linear Decay Application (src/physics_metadata.c)
**Current Code** (line 176):
```c
/* L8 FINAL INTEGRATION: L3 linear decay is always-on */
void physics_metadata_apply_linear_decay(DictEntry *entry, uint64_t elapsed_ns, VM *vm) {
if (!entry || !vm) {
return;
}
/* Frozen words do not decay */
if (entry->flags & WORD_FROZEN) {
return;
}
/* Don't bother with insignificant time intervals */
if (elapsed_ns < DECAY_MIN_INTERVAL) {
return;
}
// ... decay logic always executes
}
```
**Required Change**:
Add L3 gate check after frozen/interval checks:
```c
/* L8 FINAL INTEGRATION: L3 linear decay - runtime gated by Jacquard */
void physics_metadata_apply_linear_decay(DictEntry *entry, uint64_t elapsed_ns, VM *vm) {
if (!entry || !vm) {
return;
}
/* Frozen words do not decay */
if (entry->flags & WORD_FROZEN) {
return;
}
/* Don't bother with insignificant time intervals */
if (elapsed_ns < DECAY_MIN_INTERVAL) {
return;
}
/* L8 GATE: Check if L3 (linear decay) is enabled by Jacquard mode */
if (vm->ssm_config) {
ssm_config_t *config = (ssm_config_t*)vm->ssm_config;
if (!config->L3_linear_decay) {
return; /* L3 disabled by L8 mode selector */
}
}
// ... existing decay logic
}
```
**Files to modify**:
- `src/physics_metadata.c:176` - Add L3 gate check (line ~188-193)
- Comment update at line 175 to reflect runtime gating
**Call sites** (no changes needed, gate is internal):
- `src/vm.c:737` - `vm_tick_apply_background_decay()` (heartbeat loop)
- Other call sites inherit the gate
---
### 3. L5: Window Width Inference (src/inference_engine.c)
**Current Code** (line 538):
```c
/* === PHASE 2C: Window Width Inference === */
/* L8 FINAL INTEGRATION: Loop #5 always-on */
uint32_t inferred_width = find_variance_inflection(
trajectory,
traj_len,
current_variance
);
```
**Required Change**:
Gate inference execution with conditional block:
```c
/* === PHASE 2C: Window Width Inference === */
/* L8 FINAL INTEGRATION: Loop #5 runtime-gated by Jacquard */
uint32_t inferred_width = outputs->adaptive_window_width; /* Keep previous value as default */
if (inputs->vm && inputs->vm->ssm_config) {
ssm_config_t *config = (ssm_config_t*)inputs->vm->ssm_config;
if (config->L5_window_inference) {
/* L5 enabled - run inference */
inferred_width = find_variance_inflection(
trajectory,
traj_len,
current_variance
);
}
/* else: L5 disabled, keep previous adaptive_window_width */
}
```
**Files to modify**:
- `src/inference_engine.c:538-544` - Add L5 conditional gate
- Comment update at line 539 to reflect runtime gating
---
### 4. L6: Decay Slope Inference (src/inference_engine.c)
**Current Code** (line 546):
```c
/* === PHASE 2D: Decay Slope Inference === */
/* L8 FINAL INTEGRATION: Loop #6 always-on */
uint64_t inferred_slope = infer_decay_slope_q48(trajectory, traj_len);
```
**Required Change**:
Gate inference execution with conditional block:
```c
/* === PHASE 2D: Decay Slope Inference === */
/* L8 FINAL INTEGRATION: Loop #6 runtime-gated by Jacquard */
uint64_t inferred_slope = outputs->adaptive_decay_slope; /* Keep previous value as default */
if (inputs->vm && inputs->vm->ssm_config) {
ssm_config_t *config = (ssm_config_t*)inputs->vm->ssm_config;
if (config->L6_decay_inference) {
/* L6 enabled - run inference */
inferred_slope = infer_decay_slope_q48(trajectory, traj_len);
}
/* else: L6 disabled, keep previous adaptive_decay_slope */
}
```
**Files to modify**:
- `src/inference_engine.c:546-548` - Add L6 conditional gate
- Comment update at line 547 to reflect runtime gating
---
## Implementation Order (Recommended)
1. **L3 (Decay)** - Simplest, internal to one function
- Modify `src/physics_metadata.c:176-220`
- Add gate check after line 188
- Test with mode C0 (all off) vs C4 (L3 on)
2. **L5 (Window Inference)** - Medium complexity
- Modify `src/inference_engine.c:538-544`
- Preserve previous value when disabled
- Test early-exit behavior with mode C0 vs C2
3. **L6 (Decay Inference)** - Similar to L5
- Modify `src/inference_engine.c:546-548`
- Preserve previous value when disabled
- Test with mode C0 vs C1
4. **L2 (Rolling Window)** - Most complex (data structure)
- Decide on Option A (call-site gate) vs Option B (internal gate)
- Modify call site `src/vm.c:1652` OR function signature
- Test with mode C0 vs C8 (high entropy)
---
## Testing Strategy
### Unit Tests
For each loop, verify:
1. **Mode C0 (0000)**: All loops disabled → function returns early or skips logic
2. **Single-bit modes**: Enable one loop at a time (C1, C2, C4, C8)
3. **Top 5% modes**: C4, C7, C9, C11, C12 work correctly
### Integration Tests
1. **Mode transitions**: L8 changes mode → loops respond correctly
2. **Hysteresis**: Mode doesn't flap rapidly (5-tick delay)
3. **Metrics**: Verify entropy/CV/temporal drive correct mode selection
### Regression Tests
1. Run existing 936+ test suite with different L8 modes
2. Verify deterministic behavior (no variance across runs)
3. Check performance matches DoE predictions
---
## Validation Checklist
After implementation, verify:
- [ ] L2: `rolling_window_record_execution()` only executes when L2 bit set
- [ ] L3: `physics_metadata_apply_linear_decay()` returns early when L3 bit clear
- [ ] L5: `find_variance_inflection()` only called when L5 bit set
- [ ] L6: `infer_decay_slope_q48()` only called when L6 bit set
- [ ] L8: Mode transitions logged in debug output
- [ ] All existing tests pass with default mode (C0 or C12)
- [ ] DoE experiment can reproduce 128 configurations via L8 mode forcing
---
## Impact Assessment
**Lines of code changed**: ~40-60 lines across 3 files
**Risk**: Low (adding guards, not changing logic)
**Performance**: Neutral to positive (skips work when disabled)
**Correctness**: High confidence (L8 already validated via DoE)
---
## Open Questions
1. **Default mode**: Should VM initialize with C0 (minimal) or C12 (DoE optimal)?
- Current: `ssm_l8_init(..., SSM_MODE_C0)` in `vm.c:387`
- Top 5%: C4, C7, C9, C11, C12
- Recommendation: Start C0, let L8 adapt
2. **Mode override**: Should FORTH words exist to force L8 mode?
- `L8-MODE!` to set mode manually
- `L8-MODE@` to query current mode
- Useful for debugging/benchmarking
3. **Metrics exposure**: Should heartbeat CSV export include L8 mode?
- Currently exports: executions, heat, variance, slope
- Add: `l8_mode` column (0-15 integer)
---
## Success Criteria
**Definition of Done**:
1. All 4 feedback loops check `ssm_config` flags before executing
2. L8 mode selector controls L2/L3/L5/L6 runtime behavior
3. Mode transitions visible in debug logs
4. Existing test suite passes with default L8 mode
5. Can reproduce DoE configurations by forcing L8 modes 0-15
**Verification**:
```bash
# Build with L8 control wired
make clean && make fastest
# Test mode C0 (all off) - should be fastest, least adaptive
./build/amd64/fastest/starforth --benchmark 1000
# Test mode C15 (all on) - should be slowest, most adaptive
# (requires word to set L8 mode or environment variable)
# Run full test suite
make test
```
---
## Next Steps
**Before coding**:
1. Review this plan with stakeholders
2. Decide on L2 implementation strategy (Option A vs B)
3. Choose default L8 mode (C0 vs C12)
**During implementation**:
1. Follow order: L3 → L5 → L6 → L2
2. Test each loop individually before moving to next
3. Update comments to reflect runtime gating
**After completion**:
1. Run full DoE validation (128 configs × 300 reps)
2. Verify 0% algorithmic variance maintained
3. Document L8 control API for future developers
docs/working/archive/superseded/factorial-doe/guide.md
+265
View File
@@ -0,0 +1,265 @@
> **SUPERSEDED (docs triage 2026-10-14).** The 2^6 factorial DoE guide, flagged obsolete 2026-10-09.
> Instead: experiments/bare_metal/README.md for the DoE as it runs now. Moved here from `docs/working/experiments/02-experiments/factorial-doe/guide.md`.
<!-- Moved from docs/02-experiments/factorial-doe/guide.md to docs/working/archive/superseded/factorial-doe/guide.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ OBSOLETE (flagged 2026-10-09).** Describes the retired per-loop Makefile-toggle
> architecture, replaced by the L8 Jacquard runtime mode selector. Kept for historical record.
# StarForth Complete 2^6 Factorial Design of Experiments
## Overview
**One-and-done comprehensive factorial testing** - collect all data systematically, analyze separately.
This guide describes how to run a **complete 2^6 factorial design of experiments** across all 6 feedback loops, yielding:
- **64 unique configurations** (all binary combinations)
- **1,920+ runs** (30 samples per configuration)
- **Single unified dataset** for factorial analysis
## The Why: Factorial DoE
Instead of incrementally tuning parameters, we collect **comprehensive experimental data** in one go:
### Traditional Approach (Old)
- Run baseline
- Run with Loop #1
- Run with Loops #1-2
- Run with Loops #1-3
- ... (sequential, incremental)
- **Problem:** Can't see interactions, trade-offs, or surprises
### Factorial Approach (New)
- Run ALL 2^6 = 64 combinations in randomized order
- Collect all data at once
- Analyze factorial effects, interactions, and main effects
- **Advantage:** Complete picture of how loops interact
## The 6 Feedback Loops
Each loop is a binary toggle in the Makefile:
| Loop | Makefile Flag | Purpose |
|------|---|---|
| **#1** | `ENABLE_LOOP_1_HEAT_TRACKING` | Count word execution frequency |
| **#2** | `ENABLE_LOOP_2_ROLLING_WINDOW` | Capture execution history (circular buffer) |
| **#3** | `ENABLE_LOOP_3_LINEAR_DECAY` | Age words by exponential decay over time |
| **#4** | `ENABLE_LOOP_4_PIPELINING_METRICS` | Track word-to-word transitions |
| **#5** | `ENABLE_LOOP_5_WINDOW_INFERENCE` | Infer optimal window width (Levene's test) |
| **#6** | `ENABLE_LOOP_6_DECAY_INFERENCE` | Infer decay slope via linear regression |
## Configuration Naming
Each of 64 configurations is named by its loop status:
```
L1_L2_L3_L4_L5_L6
Where each digit is 0 (OFF) or 1 (ON)
Examples:
000000 = Baseline (all loops OFF - pure FORTH-79)
000001 = Only heat tracking
111110 = All except decay inference
111111 = All loops ON (likely optimal)
101010 = Alternating pattern (for research)
```
## Quick Start
### Minimal Test (validate script)
```bash
./scripts/run_factorial_doe.sh --runs-per-config 1 TEST_RUN
# 64 builds × 1 run = 64 total runs
# Runtime: ~30-45 minutes
# Output: /home/rajames/CLionProjects/StarForth-DoE/experiments/TEST_RUN/
```
### Standard Comprehensive Run
```bash
./scripts/run_factorial_doe.sh --runs-per-config 30 2025_11_19_FACTORIAL
# 64 builds × 30 runs = 1,920 total runs
# Runtime: ~2-4 hours
# Output: /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FACTORIAL/
```
### High-Precision Run
```bash
./scripts/run_factorial_doe.sh --runs-per-config 100 HIGH_PRECISION_DOE
# 64 builds × 100 runs = 6,400 total runs
# Runtime: ~6-12 hours
# Output: /home/rajames/CLionProjects/StarForth-DoE/experiments/HIGH_PRECISION_DOE/
```
## Output Structure
```
/home/rajames/CLionProjects/StarForth-DoE/experiments/<EXPERIMENT_LABEL>/
├── experiment_results.csv # All measurements (1920+ rows)
├── experiment_summary.txt # Timing, metadata, statistics
├── configuration_manifest.txt # All 64 configs documented
├── test_matrix.txt # Randomized execution order
└── run_logs/
├── 000000_run_1.log
├── 000000_run_2.log
├── 000001_run_1.log
├── ...
└── 111111_run_30.log # 1,920 total log files
```
## CSV Schema
Each row in `experiment_results.csv` contains:
```
timestamp,configuration,run_number,
<35+ metrics from test harness>,
enable_loop_1_heat_tracking,
enable_loop_2_rolling_window,
enable_loop_3_linear_decay,
enable_loop_4_pipelining_metrics,
enable_loop_5_window_inference,
enable_loop_6_decay_inference
```
This lets you correlate every metric against every loop combination.
## Analysis Strategy
After collection, analyze in the separate `StarForth-DoE` repository:
### Step 1: Load Data
```python
import pandas as pd
df = pd.read_csv("experiment_results.csv")
print(df.shape) # Should be (1920, 38+)
print(df.columns)
```
### Step 2: Separate Metrics from Factors
```python
factor_cols = ['enable_loop_1_heat_tracking', ..., 'enable_loop_6_decay_inference']
metric_cols = ['total_lookups', 'cache_hit_percent', ..., 'vm_workload_duration_ns_q48']
factors = df[factor_cols]
metrics = df[metric_cols]
```
### Step 3: Factorial Analysis
```python
# Main effects: What's the average change when Loop X goes from 0→1?
main_effects = {}
for loop in factor_cols:
main_effects[loop] = metrics[factors[loop]==1].mean() - metrics[factors[loop]==0].mean()
# Interactions: Do loops work better/worse together?
# Can use statsmodels.formula.api for ANOVA/regression
# Optimal subset: Which combinations of loops give best throughput?
best_configs = df.nlargest(10, 'vm_workload_duration_ns_q48')
```
## Key Design Choices
### 1. **Rebuild Every Pass** ✓
- Each configuration requires a `make clean && make`
- Ensures clean, isolated state
- No cross-contamination
- True ground truth for each variant
### 2. **Randomized Execution**
- All 1,920 runs randomized into single matrix (not 64 separate runs)
- Prevents thermal ramp, temporal bias, or systematic trends
- DoE principle: randomization eliminates confounds
### 3. **Full Factorial** (not fractional)
- Testing ALL 2^6 = 64 combinations
- Can detect unexpected interactions
- Most expensive, most informative
### 4. **One-and-Done Collection**
- Single experiment, all data
- No progressive tuning during collection
- Separates measurement from analysis (good practice)
## Expected Results
### Hypothetical Output
Assuming typical performance distribution:
```
Configuration | Avg Time (ns) | Speed vs Baseline
─────────────────────────────────────────────────
000000 (base) | 10,000,000 | 1.0x
000001 (L1) | 9,950,000 | 1.005x
111111 (all) | 8,500,000 | 1.176x ← Likely best
```
### Expected Patterns
1. **Loop #1 (heat tracking)**: ~2-5% improvement alone
2. **Loop #2 (rolling window)**: ~1-3% improvement (depends on L1)
3. **Loops #3-6**: Increasingly synergistic (work better together)
4. **Interactions**: Loop #5 + #6 likely synergistic (inference loops)
## Troubleshooting
### Build Fails for a Configuration
```bash
# Check specific config build log
cat /home/rajames/CLionProjects/StarForth-DoE/experiments/<LABEL>/run_logs/build_<CONFIG>.log
```
### Test Harness Crashes on Some Configs
- Check if loop combination is invalid (e.g., Loop #5 without Loop #2)
- Verify Makefile flags are correctly parsed
- See: `include/vm.h` for interdependencies
### Results Look Suspicious
- Check `test_matrix.txt` - ensure randomization is present
- Verify CSV row counts match expected total
- Check for missing runs in log directory
## Performance Notes
### Timing Estimates
- **Minimal (1 run/config)**: 30-45 minutes
- **Standard (30 runs/config)**: 2-4 hours
- **High-precision (100 runs/config)**: 6-12 hours
- **Depends on:** CPU speed, disk I/O, test harness overhead
### Optimization
- Run overnight for high-precision experiments
- Use `screen` or `tmux` to keep session alive
- Monitor disk space (~100 MB for results + logs)
## Next Steps
1. **Run the experiment:**
```bash
./scripts/run_factorial_doe.sh 2025_11_19_COMPLETE_FACTORIAL
```
2. **Monitor progress:**
```bash
tail -f /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_COMPLETE_FACTORIAL/experiment_summary.txt
```
3. **Move to analysis repo** (once complete):
```bash
cp -r /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_COMPLETE_FACTORIAL \
/home/rajames/CLionProjects/StarForth-DoE-Analysis/data/
cd /home/rajames/CLionProjects/StarForth-DoE-Analysis
python3 analyze_factorial.py data/2025_11_19_COMPLETE_FACTORIAL/experiment_results.csv
```
4. **Generate reports:**
- Main effects plots
- Interaction matrices
- Pareto optimal configurations
- Recommendations for production builds
---
**Version:** 1.0
**Last Updated:** 2025-11-19
**Author:** Claude Code (with Robert A. James)
docs/working/archive/superseded/factorial-doe/incompatible-configs.md
+404
View File
@@ -0,0 +1,404 @@
> **SUPERSEDED (docs triage 2026-10-14).** The 2^6 factorial DoE guide, flagged obsolete 2026-10-09.
> Instead: experiments/bare_metal/README.md for the DoE as it runs now. Moved here from `docs/working/experiments/02-experiments/factorial-doe/incompatible-configs.md`.
<!-- Moved from docs/02-experiments/factorial-doe/incompatible-configs.md to docs/working/archive/superseded/factorial-doe/incompatible-configs.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ OBSOLETE (flagged 2026-10-09).** Build-time incompatibility handling for the retired 2^6
> toggle experiment, replaced by the L8 Jacquard mode selector. Kept for historical record.
# Handling Incompatible Configurations in Factorial DoE
## Overview
Your intuition was correct: **some loop combinations can't coexist**. Rather than failing the entire experiment, the updated script treats incompatibility as valuable data.
This document explains how the script detects and records incompatible configurations during the factorial experiment.
---
## What is an Incompatible Configuration?
An incompatible configuration is a combination of feedback loop toggles that:
1. **Fails to compile** - Makefile target produces build errors
2. **Causes StarForth to crash** - Binary exits abnormally (segfault, signal, etc.)
3. **Produces invalid output** - Test harness runs but produces no metrics
Examples (hypothetical):
- Loop 5 (window inference) without Loop 2 (rolling window) = no data to infer from
- Loop 6 (decay inference) without Loop 3 (linear decay) = no slope to measure
- Certain combinations trigger memory conflicts or uninitialized states
---
## How the Script Handles It
### Detection Strategy
**Build Phase:**
```
Configuration X attempted
├─ make TARGET=fastest ENABLE_LOOP_1=1 ...ENABLE_LOOP_6=0
├─ Check exit code
├─ If success: mark as "OK", proceed to testing
└─ If failure: mark as "BUILD_FAILED", skip all runs for this config
```
**Execution Phase:**
```
Run configuration X, iteration N
├─ Execute: ./starforth --doe-experiment
├─ Check exit code
├─ Try to extract metrics from output
├─ If success: mark as "OK", record metrics
└─ If failure: mark as "CRASH", record incompatibility
```
### CSV Recording
Incompatible configurations are recorded with status flags:
**Successful run (viable config):**
```
timestamp,configuration,run_number,build_status,run_status,
2025-11-19T10:30:45,1_1_0_1_0_0,5,OK,OK,
<35+ metrics>,1,1,0,1,0,0
```
**Build failure (incompatible):**
```
timestamp,configuration,run_number,build_status,run_status,
2025-11-19T10:35:20,0_1_1_1_1_0,1,BUILD_FAILED,<skipped>,
<empty>,0,1,1,1,1,0
```
**Runtime crash (incompatible):**
```
timestamp,configuration,run_number,build_status,run_status,
2025-11-19T10:40:15,1_0_0_1_1_1,3,OK,CRASH,
<empty>,1,0,0,1,1,1
```
### Experiment Flow
The experiment **continues** when incompatibility is detected:
```
Run 1: Config A (OK) → Test completes, metrics recorded
Run 2: Config B (BUILD_FAILED) → Build fails, skip all runs for B, continue
Run 3: Config A (OK) → Test completes, metrics recorded
Run 4: Config C (CRASH) → Crashes, record as incompatible, continue
...
Run N: Config D (OK) → Test completes, metrics recorded
SUMMARY:
✓ Successful runs: 847/1920
✗ Failed configs: 17 (incompatible combinations detected)
✓ Experiment completed despite incompatibilities
```
---
## CSV Schema
### Two New Columns
| Column | Values | Meaning |
|--------|--------|---------|
| `build_status` | OK, BUILD_FAILED | Did Makefile compilation succeed? |
| `run_status` | OK, CRASH | Did StarForth execute without crashing? |
### Analysis Filter Rules
```python
import pandas as pd
df = pd.read_csv('experiment_results.csv')
# Only viable configurations
viable = df[(df['build_status'] == 'OK') & (df['run_status'] == 'OK')]
# Incompatible configs (at compile time)
build_failures = df[df['build_status'] == 'BUILD_FAILED']
# Incompatible configs (at runtime)
runtime_crashes = df[df['run_status'] == 'CRASH']
# All incompatible
incompatible = df[(df['build_status'] != 'OK') | (df['run_status'] != 'OK')]
```
---
## Interpreting Results
### Experiment Summary
After completion, you'll see:
```
═══════════════════════════════════════════════════════════
EXPERIMENT COMPLETE (with incompatibility detection)
═══════════════════════════════════════════════════════════
✓ Successful runs: 1847/1920
✓ Failed configurations (incompatible): 17
→ Incompatible configurations detected:
✗ 0_1_1_1_1_0 (BUILD_FAILED)
✗ 1_0_0_1_1_1 (CRASH)
✗ 0_1_0_1_1_1 (CRASH)
... (total 17)
→ These configs are marked in CSV with build_status or run_status
→ They represent configurations where loops cannot coexist without conflict
```
### What This Means
- **1847 successful runs** = viable, measurable configurations
- **17 incompatible** = design space constraints discovered
- **Experiment completed** = data collection succeeded despite incompatibilities
- **CSV includes both** = full picture of parameter space
---
## Analysis Examples
### Example 1: Summary by Viability
```python
import pandas as pd
df = pd.read_csv('experiment_results.csv')
# Count by status
viable = len(df[(df['build_status'] == 'OK') & (df['run_status'] == 'OK')])
build_fail = len(df[df['build_status'] == 'BUILD_FAILED'].drop_duplicates('configuration'))
runtime_crash = len(df[df['run_status'] == 'CRASH'].drop_duplicates('configuration'))
print(f"Viable configurations: {viable} runs")
print(f"Build failures: {build_fail} configurations")
print(f"Runtime crashes: {runtime_crash} configurations")
print(f"Total tested: {64} configurations, {64*30} runs")
print(f"Success rate: {viable}/{len(df)} = {100*viable/len(df):.1f}%")
```
Expected output:
```
Viable configurations: 1847 runs
Build failures: 5 configurations
Runtime crashes: 12 configurations
Total tested: 64 configurations, 1920 runs
Success rate: 1847/1920 = 96.2%
```
### Example 2: Identify Problematic Loop Combinations
```python
# Find configs that always crash
crashes = df[df['run_status'] == 'CRASH'].drop_duplicates('configuration')
print("Configurations that crash:")
for config in crashes['configuration']:
parts = config.split('_')
loops_on = [f"Loop{i+1}" for i, bit in enumerate(parts) if bit == '1']
print(f" {config}: {', '.join(loops_on) or 'None'}")
```
Expected output:
```
Configurations that crash:
0_1_1_1_1_0: Loop2, Loop3, Loop4, Loop5
1_0_0_1_1_1: Loop1, Loop4, Loop5, Loop6
0_1_0_1_1_1: Loop2, Loop4, Loop5, Loop6
... etc
```
**Insight:** Loop 5 (window inference) appears in many crashes → suggests window inference has prerequisites or conflicts.
### Example 3: Feasibility Matrix
```python
# Which loop combinations are feasible?
df_viable = df[(df['build_status'] == 'OK') & (df['run_status'] == 'OK')].copy()
# For each config, mark as viable (1) or not (0)
feasible = {}
for config in df['configuration'].unique():
is_viable = len(df_viable[df_viable['configuration'] == config]) > 0
feasible[config] = 1 if is_viable else 0
# Show as binary matrix
print("Feasibility (1=viable, 0=incompatible):")
for i in range(64):
config = f"{i%2}_{(i//2)%2}_{(i//4)%2}_{(i//8)%2}_{(i//16)%2}_{(i//32)%2}"
print(f"{config}: {'✓' if feasible[config] else '✗'}")
```
Result: Feasibility map of 64 configurations for maintenance and design decisions.
---
## Maintenance Insights
### Why Some Combinations Are Incompatible
Common reasons:
1. **Missing Prerequisites:**
- Loop 5 (window inference) needs Loop 2 (rolling window) data
- Loop 6 (decay inference) needs Loop 3 (linear decay) to measure
2. **Memory/Resource Conflicts:**
- Multiple loops allocating same memory region
- Circular dependencies in initialization
3. **Semantic Contradictions:**
- Loop X disables feature that Loop Y requires
- Incompatible macro definitions
4. **Bugs in Specific Combinations:**
- Race conditions appearing only with certain feature combinations
- Uninitialized state when paths not taken
### Using This Information
**For Future Development:**
```
Findings: Loop 5 incompatible without Loop 2
→ Check: Does window inference code assume rolling window exists?
→ Action: Either add guard (if !LOOP_5 without LOOP_2) or document prerequisite
→ Test: Add compile-time or runtime check
```
**For Optimization:**
```
Result: All configs with Loop 6 crash if Loop 1 is OFF
→ Issue: Decay inference needs heat tracking for slope calculation
→ Fix: Add LOOP_1_REQUIRED_BY_LOOP_6 guard in code
→ Verify: Re-run factorial (expect fewer crashes)
```
**For Production:**
```
Feasible subset: {0_0_0_0_0_0, 1_1_0_0_0_0, 1_1_1_0_0_0, 1_1_1_1_1_1}
→ Recommendation: Use 1_1_1_1_1_1 (all compatible, best performance)
→ Fallback: Use 1_1_0_0_0_0 (compatible, simpler)
→ Baseline: Use 0_0_0_0_0_0 (no feedback, minimal overhead)
```
---
## Next Steps
1. **Run the experiment** to completion (incompatibilities are discovered organically)
2. **Analyze the CSV** to identify which configs are viable
3. **Examine incompatible configs** to understand why they fail
4. **Document prerequisites** (which loops require others)
5. **Fix or guard** the code to eliminate incompatibilities
6. **Re-run** experiment to validate fixes
---
## Technical Details
### Build Failure Detection
```bash
if make TARGET=fastest ENABLE_LOOP_1=1 ... > build_log 2>&1; then
# Build successful
BUILD_STATUS="OK"
else
# Build failed - mark incompatible
BUILD_STATUS="BUILD_FAILED"
# Skip all runs for this config
continue
fi
```
### Runtime Crash Detection
```bash
if "${BINARY}" --doe-experiment > run_log 2>&1; then
# Binary executed (exit code 0)
if extract_csv_row(run_log) succeeds:
RUN_STATUS="OK"
else:
RUN_STATUS="CRASH" # Binary ran but no output
else
# Binary crashed (non-zero exit)
RUN_STATUS="CRASH"
fi
```
### CSV Recording
```bash
# Viable run
printf '%s,%s,%s,%s,%s,%s,%s\n' \
"${timestamp}" "${config}" "${run_num}" \
"OK" "OK" \
"${csv_row}" "${flags}" >> results_run_01_2025_12_08.csv
# Incompatible run
printf '%s,%s,%s,%s,%s,%s\n' \
"${timestamp}" "${config}" "${run_num}" \
"${build_status}" "${run_status}" \
"${flags}" >> results_run_01_2025_12_08.csv
```
---
## Example: Real Incompatibility
**Scenario:** Loop 5 (window inference) without Loop 2 (rolling window)
```c
// Loop 5: Window Inference Code
void vm_tick_inference_engine(VM *vm) {
// Try to infer window width from rolling window
RollingWindowOfTruth *window = &vm->rolling_window; // LOOP_2 data
// BUG: If LOOP_2 is OFF, window is uninitialized!
uint32_t window_pos = window->window_pos; // CRASH: uninitialized read
}
```
**Symptom in CSV:**
```
Configuration 0_0_0_0_1_0 marked as CRASH
Loop1=0, Loop2=0, Loop3=0, Loop4=0, Loop5=1, Loop6=0
↑ ↑
Missing prerequisite Problematic loop
```
**Fix:**
```c
// Add guard in code
#if ENABLE_LOOP_5_WINDOW_INFERENCE
#if !ENABLE_LOOP_2_ROLLING_WINDOW
#error "Loop 5 requires Loop 2: window inference needs rolling window data"
#endif
#endif
```
**Result after fix:** Configuration 0_0_0_0_1_0 removed from viable set (prevented at compile time), reduces crash count.
---
## Summary
| Aspect | Approach |
|--------|----------|
| **Detection** | Build exit codes + runtime crash detection |
| **Recording** | New `build_status` and `run_status` columns |
| **Handling** | Skip incompatible config, continue experiment |
| **Impact** | Partial dataset (viable configs) + incompatibility map |
| **Analysis** | Filter by status, identify patterns |
| **Action** | Fix code, re-run, measure improvement |
This design enables discovering incompatibilities without sacrificing experimental continuity.
---
**Version:** 1.0
**Last Updated:** 2025-11-19
**Status:** Integrated into run_factorial_doe.sh
docs/working/archive/superseded/factorial-doe/index.md
+271
View File
@@ -0,0 +1,271 @@
> **SUPERSEDED (docs triage 2026-10-14).** The 2^6 factorial DoE guide, flagged obsolete 2026-10-09.
> Instead: experiments/bare_metal/README.md for the DoE as it runs now. Moved here from `docs/working/experiments/02-experiments/factorial-doe/index.md`.
<!-- Moved from docs/02-experiments/factorial-doe/index.md to docs/working/archive/superseded/factorial-doe/index.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ OBSOLETE (flagged 2026-10-09).** Index for the retired 2^6 factorial toggle
> architecture, replaced by the L8 Jacquard mode selector. Kept for historical record.
# StarForth Complete 2^6 Factorial DoE - Documentation Index
## Overview
This is a **one-and-done comprehensive factorial design of experiments** to systematically test all 64 possible combinations of 6 feedback loops.
**Key insight:** Don't toggle back-and-forth. Instead, systematically test ALL 2^6 combinations in a single experiment, then analyze the results afterward.
## Documentation Structure
### 1. **START HERE** 📋
- **File:** `docs/QUICK_START_FACTORIAL_DoE.md`
- **Purpose:** Entry point for quick reference
- **Read time:** 5 minutes
- **Contains:**
- 3 command options (quick/standard/high-precision)
- Time estimates
- What to expect during execution
- Basic troubleshooting
### 2. **Complete Guide** 📚
- **File:** `docs/FACTORIAL_DoE_GUIDE.md`
- **Purpose:** Full design documentation and rationale
- **Read time:** 20-30 minutes
- **Contains:**
- Why factorial DoE (vs. incremental tuning)
- 6 feedback loop definitions
- Configuration naming scheme (0_0_0_0_0_0 → 1_1_1_1_1_1)
- Design choices explained
- Expected results and patterns
- Troubleshooting guide
- Performance notes
### 3. **Analysis Workflow** 🔍
- **File:** `docs/FACTORIAL_DoE_ANALYSIS_WORKFLOW.md`
- **Purpose:** End-to-end pipeline from collection to insights
- **Read time:** 15-20 minutes (or 1-2 hours if coding along)
- **Contains:**
- Phase 1: Data collection (run script)
- Phase 2: Data transfer (to analysis repo)
- Phase 3: Statistical analysis (main effects, interactions)
- Phase 4: Reporting (generating insights)
- Python code examples:
* Load data and validate
* Compute main effects (individual loop impacts)
* Detect interactions (loop synergies)
* Optimize configuration subset
* Visualization code
- Example report structure
- Checklist for completion
### 4. **Main Script** ⚙️
- **File:** `scripts/run_factorial_doe.sh`
- **Purpose:** Executable for running the entire experiment
- **Read time:** 10 minutes (code + comments)
- **Contains:**
- Generates all 64 configurations
- Rebuilds for each unique configuration
- Randomizes execution order (1,920+ runs)
- Collects metrics to CSV
- Full instrumentation and progress reporting
---
## Reading Guide by Use Case
### "I just want to run the experiment"
1. Read: `QUICK_START_FACTORIAL_DoE.md` (5 min)
2. Execute: Choose command option (quick/standard/high-precision)
3. Wait: 30 min - 12 hours depending on option
4. Done!
### "I want to understand the design before running"
1. Read: `QUICK_START_FACTORIAL_DoE.md` (5 min)
2. Read: `FACTORIAL_DoE_GUIDE.md` (30 min)
3. Execute: Run with confidence
4. Analyze: Follow `FACTORIAL_DoE_ANALYSIS_WORKFLOW.md`
### "I need to analyze the results"
1. Read: `QUICK_START_FACTORIAL_DoE.md` → understand naming
2. Read: `FACTORIAL_DoE_ANALYSIS_WORKFLOW.md` → statistical methods
3. Code along: Use Python examples in Analysis Workflow
4. Interpret: Reference `FACTORIAL_DoE_GUIDE.md` for expected patterns
### "I want complete understanding"
Read in order:
1. `QUICK_START_FACTORIAL_DoE.md` (overview)
2. `FACTORIAL_DoE_GUIDE.md` (design rationale)
3. `scripts/run_factorial_doe.sh` (implementation)
4. `FACTORIAL_DoE_ANALYSIS_WORKFLOW.md` (analysis methods)
---
## Quick Command Reference
### Run Experiment
```bash
# Option A: Quick validation (30 min)
./scripts/run_factorial_doe.sh --runs-per-config 1 TEST_RUN
# Option B: Standard production (2-4 hours)
./scripts/run_factorial_doe.sh --runs-per-config 30 2025_11_19_FULL_FACTORIAL
# Option C: High-precision overnight (6-12 hours)
./scripts/run_factorial_doe.sh --runs-per-config 100 HIGH_PRECISION_FACTORIAL
```
### Monitor Progress
```bash
# Check results_run_01_2025_12_08 CSV growth
watch -n 5 'wc -l /path/to/experiment_results.csv'
# View summary
tail /path/to/experiment_summary.txt
# Count completed configurations
ls /path/to/run_logs/build_*.log | wc -l
```
### Analyze Results
```bash
# Load and analyze in Python
cd /path/to/StarForth-DoE-Analysis
python3 << 'EOF'
import pandas as pd
df = pd.read_csv('data/2025_11_19_FULL_FACTORIAL/experiment_results.csv')
print(f"Loaded {len(df)} runs")
# See FACTORIAL_DoE_ANALYSIS_WORKFLOW.md for full analysis code
EOF
```
---
## Key Concepts
### 64 Configurations
Each configuration is a unique combination of 6 binary factors:
- **0_0_0_0_0_0** = Baseline (all loops OFF)
- **1_0_0_0_0_0** = Only heat tracking
- **1_1_0_0_0_0** = Heat + rolling window
- ...
- **1_1_1_1_1_1** = All loops ON (current optimal)
### Randomized Execution
All 1,920+ runs (64 configs × 30 samples) are randomized into a single execution matrix.
- Eliminates temporal bias
- Prevents thermal ramp (if CPU heats during night)
- Proper DoE methodology
### One-and-Done Collection
Collect all data in single experiment, analyze separately.
- No tuning during collection (avoid bias)
- Single pass through parameter space
- Separates measurement from analysis
### Main Effects
Individual impact of each loop:
```
Loop #1: Loop turns ON, average metric changes by X%
Loop #2: Loop turns ON, average metric changes by Y%
... etc
```
### Interactions
How loops work together (amplify or suppress each other):
```
Loop #1 + Loop #2: Do they work better together than independently?
```
---
## Files at a Glance
| File | Size | Purpose |
|------|------|---------|
| `scripts/run_factorial_doe.sh` | 19 KB | Main executable script |
| `docs/QUICK_START_FACTORIAL_DoE.md` | 4.9 KB | Quick reference |
| `docs/FACTORIAL_DoE_GUIDE.md` | 8.0 KB | Design documentation |
| `docs/FACTORIAL_DoE_ANALYSIS_WORKFLOW.md` | 9.6 KB | Analysis pipeline |
| `docs/FACTORIAL_DoE_INDEX.md` | This file | Documentation index |
---
## Execution Timeline
### Phase 1: Preparation (5-10 min)
- Read `QUICK_START_FACTORIAL_DoE.md`
- Choose execution option
- Run command
### Phase 2: Collection (30 min - 12 hours)
- Script builds all 64 configs (as needed)
- Randomized execution of 1,920+ runs
- Metrics collected to CSV
- Progress shown every run
### Phase 3: Transfer (5 min)
- Copy results to analysis repo
- Verify file integrity
### Phase 4: Analysis (1-2 hours)
- Load CSV in Python
- Compute main effects
- Detect interactions
- Find optimal configurations
- Generate visualizations
### Phase 5: Reporting (1-2 hours)
- Summarize findings
- Create report
- Recommendations for production
**Total time:** Collection (2-4 hrs) + Analysis (2-4 hrs) = **4-8 hours**
---
## The Design Philosophy
### Problem Statement
How do we systematically understand the performance contribution and interactions of 6 interdependent feedback loops?
### Solution
**Complete 2^6 factorial design of experiments**
### Why This Approach
1. **Complete:** Tests all 64 combinations (no gaps)
2. **Systematic:** Proper experimental design methodology
3. **Separable:** Collection ≠ Analysis (reduces bias)
4. **Reproducible:** Randomized execution, full logging
5. **Scalable:** Single script, extensible to more factors
### Alternatives Rejected
- ❌ Incremental tuning ("baseline → +loop1 → +loop1+2"): Misses interactions
- ❌ Dynamic toggle without rebuild: Introduces state contamination
- ❌ Partial factorial: Misses key interactions
- ❌ One-at-a-time testing: Expensive and confounded
---
## Next Steps
1. **Read** `QUICK_START_FACTORIAL_DoE.md` (5 minutes)
2. **Decide** which execution option (quick/standard/precise)
3. **Run** the command and let it collect data
4. **Monitor** progress in another terminal
5. **Analyze** results using `FACTORIAL_DoE_ANALYSIS_WORKFLOW.md`
6. **Report** findings to stakeholders
---
## Questions?
- **"How long will this take?"** → See `QUICK_START_FACTORIAL_DoE.md`
- **"What does configuration X mean?"** → See `FACTORIAL_DoE_GUIDE.md`
- **"How do I analyze the results?"** → See `FACTORIAL_DoE_ANALYSIS_WORKFLOW.md`
- **"Why this approach?"** → See `FACTORIAL_DoE_GUIDE.md` - "The Why"
- **"How do I run it?"** → `./scripts/run_factorial_doe.sh`
---
**Version:** 1.0
**Created:** 2025-11-19
**Status:** Ready for production use
docs/working/archive/superseded/factorial-doe/quick-start.md
+158
View File
@@ -0,0 +1,158 @@
> **SUPERSEDED (docs triage 2026-10-14).** Quick start for the same 2^6 factorial DoE, flagged obsolete 2026-10-09.
> Instead: experiments/bare_metal/README.md. Moved here from `docs/working/architecture/getting-started/quick-start/factorial-doe.md`.
<!-- Moved from docs/01-getting-started/quick-start/factorial-doe.md to docs/working/archive/superseded/factorial-doe/quick-start.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ OBSOLETE (flagged 2026-10-09).** Describes the retired per-loop `ENABLE_LOOP_1..6`
> compile-time toggle architecture and hardcoded local paths; the physics adaptation is now
> driven by the L8 Jacquard runtime mode selector. Kept for historical record.
# Quick Start: Complete 2^6 Factorial DoE
## TL;DR
Run a **complete, exhaustive 2^6 = 64 configuration experiment** in one go, collecting all data for later analysis.
## Three Command Options
### Option 1: Quick Validation (30 min)
Test the script and infrastructure:
```bash
./scripts/run_factorial_doe.sh --runs-per-config 1 QUICK_TEST
```
- 64 builds × 1 run = 64 total runs
- Fast feedback on whether setup works
- Output: `/home/rajames/CLionProjects/StarForth-DoE/experiments/QUICK_TEST/`
### Option 2: Standard Full DoE (2-4 hours)
Comprehensive factorial collection:
```bash
./scripts/run_factorial_doe.sh --runs-per-config 30 2025_11_19_FULL_FACTORIAL
```
- 64 builds × 30 runs = 1,920 total runs
- Good statistical power for main effects
- Standard production-grade experiment
- Output: `/home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/`
### Option 3: High-Precision Run (6-12 hours)
Maximum data for deep analysis:
```bash
./scripts/run_factorial_doe.sh --runs-per-config 100 HIGH_PRECISION_FACTORIAL
```
- 64 builds × 100 runs = 6,400 total runs
- Best for detecting subtle interactions
- Run overnight for best results
- Output: `/home/rajames/CLionProjects/StarForth-DoE/experiments/HIGH_PRECISION_FACTORIAL/`
## What Gets Generated
```
/home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/
├── experiment_results.csv ← All 1,920 rows of measurements
├── experiment_summary.txt ← Timing and metadata
├── configuration_manifest.txt ← All 64 configs listed
├── test_matrix.txt ← Randomized execution order
└── run_logs/
├── 0_0_0_0_0_0_run_1.log
├── 0_0_0_0_0_0_run_2.log
├── ...
└── 1_1_1_1_1_1_run_30.log
```
## What's Being Tested
6 **independent binary feedback loops**:
| Loop | Toggle | What It Does |
|------|--------|---|
| 1 | `ENABLE_LOOP_1_HEAT_TRACKING` | Count execution frequency |
| 2 | `ENABLE_LOOP_2_ROLLING_WINDOW` | Capture execution history |
| 3 | `ENABLE_LOOP_3_LINEAR_DECAY` | Age words over time |
| 4 | `ENABLE_LOOP_4_PIPELINING_METRICS` | Track word transitions |
| 5 | `ENABLE_LOOP_5_WINDOW_INFERENCE` | Infer window width |
| 6 | `ENABLE_LOOP_6_DECAY_INFERENCE` | Infer decay slope |
**All 2^6 = 64 combinations** tested:
- `0_0_0_0_0_0` = Baseline (all OFF)
- `1_0_0_0_0_0` = Only heat tracking
- `0_1_0_0_0_0` = Only rolling window
- ...
- `1_1_1_1_1_1` = All loops ON (current optimal)
## During Execution
You'll see progress like:
```
═══════════════════════════════════════════════════════════
STARFORTH COMPLETE 2^6 FACTORIAL DoE
═══════════════════════════════════════════════════════════
→ Factors: 6 feedback loops (LOOP_1 through LOOP_6)
→ Configurations: 2^6 = 64
→ Runs per config: 30
→ TOTAL RUNS: 1,920
→ Output: /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL
>>> Generating randomized test matrix...
✓ Test matrix generated
→ Run 1/1920 - config 0_1_0_1_1_0 #14...
✓ Run 1/1920 completed (3s)
→ Run 2/1920 - config 1_1_1_0_0_1 #2...
✓ Run 2/1920 completed (2s)
...
```
The script:
1. **Builds** each of 64 configurations (rebuilds when config changes)
2. **Runs** tests in randomized order (no systematic bias)
3. **Collects** metrics to CSV
4. **Outputs** progress every run
## After Completion
Move to analysis repo to understand results:
```bash
cd /home/rajames/CLionProjects/StarForth-DoE-Analysis
python3 analyze_factorial.py \
data/2025_11_19_FULL_FACTORIAL/experiment_results.csv
```
Expected analysis outputs:
- **Main effects:** Which individual loops help most?
- **Interactions:** Do loops work better/worse together?
- **Optimal subset:** Best combination of loops?
- **Trade-offs:** Performance vs complexity?
## Key Design Decisions
✅ **Rebuild every pass** - ensures clean, isolated testing
✅ **Randomized execution** - eliminates temporal/thermal bias
✅ **Complete factorial** - captures all interactions
✅ **One-and-done** - collect all data, analyze later
## Estimated Time
| Runs/Config | Total | Time | When |
|---|---|---|---|
| 1 | 64 | 30-45 min | Quick test |
| 30 | 1,920 | 2-4 hours | Standard |
| 100 | 6,400 | 6-12 hours | Overnight |
Times depend on CPU, disk speed, and test harness overhead.
## Example Command
**Run it now:**
```bash
cd /home/rajames/CLionProjects/StarForth
./scripts/run_factorial_doe.sh 2025_11_19_FULL_FACTORIAL
```
**Monitor progress:**
```bash
tail -f /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/experiment_summary.txt
```
---
**Ready?** Run the command above and let the experiment collect comprehensive data for downstream analysis.
docs/working/archive/superseded/factorial-doe/workflow.md
+300
View File
@@ -0,0 +1,300 @@
> **SUPERSEDED (docs triage 2026-10-14).** The 2^6 factorial DoE guide, flagged obsolete 2026-10-09.
> Instead: experiments/bare_metal/README.md for the DoE as it runs now. Moved here from `docs/working/experiments/02-experiments/factorial-doe/workflow.md`.
<!-- Moved from docs/02-experiments/factorial-doe/workflow.md to docs/working/archive/superseded/factorial-doe/workflow.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ HISTORICAL (flagged 2026-10-09).** Documents the completed 2025-11-19 factorial
> campaign whose data informed the current L8 design; the toggle architecture it runs on is
> retired. Kept as a record.
# Factorial DoE Analysis Workflow
## End-to-End Process
```
StarForth Repo
└─ ./scripts/run_factorial_doe.sh
│
├─ Builds all 64 configurations
├─ Runs 1,920+ tests (randomized)
└─ Outputs: experiment_results.csv + logs
│
└─→ Copy to StarForth-DoE-Analysis repo
│
├─ Load CSV into Python/R
├─ Separate factors from metrics
├─ Compute main effects & interactions
├─ Identify optimal subsets
└─ Generate reports & visualizations
```
## Phase 1: Data Collection (StarForth Repo)
### Step 1: Run Experiment
```bash
cd /home/rajames/CLionProjects/StarForth
# Quick validation (optional)
./scripts/run_factorial_doe.sh --runs-per-config 1 TEST_01
# Full production run
./scripts/run_factorial_doe.sh --runs-per-config 30 2025_11_19_FULL_FACTORIAL
```
### Step 2: Wait for Completion
Monitor in another terminal:
```bash
watch -n 5 'wc -l /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/experiment_results.csv'
```
Expected progression:
- 0 min: 1 line (header)
- 30 min: ~500 lines
- 2 hours: ~1,500 lines
- 2.5 hours: 1,921 lines (complete)
### Step 3: Verify Completion
```bash
tail /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/experiment_results.csv
echo "---"
cat /home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL/experiment_summary.txt
```
## Phase 2: Data Transfer & Setup
### Step 1: Copy to Analysis Repo
```bash
RESULTS_DIR="/home/rajames/CLionProjects/StarForth-DoE/experiments/2025_11_19_FULL_FACTORIAL"
ANALYSIS_REPO="/home/rajames/CLionProjects/StarForth-DoE-Analysis"
mkdir -p "$ANALYSIS_REPO/data"
cp -r "$RESULTS_DIR" "$ANALYSIS_REPO/data/2025_11_19_FULL_FACTORIAL"
```
### Step 2: Verify Transfer
```bash
ls -lah "$ANALYSIS_REPO/data/2025_11_19_FULL_FACTORIAL/"
wc -l "$ANALYSIS_REPO/data/2025_11_19_FULL_FACTORIAL/experiment_results.csv"
```
## Phase 3: Analysis (StarForth-DoE-Analysis Repo)
### Load Data (Python)
```python
import pandas as pd
import numpy as np
from scipy import stats
import matplotlib.pyplot as plt
# Load results_run_01_2025_12_08
df = pd.read_csv("data/2025_11_19_FULL_FACTORIAL/experiment_results.csv")
print(f"Loaded {len(df)} runs across {len(df.groupby('configuration'))} configurations")
# Separate factors and metrics
factors = df[['enable_loop_1_heat_tracking',
'enable_loop_2_rolling_window',
'enable_loop_3_linear_decay',
'enable_loop_4_pipelining_metrics',
'enable_loop_5_window_inference',
'enable_loop_6_decay_inference']]
metrics = df[['total_lookups', 'cache_hit_percent', 'vm_workload_duration_ns_q48', ...]]
print(f"Shape: {len(df)} rows × {len(df.columns)} columns")
print(f"Factors: {len(factors.columns)}")
print(f"Metrics: {len(metrics.columns)}")
```
### Main Effects Analysis
```python
# For each loop, compute average metric change when loop goes 0→1
main_effects = {}
for loop in factors.columns:
loop_name = loop.replace('enable_', '').replace('_', ' ').title()
metric_when_on = df[df[loop] == 1]['vm_workload_duration_ns_q48'].mean()
metric_when_off = df[df[loop] == 0]['vm_workload_duration_ns_q48'].mean()
effect = metric_when_on - metric_when_off
pct_change = 100 * (effect / metric_when_off)
main_effects[loop_name] = {
'effect': effect,
'pct_change': pct_change,
'on_mean': metric_when_on,
'off_mean': metric_when_off,
}
# Print sorted by impact
for loop, effect in sorted(main_effects.items(), key=lambda x: abs(x[1]['pct_change']), reverse=True):
print(f"{loop:30} {effect['pct_change']:+7.2f}% (ON: {effect['on_mean']:.0f}, OFF: {effect['off_mean']:.0f})")
```
Expected output:
```
Loop 6 Decay Inference +12.3% (ON: 8,800,000, OFF: 7,840,000)
Loop 5 Window Inference +8.5% (ON: 8,600,000, OFF: 7,920,000)
Loop 4 Pipelining Metrics +6.2% (ON: 8,400,000, OFF: 7,900,000)
Loop 3 Linear Decay +4.1% (ON: 8,200,000, OFF: 7,870,000)
Loop 2 Rolling Window +3.0% (ON: 8,100,000, OFF: 7,850,000)
Loop 1 Heat Tracking +2.1% (ON: 8,010,000, OFF: 7,850,000)
```
### Interaction Analysis
```python
# Two-way interactions: Do loops amplify/suppress each other?
interactions = {}
for i, loop_a in enumerate(factors.columns):
for j, loop_b in enumerate(factors.columns):
if i >= j:
continue # Only upper triangle
# Four cases: (0,0), (0,1), (1,0), (1,1)
case_00 = df[(df[loop_a] == 0) & (df[loop_b] == 0)]['vm_workload_duration_ns_q48'].mean()
case_01 = df[(df[loop_a] == 0) & (df[loop_b] == 1)]['vm_workload_duration_ns_q48'].mean()
case_10 = df[(df[loop_a] == 1) & (df[loop_b] == 0)]['vm_workload_duration_ns_q48'].mean()
case_11 = df[(df[loop_a] == 1) & (df[loop_b] == 1)]['vm_workload_duration_ns_q48'].mean()
# Interaction effect: (1,1) - (1,0) - (0,1) + (0,0)
interaction = (case_11 - case_10 - case_01 + case_00)
loop_a_name = loop_a.replace('enable_', '').replace('_', ' ')
loop_b_name = loop_b.replace('enable_', '').replace('_', ' ')
if abs(interaction) > 100000: # Only report large interactions
interactions[f"{loop_a_name} × {loop_b_name}"] = interaction
# Print sorted interactions
for pair, interaction in sorted(interactions.items(), key=lambda x: abs(x[1]), reverse=True):
print(f"{pair:40} {interaction:+12,.0f}")
```
### Optimal Configuration Search
```python
# Find best k configurations by performance metric
k = 10
best_configs = df.nlargest(k, 'vm_workload_duration_ns_q48')[
['configuration', 'vm_workload_duration_ns_q48', 'enable_loop_1_heat_tracking',
'enable_loop_2_rolling_window', 'enable_loop_3_linear_decay',
'enable_loop_4_pipelining_metrics', 'enable_loop_5_window_inference',
'enable_loop_6_decay_inference']
].drop_duplicates(subset=['configuration'])
print("Top 10 Configurations by Performance:")
print("─" * 100)
for idx, row in best_configs.iterrows():
loops = (
f"L1:{row['enable_loop_1_heat_tracking']} " +
f"L2:{row['enable_loop_2_rolling_window']} " +
f"L3:{row['enable_loop_3_linear_decay']} " +
f"L4:{row['enable_loop_4_pipelining_metrics']} " +
f"L5:{row['enable_loop_5_window_inference']} " +
f"L6:{row['enable_loop_6_decay_inference']}"
)
print(f"{row['configuration']:20} | {row['vm_workload_duration_ns_q48']:12,.0f} ns | {loops}")
```
### Visualization Example
```python
# Main effects plot
import matplotlib.pyplot as plt
fig, axes = plt.subplots(2, 3, figsize=(15, 10))
fig.suptitle('2^6 Factorial DoE Main Effects', fontsize=16)
loop_names = [f"Loop {i+1}" for i in range(6)]
metric_key = 'vm_workload_duration_ns_q48'
for idx, loop in enumerate(factors.columns):
ax = axes[idx // 3, idx % 3]
off_mean = df[df[loop] == 0][metric_key].mean()
on_mean = df[df[loop] == 1][metric_key].mean()
bars = ax.bar(['OFF', 'ON'], [off_mean, on_mean], color=['lightcoral', 'lightgreen'])
ax.set_ylabel('Workload Duration (ns)')
ax.set_title(f'{loop_names[idx]}')
ax.set_ylim([7000000, 10000000])
# Add value labels
for bar in bars:
height = bar.get_height()
ax.text(bar.get_x() + bar.get_width()/2., height,
f'{height/1e6:.1f}M', ha='center', va='bottom')
plt.tight_layout()
plt.savefig('main_effects.png', dpi=150)
print("Saved: main_effects.png")
```
## Phase 4: Reporting
### Key Outputs to Generate
1. **Main Effects Summary** (which loops matter most?)
- Table of individual loop impacts
- Sorted by effect size
2. **Interaction Matrix** (do loops amplify/suppress?)
- 6×6 table of two-way interactions
- Highlights synergies/conflicts
3. **Optimal Configuration** (best 10-20 subsets)
- Recommend top 3-5 production configs
- Show Pareto frontier (performance vs complexity)
4. **Statistical Summary**
- Variance by configuration
- Confidence intervals on estimates
- Regression model R²
### Example Report Structure
```markdown
# Factorial DoE Analysis Report - 2025-11-19
## Executive Summary
- Tested all 64 feedback loop combinations
- 1,920 experimental runs collected
- Key finding: Loops 5+6 show strong synergy (+18% combined)
## Main Effects
| Loop | Impact | Confidence | Recommendation |
|------|--------|-----------|---|
| Loop 6 (Decay Inference) | +12.3% | p<0.001 | **Enabled** |
| Loop 5 (Window Inference) | +8.5% | p<0.001 | **Enabled** |
| Loop 4 (Pipelining) | +6.2% | p<0.01 | Enabled |
| Loop 3 (Decay) | +4.1% | p<0.05 | Enabled |
| Loop 2 (Window) | +3.0% | p<0.05 | Enabled |
| Loop 1 (Heat) | +2.1% | p>0.05 | Optional |
## Recommended Configurations
1. **111111** (All loops): +30% vs baseline
2. **111110** (Except heat): +28% vs baseline, simpler
3. **011111** (Except heat, decay): +24% vs baseline
## Trade-off Analysis
- Full stack (111111): Best performance, highest code complexity
- Recommended (111110): 28% improvement, 95% of benefit
- Minimal set (001110): 18% improvement, 50% code overhead
---
```
## Checklist
- [ ] Run factorial DoE script to completion
- [ ] Verify 1,920 rows in experiment_results.csv
- [ ] Copy results to analysis repo
- [ ] Load data in Python/R
- [ ] Compute main effects (6 loops)
- [ ] Compute interactions (2-way or higher)
- [ ] Identify top 5-10 configurations
- [ ] Generate visualizations
- [ ] Write summary report
- [ ] Decide on production configuration
---
**Timeline:** Collection (2-4 hrs) + Analysis (2-4 hrs) = **6-8 hours total** for comprehensive factorial DoE.
docs/working/archive/superseded/feedback-loops-analysis.md
+365
View File
@@ -0,0 +1,365 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09.
> Instead: its own header. Moved here from `docs/working/architecture/03-architecture/physics-engine/feedback-loops-analysis.md`.
<!-- Moved from docs/03-architecture/physics-engine/feedback-loops-analysis.md to docs/working/archive/superseded/feedback-loops-analysis.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** Precursor bug-hunt audit covering six feedback loops;
> the canonical, current reference is `feedback-loops.md` (all seven loops, tuning knobs, and
> source files). Kept for historical record.
# StarForth Feedback Loops & Self-Optimization Analysis
## Executive Summary
StarForth implements **6 feedback loops** for automatic self-tuning during execution. Of these:
- **3 fully wired AND operational** ✅ (actively driving optimization decisions)
1. ✅ Hot-words cache promotion (1.78× speedup measured)
2. ✅ Rolling window adaptive shrinking (FIXED 2025-11-08, enforcement now works)
4. ✅ Pipelining transition speculation (WIRED 2025-11-08, speculative prefetch active)
- **1 wired AND utilized, but NOT validated** ⚠️ (running, but unsupported design choices)
3. ⚠️ Heat decay of execution history (arbitrary linear function, no tuning knob, unvalidated)
- **1 wired but NOT utilized** ⏳ (Phase 2 experimental, infrastructure ready)
5. Context-aware window tuning (binary chop algorithm stub, needs VM-level metrics aggregation)
- **1 observability-only** (metrics collected for formal verification)
6. Physics metadata temperature tracking
**CURRENT STATUS:**
- A_BASELINE, A_B_CACHE, A_C_FULL, A_B_C_FULL configurations now test complete optimization stack:
- Loop #1 (cache) enabled via A_B_CACHE flag
- Loop #2 (window shrinking) always enabled, active in all configs
- Loop #3 (decay) always enabled, active in all configs
- Loop #4 (speculation) disabled by default (ENABLE_PIPELINING=0), ready for A_C_FULL testing
- **BUT:** Loop #3 (heat decay) lacks principled design—slope is arbitrary, function choice unvalidated
- **Gap:** No measurements showing decay actually helps; DECAY_RATE has NO DoE knob to tune it
---
## FULLY WIRED & UTILIZED FEEDBACK LOOPS ✅
### ~~1. Hot-Words Cache Promotion Feedback Loop~~ ✅ DONE
**What's Measured:**
- `execution_heat` counter on each dictionary entry (incremented per execution)
**Trigger Condition:**
- When `execution_heat > HOTWORDS_EXECUTION_HEAT_THRESHOLD` (default: 50)
**Automatic Action:**
- `hotwords_cache_promote()` moves word to 32-entry LRU cache
- Subsequent lookups hit cache (1-2 ns vs 20-30 ns in bucket)
**Code Locations:**
- Metric tracking: `src/dictionary_management.c:170`
- Threshold check: `src/physics_hotwords_cache.c:155`
- Promotion logic: `src/physics_hotwords_cache.c:420-450`
**Observed Impact:**
- 1.78× speedup on cache-hit paths (measured in DoE phase 1)
- **Status: FULLY OPERATIONAL** ✅
---
### ~~2. Rolling Window Adaptive Shrinking Feedback Loop~~ ✅ FIXED
**What's Measured:**
- Pattern diversity: count of unique word transitions in execution history
**Measurement Function:**
```c
// src/rolling_window_of_truth.c:546-577 (FIXED)
uint64_t rolling_window_measure_diversity(const RollingWindowOfTruth* window)
{
// Count unique adjacent transitions (word_a → word_b)
// NOW RESPECTS effective_window_size - only scans recent entries
}
```
**Trigger Condition:**
- When pattern diversity growth rate < `ADAPTIVE_GROWTH_THRESHOLD` (default: 1%)
- Checked every `ADAPTIVE_CHECK_FREQUENCY` executions (default: 256)
**Automatic Action:**
- Shrink `effective_window_size` to `ADAPTIVE_SHRINK_RATE` percent of current (default: 75%)
- Window shrinks: 4096 → 3072 → 2304 → 1728 → ... → `ADAPTIVE_MIN_WINDOW_SIZE` (256)
**Code Locations:**
- Check function: `src/rolling_window_of_truth.c:589-645`
- Shrink calculation: Line 626
- Knobs: `include/rolling_window_knobs.h`
**Integration:**
- Called from `rolling_window_record_execution()` on every word execution
- **Status: FULLY OPERATIONAL** ✅
**Fix Applied (2025-11-08):**
- Line 559: Now limits scan to `effective_window_size` when window is warm
- Lines 561-564: Circular indexing ensures only recent entries within effective window are scanned
- **Effect:** New data arrives every tick, old data falls off after `effective_window_size` ticks (was: always 4096)
---
### 3. Linear Decay of Execution Heat Feedback Loop
**What's Measured:**
- `execution_heat` value per word (time-sensitive metric)
**Decay Formula (INTENTIONALLY BASELINE):**
- `H(t) = max(0, H_0 - decay_rate × Δt)` — **LINEAR decay**
- Configurable decay rate: `DECAY_RATE_PER_US_Q16` (default: 1 = 1/65536 heat/µs)
- Derived half-life: ~6-7 seconds for 100-heat word
**Automatic Action:**
- Stale words automatically lose heat weight
- Only frequently-executed words stay above promotion threshold
**Code Locations:**
- Decay logic: `src/physics_metadata.c:164-203` (`physics_metadata_apply_linear_decay()`)
- Integration: `src/vm.c:524` (called before each execution in inner loop)
**Design Rationale (From Phase 2 Specification):**
- Chose LINEAR over exponential for:
1. Performance (integer arithmetic, no floating-point)
2. Predictability (bounded convergence time)
3. Simplicity (fewer lines, easier to verify formally)
4. StarshipOS applicability (models "task context switch forces heat reset")
- **Explicitly noted as baseline:** "Can always switch to halflife after validating linear approach"
**STATUS: WIRED & UTILIZED, BUT NOT VALIDATED** ⚠️
**CRITICAL GAPS:**
1. **No empirical validation**: Does linear decay actually help? No measurements.
2. **Arbitrary initial slope**: Why 1 µs⁻¹? No evidence this is optimal.
3. **Wrong function?**: Could exponential, log, parabolic, or sinusoidal be better?
4. **Tuning blind**: DECAY_RATE_PER_US_Q16 has NO KNOB in DoE configurations
5. **No feedback loop closure**: Decay happens, but no measurement of whether it improved anything
---
## ✅ IMPLEMENTATION FIXES APPLIED
### Rolling Window Shrinking - FIXED
**The Bug (Now Fixed):**
The adaptive shrinking mechanism was shrinking the reported metric without enforcing the limit.
**Old Code (Line 557):**
```c
for (uint32_t i = 0; i < ROLLING_WINDOW_SIZE; i++) // ← ALWAYS full size (4096)
{
uint32_t current = window->execution_history[i];
// Count unique transitions...
}
```
**Fixed Code (Lines 559-574):**
```c
uint32_t scan_limit = (window->is_warm) ? window->effective_window_size : ROLLING_WINDOW_SIZE;
for (uint32_t i = 0; i < scan_limit; i++)
{
/* Index into the recent window (window_pos is the next write position) */
uint32_t idx = (window->window_pos + ROLLING_WINDOW_SIZE - scan_limit + i) % ROLLING_WINDOW_SIZE;
uint32_t current = window->execution_history[idx];
// Count unique transitions in recent entries only
}
```
**What Changed:**
- Now limits scan to `effective_window_size` when window is warm
- Uses circular indexing to ensure only RECENT entries are scanned
- New data arrives every tick, old data falls off after `effective_window_size` ticks
**User's Requirement Met:**
> "The window should advance with every tick so new data comes in and stale data falls off."
✅ Now data falls off after `effective_window_size` ticks, not always 4096
---
## ~~4. Pipelining Transition Metrics~~ ✅ NOW FUNCTIONAL
**What's Measured:**
- Transition patterns: (word_A → word_B) frequencies via `transition_metrics_record()`
- Probability calculations: P(B | A) in Q48.16 fixed-point format
- Prefetch accuracy: Hits vs. misses from speculative cache promotions
**Trigger Conditions:**
- `SPECULATION_THRESHOLD_Q48` (50% minimum confidence)
- `MIN_SAMPLES_FOR_SPECULATION` (10 minimum observations)
**Automatic Action (NOW WIRED):**
- After recording each transition, update probability cache
- Check: Does next word probability exceed threshold?
- If yes: Find dictionary entry and pre-promote to hotwords cache
- Effect: When next word is actually looked up, it's cache-warm
- Record: Track prefetch_attempts for feedback measurement
**Code Locations:**
- Metric collection: `src/physics_pipelining_metrics.c:92-109` (`transition_metrics_record()`)
- Probability calculation: `src/physics_pipelining_metrics.c:112-123`
- Decision logic: `src/physics_pipelining_metrics.c:164-186` (`transition_metrics_should_speculate()`)
- **Action wiring: `src/vm.c:544-583`** (execute_colon_word, speculative prefetch logic)
**Status: FULLY WIRED & FUNCTIONAL** ✅
**Wiring Details (2025-11-08):**
- Line 548: `transition_metrics_record()` - record the transition
- Line 551: `transition_metrics_update_cache()` - find most likely next word
- Line 554-555: `transition_metrics_should_speculate()` - check confidence threshold
- Lines 560-581: If confident, look up predicted word and promote to hotwords cache
- Line 580: Record the prefetch attempt for metrics feedback
**Loop Closure:**
1. ✅ Metrics collected: transition_heat array counts (word_B | word_A)
2. ✅ Probability calculated: Q48.16 format
3. ✅ Decision made: Confidence check against threshold
4. ✅ Action taken: Speculative promotion to hotwords cache
5. ✅ Feedback recorded: prefetch_attempts counter
### 5. Context-Aware Window Size Tuning (Phase 2 Placeholder)
**What's Measured:**
- Multi-word context patterns (sequences of 2-4 consecutive words)
**Intended Action (Not Connected):**
- Use binary search to find optimal `effective_window_size` for given workload
- Function: `transition_metrics_binary_chop_suggest_window()` (lines 340-357)
**Why Not Utilized:**
- Phase 2 research feature
- Experimental algorithm not yet validated
- No integration point in main execution path
**Status: PROTOTYPE, NOT INTEGRATED** ⏳
---
## OBSERVABILITY-ONLY (Not Decision-Driving)
### 6. Physics Metadata Temperature Tracking
**What's Measured:**
- `temperature_q8`: Exponential moving average of execution_heat
- Used for observability and formal verification model tracing
**Current Use:**
- Profiling and debugging (not driving optimization decisions)
- Forms foundation for Isabelle/HOL physics state machine proofs
**Status: WORKING AS INTENDED** (observability, not optimization)
---
## Bootstrap Seeding Mechanisms
These run once at startup to initialize feedback loop state:
### POST (Power-On Self-Test) Seeding
- **Function**: `rolling_window_seed_hotwords_cache()` (lines 258-329)
- **Action**: Records POST execution, promotes high-heat words to cache
- **Effect**: Cache is "warm" when user workload begins
### Pipelining Context Seeding
- **Function**: `rolling_window_seed_pipelining_context()` (lines 341-415)
- **Action**: Records POST sequences, establishes transition baseline
- **Effect**: Pipelining has initial pattern knowledge
---
## Configuration Knobs for Feedback Loops
All knobs are **tunable via Makefile**:
```bash
# Hot-words cache promotion
make HOTWORDS_EXECUTION_HEAT_THRESHOLD=75
# Adaptive window shrinking
make ADAPTIVE_SHRINK_RATE=75 # Keep 75%, shrink by 25%
make ADAPTIVE_MIN_WINDOW_SIZE=256 # Never shrink below 256
make ADAPTIVE_CHECK_FREQUENCY=256 # Check every 256 executions
make ADAPTIVE_GROWTH_THRESHOLD=1 # Shrink when growth < 1%
# Heat decay
make DECAY_RATE_PER_US_Q16=16384 # Decay rate (units: heat/μs)
make DECAY_MIN_INTERVAL=1000000 # Min interval between decays (ns)
# Pipelining (Phase 2, disabled by default)
make ENABLE_PIPELINING=1
make SPECULATION_THRESHOLD_Q48=$((50 << 16)) # 50% confidence
make MIN_SAMPLES_FOR_SPECULATION=10
```
---
## Summary Table: All Feedback Loops
| Loop | Measured | Threshold | Action | Wired? | Utilized? | Status |
|------|----------|-----------|--------|--------|-----------|--------|
| ~~1. Hot-words promotion~~ | `execution_heat` | > 50 | Add to cache | ✅ | ✅ | ✅ VALIDATED |
| ~~2. Adaptive shrinking~~ | `pattern_diversity` | growth < 1% | Shrink window | ✅ | ✅ | ✅ FIXED |
| 3. Heat decay | Time since last execution | ~6-7s half-life | Reduce heat | ✅ | ✅ | ⚠️ NOT VALIDATED |
| 4. Pipelining transitions | `transition_heat[]` | > 50% confidence | Speculative prefetch | ✅ | ❌ | ⏳ Phase 2 |
| 5. Context window tuning | Multi-word sequences | N/A | Binary chop | ✅ | ❌ | ⏳ Phase 2 |
| 6. Temperature tracking | `temperature_q8` | N/A | Observability only | ✅ | ✅ | ✅ OBSERVABILITY |
---
## Completed Actions ✅
### ✅ DONE: Fix Rolling Window Shrinking Implementation
**Status: APPLIED (2025-11-08)**
Modified `src/rolling_window_of_truth.c:546-577` (function `rolling_window_measure_diversity()`):
- Now respects `effective_window_size` when window is warm
- Uses circular indexing to scan only recent entries
- New data arrives every tick, old data falls off after `effective_window_size` ticks (not always 4096)
- A_B_C_FULL configuration now tests both hot-words cache + adaptive window shrinking together
---
## Next Steps for Phase 2 (If Proceeding)
### OPTIONAL 1: Connect Pipelining Decision Logic
**When `ENABLE_PIPELINING=1`:**
- Call `transition_metrics_should_speculate()` from vm inner loop
- If returns true: Pre-load next word from context prediction
- Measure cache miss reduction vs. A_BASELINE (Phase 2 experimental validation)
**Status:** Metrics collection ready, decision logic not wired
### OPTIONAL 2: Validate Context Window Tuning (Phase 2)
**For adaptive window size optimization:**
- Implement binary chop search algorithm in main execution path
- Tune `effective_window_size` dynamically based on context complexity
- Measure pattern capture efficiency vs. memory overhead
**Status:** Binary chop function exists, integration point not defined
---
## Files & Line Numbers Reference
| Purpose | File | Lines |
|---------|------|-------|
| Rolling window structure | `include/vm.h` | 58-70 |
| Rolling window implementation | `src/rolling_window_of_truth.c` | - |
| Adaptive shrinking | `src/rolling_window_of_truth.c` | 583-639 |
| **Bug location** | `src/rolling_window_of_truth.c` | **557** |
| Diversity measurement | `src/rolling_window_of_truth.c` | 546-571 |
| Shrinking knobs | `include/rolling_window_knobs.h` | - |
| Hot-words cache | `src/physics_hotwords_cache.c` | - |
| Promotion threshold | `src/physics_hotwords_cache.c` | 155 |
| Heat decay | `src/physics_metadata.c` | 164-203 |
| Pipelining metrics | `src/physics_pipelining_metrics.c` | - |
| Pipelining decision | `src/physics_pipelining_metrics.c` | 164-186 |
| Main integration | `src/vm.c` | 524, 542-548 |
---
**Analysis Complete:** 2025-11-08
**User Request:** Investigate feedback loops, identify implementation problems
**Finding:** 3 active + 2 dormant + 1 observability feedback loops; 1 critical bug in rolling window shrinking
docs/working/archive/superseded/heartbeat-doe-corrected-design.md
+344
View File
@@ -0,0 +1,344 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09.
> Instead: its own header. Moved here from `docs/working/experiments/02-experiments/heartbeat-doe/corrected-design.md`.
<!-- Moved from docs/02-experiments/heartbeat-doe/corrected-design.md to docs/working/archive/superseded/heartbeat-doe-corrected-design.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** Superseded by `phase2-corrected.md` (the
> "corrected and finalized" revision). Kept for historical record.
# StarForth Heartbeat DoE – CORRECTED Design (Heartrate Focus)
**Date:** November 20, 2025
**CORRECTION:** Previous design missing critical metric: **HEARTBEAT RATE (tick_ns)**
**Status:** Ready for implementation
---
## The Critical Insight You Caught
The heartbeat isn't just a **fixed tick**. The system has a **variable heartbeat rate** controlled by:
```c
/* HeartbeatWorker.tick_ns — this is DYNAMIC, not constant! */
uint64_t tick_ns; /* Nanoseconds between ticks — VARIES with load! */
```
**What this means:**
- Default: `HEARTBEAT_TICK_NS` (~1 millisecond)
- **But:** The inference engine can adaptively adjust `tick_target_ns`
- Different configurations may modulate heartbeat frequency differently
- **This is the KEY physics behavior we must measure!**
---
## Stage 1 Problem (What We Got Wrong)
Stage 1 baseline ran **without heartbeat thread** (HEARTBEAT_THREAD_ENABLED=0):
- Fixed 1ms tick (no modulation)
- No adaptive heartrate
- Doesn't reflect StarForth's intended physics-driven behavior
**Stage 1 conclusion:** "System is deterministic, no heartbeat variation"
**Stage 1 was incomplete** because it disabled the very mechanism that makes the system adaptive!
---
## Stage 2 Correction (What We Should Do)
### Primary Hypothesis
The 5 elite configurations differ in **how they modulate heartbeat rate**:
| Config | Expected Behavior |
|--------|-------------------|
| 1_0_1_1_1_0 | Tight, responsive heartrate (low jitter, fast modulation) |
| 1_0_1_1_1_1 | Similar, with decay inference slowing response? |
| 1_1_0_1_1_1 | Window-driven, may have different modulation pattern |
| 1_0_1_0_1_0 | Lean, may have sluggish response |
| 0_1_1_0_1_1 | Window-without-heat, baseline behavior |
### Key Metrics for Heartrate (Not Just Intervals)
**We need to measure:**
1. **Heartbeat Rate Modulation (NOT just interval jitter)**
```
heartbeat_rate_trajectory = [tick_ns_t0, tick_ns_t1, tick_ns_t2, ...]
What we measure:
- Mean heartbeat rate (expected: ~1,000,000 ns = 1 kHz baseline)
- Responsiveness to load (does rate slow under heavy load?)
- Convergence time (how long until rate stabilizes?)
- Modulation range (min to max tick_ns observed)
```
2. **Heartrate vs. Workload Coupling**
```
For each tick:
workload_intensity = lookups_this_tick / lookups_per_second_baseline
heartrate_hz = 1_000_000_000 / tick_ns
Measure: correlation(workload_intensity, heartrate_hz)
- Positive correlation: heartrate slows under load (good physics)
- Negative correlation: heartrate speeds up under load (bad physics)
- No correlation: heartrate doesn't respond to load (broken physics)
```
3. **Heartrate Stability (Not "jitter" but "rate consistency")**
```
- Does heartrate converge to a steady state?
- How quickly does it adapt to workload changes?
- Does it oscillate or settle smoothly?
```
4. **Heartrate Efficiency**
```
- Is the system running faster/slower with adapted heartrate?
- Workload duration: should improve with optimized heartrate
- Memory: should be more efficient with tuned heartrate
```
---
## Corrected Metrics Collection
### Per-Tick Collection (During Execution)
```c
/* In heartbeat_thread_main(), before nanosleep() */
struct HeartbeatTickData {
uint64_t tick_number;
uint64_t tick_ns; /* THIS IS THE HEARTBEAT RATE! */
uint64_t workload_ops_this_tick; /* Dictionary lookups in this tick */
uint64_t wall_clock_ns; /* Actual wall-clock time */
uint64_t hot_word_count; /* Physics state */
uint64_t total_heat; /* Physics state */
};
/* Store in circular buffer or rolling window */
```
### Post-Run Analysis Metrics
```c
typedef struct {
/* === Original Metrics === */
/* ... existing 38 metrics ... */
/* === CORRECTED: Heartbeat Rate Metrics === */
/* Heartbeat rate trajectory */
uint64_t total_ticks_executed;
double mean_heartbeat_rate_hz; /* Hz = 1e9 / mean_tick_ns */
double heartbeat_rate_stddev_hz; /* Variation in Hz */
double heartbeat_rate_cv; /* Coefficient of variation */
uint64_t heartbeat_rate_min_ns; /* Slowest (highest tick_ns) */
uint64_t heartbeat_rate_max_ns; /* Fastest (lowest tick_ns) */
double heartbeat_rate_range_hz; /* max_hz - min_hz */
/* Heartrate vs Load Coupling */
double load_to_heartrate_correlation; /* Does rate respond to load? */
int load_response_direction; /* +1 = slower under load (good) */
double heartrate_adaptation_latency_ticks; /* Ticks until rate responds */
double heartrate_settling_time_ticks; /* Ticks to settle after change */
/* Heartrate Convergence */
double heartrate_convergence_time_ms; /* Time to reach stable rate */
int heartrate_converged; /* 1 = yes, 0 = still oscillating */
double heartrate_oscillation_amplitude_hz; /* Peak-to-peak variation */
/* Overall Heartrate Stability Score */
double heartrate_stability_score; /* 0-100: how stable is heartrate? */
} DoeMetrics_CORRECTED;
```
---
## The Missing Piece: Heartbeat Rate Logging
**Critical code addition needed:**
```c
/* In vm.c heartbeat_thread_main() - CAPTURE tick_ns BEFORE and AFTER */
static inline void capture_heartbeat_rate_sample(VM *vm, uint64_t tick_ns, uint64_t workload_ops) {
/* Store sample in rolling buffer */
if (!vm->heartbeat_rate_samples) {
vm->heartbeat_rate_samples = calloc(HEARTBEAT_RATE_SAMPLE_MAX, sizeof(HeartbeatRateSample));
}
HeartbeatRateSample *sample = &vm->heartbeat_rate_samples[vm->heartbeat_rate_sample_count];
sample->tick_number = vm->heartbeat.tick_count;
sample->tick_ns = tick_ns; /* THE HEARTBEAT RATE! */
sample->workload_ops = workload_ops;
sample->timestamp_ns = platform_monotonic_ns();
vm->heartbeat_rate_sample_count++;
}
/* In heartbeat_thread_main() */
static void* heartbeat_thread_main(void *arg) {
VM *vm = (VM*)arg;
while (!vm->heartbeat.worker->stop_requested) {
uint64_t tick_start_ns = platform_monotonic_ns();
vm_heartbeat_run_cycle(vm); /* May adjust tick_ns */
uint64_t tick_ns = vm->heartbeat.worker->tick_ns;
uint64_t workload_ops = vm->execution_heat_current_tick; /* Need to track */
/* CAPTURE THE HEARTBEAT RATE! */
capture_heartbeat_rate_sample(vm, tick_ns, workload_ops);
/* Sleep */
nanosleep(...);
}
}
```
---
## Corrected Analysis (R Script)
```r
# Load heartbeat rate samples from CSV
doe_data <- read.csv("experiment_results_heartbeat.csv") %>%
mutate(
# Convert heartbeat rates (tick_ns → Hz)
mean_heartbeat_hz = 1e9 / mean_heartbeat_rate_ns,
heartrate_stability = 1 / (1 + heartbeat_rate_cv), # Lower CV = higher score
)
# PRIMARY ANALYSIS: Heartrate Stability
stability_by_heartrate <- doe_data %>%
group_by(configuration) %>%
summarize(
mean_heartrate_hz = mean(mean_heartbeat_hz),
mean_heartrate_stability = mean(heartrate_stability),
mean_load_coupling = mean(load_to_heartrate_correlation),
mean_workload_ms = mean(vm_workload_duration_ns_q48 / 1e6)
) %>%
mutate(
# Golden config = best heartrate stability + strongest load coupling
golden_score = (heartrate_stability * 0.5) + (load_coupling * 0.3) + (workload_improvement * 0.2)
)
# Golden Config = Highest golden_score
```
---
## Why This Matters
### Stage 1 Got Lucky
Stage 1 found 5 good configs **by accident** because it measured final performance, not physics.
### Stage 2 Should Be Different
Stage 2 should measure **whether the physics is working correctly**:
- **Can the system modulate heartbeat rate** in response to load?
- **Is the modulation smooth** or does it oscillate?
- **Does heartrate modulation improve overall performance?**
- **Which configuration has the most responsive, stable heartrate?**
**The golden configuration should be the one where heartbeat rate modulation is:**
1. ✓ **Responsive** (correlates with workload)
2. ✓ **Smooth** (doesn't oscillate wildly)
3. ✓ **Efficient** (results in best overall performance)
---
## Revised DoE Experiment Design
### What Actually Gets Measured
Instead of just "jitter in fixed 1ms tick intervals", measure:
**Per-Run Heartbeat Rate Trajectory:**
```
Tick 0: tick_ns = 1,000,000 (1 ms baseline)
Tick 1: tick_ns = 1,500,000 (slowing under load)
Tick 2: tick_ns = 2,000,000 (further slowdown)
Tick 10: tick_ns = 1,200,000 (adapting back)
Tick 100: tick_ns = 1,050,000 (converged to steady state)
```
**What we analyze:**
1. Did heartrate change at all? (some configs might keep it fixed)
2. How much did it vary? (CV of tick_ns, not CV of fixed intervals)
3. Did it respond to load? (correlation with workload)
4. Did it converge smoothly? (no oscillation)
5. Did it improve performance? (faster execution with adaptive heartrate?)
---
## Implementation Checklist (Revised)
- [ ] Add `HeartbeatRateSample` struct to capture `tick_ns` per tick
- [ ] Modify `heartbeat_thread_main()` to log `tick_ns` trajectory
- [ ] Extend `DoeMetrics` with heartrate-specific fields (not just interval jitter)
- [ ] Update CSV schema to include heartrate trajectory metrics
- [ ] Modify R analysis to focus on **heartrate modulation patterns**
- [ ] Update golden config selection: prioritize **load responsiveness**
---
## Expected Results (Corrected)
### Hypothesis: Configs Differ in Heartrate Modulation
**Config 1 (1_0_1_1_1_0) - Expected WINNER:**
- Heartrate slowly increases under load (tick_ns increases, Hz decreases)
- Smooth adaptation (no oscillation)
- Converges within 1000 ticks
- Performance improves (faster overall execution)
- Load coupling: 0.8+ (strong)
**Config 5 (0_1_1_0_1_1) - Expected LOSER:**
- Heartrate doesn't respond to load (fixed tick_ns)
- No correlation with workload
- Performance degrades (why? system can't adapt)
---
## Key Realization
The heartbeat isn't just a **timer**. It's a **feedback mechanism**:
```
Load increases
↓
Inference engine detects change
↓
Adjusts tick_target_ns (slower heartbeat = more thinking time)
↓
Heartbeat rate modulates (tick_ns increases)
↓
Physics engine gets more time to tune parameters
↓
System adapts and performance improves
```
**We need to measure whether this feedback loop is working!**
---
## Next Steps
1. **Extend HeartbeatWorker:** Add rate sample collection
2. **Modify heartbeat_thread_main():** Log `tick_ns` per tick
3. **Update DoeMetrics:** Heartrate trajectory analysis
4. **Rewrite CSV schema:** Focus on heartrate modulation
5. **Rewrite R analysis:** Measure load-response coupling and heartrate stability
6. **Re-run experiment:** Focus on configs that modulate heartrate best
---
**This changes the entire experiment focus.**
Instead of measuring "jitter in a fixed heartbeat", we're measuring "**how well does the system modulate its heartbeat in response to load?**"
That's the real physics story.
docs/working/archive/superseded/heartbeat-doe-design.md
+434
View File
@@ -0,0 +1,434 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09.
> Instead: its own header. Moved here from `docs/working/experiments/02-experiments/heartbeat-doe/design.md`.
<!-- Moved from docs/02-experiments/heartbeat-doe/design.md to docs/working/archive/superseded/heartbeat-doe-design.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** Pre-correction draft; superseded first by
> `corrected-design.md`, then by `phase2-corrected.md` (the finalized design). Kept for
> historical record.
# StarForth Factorial DoE with Heartbeat Integration
**Date:** November 20, 2025
**Status:** Design Phase
**Experiment:** 2^6 Factorial with Heartbeat-Aware Metrics Collection
---
## Overview
This document describes the **second-generation factorial DoE** that incorporates heartbeat observability into metric collection. Building on the baseline DOE (Stage 1: 3200 runs, 6 factors, no heartbeat), this experiment captures **per-tick heartbeat dynamics** to identify configurations with optimal stability characteristics.
### Key Innovation
Instead of measuring static performance, this experiment measures **temporal stability**:
- How quickly does the system converge under load changes?
- How much does heartbeat interval modulate in response to workload?
- What is the jitter envelope after settling?
- Which configuration achieves smoothest load→heartbeat coupling?
---
## Design Points (Top 5 Configurations from Stage 1)
Based on Stage 1 analysis, the elite configurations are:
1. **Config 1:** `1_0_1_1_1_0` (F1=ON, F2=OFF, F3=ON, F4=ON, F5=ON, F6=OFF)
2. **Config 2:** `1_0_1_1_1_1` (F1=ON, F2=OFF, F3=ON, F4=ON, F5=ON, F6=ON)
3. **Config 3:** `1_1_0_1_1_1` (F1=ON, F2=ON, F3=OFF, F4=ON, F5=ON, F6=ON)
4. **Config 4:** `1_0_1_0_1_0` (F1=ON, F2=OFF, F3=ON, F4=OFF, F5=ON, F6=OFF)
5. **Config 5:** `0_1_1_0_1_1` (F1=OFF, F2=ON, F3=ON, F4=OFF, F5=ON, F6=ON)
**Factors Mapping:**
- F1 = ENABLE_LOOP_1_HEAT_TRACKING
- F2 = ENABLE_LOOP_2_ROLLING_WINDOW
- F3 = ENABLE_LOOP_3_LINEAR_DECAY
- F4 = ENABLE_LOOP_4_PIPELINING_METRICS
- F5 = ENABLE_LOOP_5_WINDOW_INFERENCE
- F6 = ENABLE_LOOP_6_DECAY_INFERENCE
---
## New Heartbeat Metrics (Per-Tick Collection)
### 1. Heartbeat Interval Dynamics
```c
struct HeartbeatTickMetrics {
uint64_t tick_count; /* Absolute tick number */
uint64_t tick_interval_ns; /* Wall-clock duration of this tick */
uint64_t instantaneous_workload_ops; /* Dictionary lookups in this tick */
double instantaneous_load_ratio; /* (ops this tick) / (avg ops/tick) */
};
```
**Purpose:** Capture how heartbeat frequency modulates in response to workload intensity.
### 2. Cache Performance (Per-Tick Window)
```c
struct CacheWindowMetrics {
uint64_t tick_start;
uint64_t cache_hits_window; /* Hits in last 100 ticks */
uint64_t cache_attempts_window; /* Attempts in last 100 ticks */
double cache_hit_percent_window; /* Rolling window %, 0-100 */
};
```
**Purpose:** Measure cache stability, not just final hit rate.
### 3. Rolling Window Diagnostics
```c
struct WindowDiagnostics {
uint32_t effective_window_size; /* Current adaptive size */
uint64_t pattern_diversity; /* Unique word IDs observed */
double diversity_growth_rate; /* % new patterns per 1000 executions */
int is_warm; /* 1 if window is representative */
};
```
**Purpose:** Track window convergence and saturation.
### 4. Jitter Envelope
```c
struct JitterMetrics {
double tick_interval_mean_ns; /* Average heartbeat interval */
double tick_interval_stddev_ns; /* StdDev of interval */
double tick_interval_cv; /* Coefficient of variation (σ/μ) */
uint64_t outlier_count; /* Ticks >3σ from mean */
double outlier_ratio; /* Outliers / total ticks */
};
```
**Purpose:** Measure temporal smoothness and jitter control.
### 5. Decay Slope Tracking
```c
struct DecayTrajectory {
uint64_t tick_sample_count; /* How many ticks sampled */
double decay_slope_observed; /* Fitted slope from observations */
double slope_convergence_rate; /* Speed of slope convergence */
int slope_direction; /* -1=decreasing, 0=stable, +1=increasing */
};
```
**Purpose:** Monitor whether decay slope inference is tracking correctly.
### 6. Load-Response Coupling
```c
struct LoadResponseCoupling {
double correlation_load_to_interval; /* Correlation(workload, heartbeat_interval) */
double response_latency_ticks; /* Ticks until interval reacts to load change */
double overshoot_ratio; /* Peak response / steady-state response */
double settling_time_ticks; /* Ticks to settle within 10% of target */
};
```
**Purpose:** Measure how quickly and smoothly heartbeat responds to workload shifts.
---
## Extended DoeMetrics Struct
Add to `include/doe_metrics.h`:
```c
typedef struct {
/* === Original Metrics (preserved) === */
/* ... existing 35+ metrics ... */
/* === NEW: Heartbeat Stability Metrics === */
/* Tick dynamics */
uint64_t total_heartbeat_ticks; /* Total ticks during run */
double tick_interval_mean_ns; /* Average heartbeat interval */
double tick_interval_stddev_ns; /* StdDev of heartbeat interval */
double tick_interval_cv; /* Coefficient of variation */
uint64_t tick_outlier_count_3sigma; /* Ticks >3σ from mean */
double tick_outlier_ratio; /* Outliers / total ticks */
/* Cache stability window */
double cache_hit_percent_rolling_mean; /* Rolling avg, last 100 ticks */
double cache_hit_percent_rolling_stddev; /* Rolling stddev */
double cache_stability_index; /* How stable is cache hit %? */
/* Window convergence */
uint32_t window_final_effective_size; /* Final adaptive window size */
double pattern_diversity_final; /* Final unique patterns observed */
double diversity_growth_final; /* Growth rate at end */
int window_warmth_achieved; /* 1 = warm, 0 = still cold */
/* Decay slope convergence */
double decay_slope_fitted_final; /* Final fitted decay slope */
double decay_slope_convergence_rate; /* How fast did slope converge? */
int decay_slope_stable; /* 1 = stable direction, 0 = oscillating */
/* Load-response */
double load_interval_correlation; /* Correlation(load, heartbeat_interval) */
double response_latency_ticks; /* Ticks until response to workload change */
double overshoot_ratio; /* Peak response / steady-state */
double settling_time_ticks; /* Time to 10% settling band */
/* Aggregate stability score */
double overall_stability_score; /* Composite metric: (low jitter + fast convergence + tight correlation) */
} DoeMetrics; /* Now extends to ~65+ columns */
```
---
## Heartbeat Data Collection Architecture
### Per-Tick Logging (Lightweight)
During VM execution, capture minimal per-tick data:
```c
/* In vm.c vm_tick_heartbeat_worker() */
void vm_tick_heartbeat_worker(VM *vm) {
HeartbeatSnapshot snapshot = {
.published_tick = vm->heartbeat.tick_count,
.published_ns = platform_monotonic_ns(),
.window_width = vm->rolling_window.effective_window_size,
.decay_slope_q48 = vm->decay_slope_q48,
.hot_word_count = vm->hot_word_count_at_check,
.stale_word_count = vm->stale_word_count_at_check,
.total_heat = vm->total_heat_at_last_check,
};
/* Publish snapshot (double-buffered, lock-free read) */
int next_slot = 1 - vm->heartbeat.snapshot_index;
vm->heartbeat.snapshots[next_slot] = snapshot;
vm->heartbeat.snapshot_index = next_slot;
}
```
### Post-Run Analysis (Batch Processing)
After test execution completes, analyze collected tick data:
```c
/* In src/doe_metrics.c metrics_from_vm() */
void analyze_heartbeat_trajectory(VM *vm, DoeMetrics *metrics) {
HeartbeatSnapshot *snapshots = vm->heartbeat.snapshots;
uint64_t total_ticks = vm->heartbeat.tick_count;
/* 1. Compute tick interval statistics */
double intervals[total_ticks];
for (size_t i = 1; i < total_ticks; i++) {
intervals[i] = snapshots[i].published_ns - snapshots[i-1].published_ns;
}
metrics->tick_interval_mean_ns = compute_mean(intervals, total_ticks);
metrics->tick_interval_stddev_ns = compute_stddev(intervals, total_ticks);
metrics->tick_interval_cv = metrics->tick_interval_stddev_ns / metrics->tick_interval_mean_ns;
/* 2. Compute outlier ratio (ticks >3σ from mean) */
double threshold_3sigma = metrics->tick_interval_mean_ns + 3.0 * metrics->tick_interval_stddev_ns;
uint64_t outliers = 0;
for (size_t i = 0; i < total_ticks; i++) {
if (intervals[i] > threshold_3sigma) outliers++;
}
metrics->tick_outlier_ratio = (double)outliers / total_ticks;
/* 3. Compute decay slope convergence */
double slopes[total_ticks];
for (size_t i = 100; i < total_ticks; i++) { /* Skip warm-up */
slopes[i] = snapshots[i].decay_slope_q48 / 65536.0;
}
metrics->decay_slope_fitted_final = slopes[total_ticks - 1];
metrics->decay_slope_convergence_rate = compute_convergence_rate(slopes, total_ticks);
/* 4. Compute overall stability score */
double stability = 0.0;
stability += (100.0 - metrics->tick_interval_cv); /* Lower CV = higher score */
stability += (100.0 - metrics->tick_outlier_ratio * 100.0); /* Fewer outliers = higher */
stability += metrics->decay_slope_convergence_rate; /* Fast convergence = higher */
metrics->overall_stability_score = stability / 3.0;
}
```
---
## CSV Output Schema (Extended)
**New columns (per heartbeat experiment run):**
```
timestamp,configuration,run_number,
[... original 38 columns ...],
total_heartbeat_ticks,
tick_interval_mean_ns,tick_interval_stddev_ns,tick_interval_cv,
tick_outlier_count_3sigma,tick_outlier_ratio,
cache_hit_percent_rolling_mean,cache_hit_percent_rolling_stddev,cache_stability_index,
window_final_effective_size,pattern_diversity_final,diversity_growth_final,window_warmth_achieved,
decay_slope_fitted_final,decay_slope_convergence_rate,decay_slope_stable,
load_interval_correlation,response_latency_ticks,overshoot_ratio,settling_time_ticks,
overall_stability_score
```
**Total:** ~60 columns per run
---
## R Analysis Template
```r
library(tidyverse)
library(ggplot2)
library(gridExtra)
# Load heartbeat DoE data
doe_data <- read.csv("experiment_results_with_heartbeat.csv") %>%
mutate(
configuration = as.factor(configuration),
timestamp = as.POSIXct(timestamp)
)
# ========== MAIN ANALYSIS ==========
# 1. Stability Ranking
stability_rank <- doe_data %>%
group_by(configuration) %>%
summarize(
mean_stability_score = mean(overall_stability_score),
mean_jitter_cv = mean(tick_interval_cv),
mean_convergence_rate = mean(decay_slope_convergence_rate),
mean_settling_time = mean(settling_time_ticks),
n_runs = n()
) %>%
mutate(
rank = rank(-mean_stability_score, ties.method = "min")
) %>%
arrange(rank)
print("=== STABILITY RANKINGS ===")
print(stability_rank)
# 2. Visualization: Stability Scores
ggplot(doe_data, aes(x = reorder(configuration, overall_stability_score, FUN=median),
y = overall_stability_score,
fill = configuration)) +
geom_boxplot(alpha = 0.7) +
geom_jitter(width = 0.2, alpha = 0.4) +
coord_flip() +
theme_minimal() +
labs(title = "Overall Stability Scores by Configuration",
x = "Configuration",
y = "Stability Score (0-100)")
ggsave("01_stability_scores.png", width = 10, height = 6)
# 3. Visualization: Jitter Control (CV)
ggplot(doe_data, aes(x = reorder(configuration, tick_interval_cv, FUN=median),
y = tick_interval_cv,
fill = configuration)) +
geom_boxplot(alpha = 0.7) +
coord_flip() +
theme_minimal() +
labs(title = "Heartbeat Jitter (Coefficient of Variation)",
x = "Configuration",
y = "CV (lower = less jitter)")
ggsave("02_jitter_control.png", width = 10, height = 6)
# 4. Visualization: Convergence Speed
ggplot(doe_data, aes(x = reorder(configuration, decay_slope_convergence_rate, FUN=median),
y = decay_slope_convergence_rate,
fill = configuration)) +
geom_boxplot(alpha = 0.7) +
coord_flip() +
theme_minimal() +
labs(title = "Decay Slope Convergence Speed",
x = "Configuration",
y = "Convergence Rate (ticks)")
ggsave("03_convergence_speed.png", width = 10, height = 6)
# 5. Correlation: Load Response Coupling
ggplot(doe_data, aes(x = load_interval_correlation,
y = overall_stability_score,
color = configuration)) +
geom_point(alpha = 0.6, size = 2) +
facet_wrap(~configuration) +
theme_minimal() +
labs(title = "Load-Response Coupling vs Stability",
x = "Correlation(load, interval)",
y = "Overall Stability Score")
ggsave("04_load_response_coupling.png", width = 12, height = 8)
# 6. Statistical Test: Is top config significantly better?
top_config <- stability_rank$configuration[1]
other_configs <- stability_rank$configuration[-1]
top_data <- doe_data %>% filter(configuration == top_config) %>% pull(overall_stability_score)
other_data <- doe_data %>% filter(configuration %in% other_configs) %>% pull(overall_stability_score)
t_test <- t.test(top_data, other_data)
print(sprintf("\nT-test: %s vs others\np-value: %.4f", top_config, t_test$p.value))
# 7. Golden Configuration Summary
cat("\n=== GOLDEN CONFIGURATION CANDIDATE ===\n")
golden <- stability_rank %>% slice(1)
print(golden)
cat("\nInterpretation:")
cat(sprintf("\n- Lowest jitter (CV): %.4f", golden$mean_jitter_cv))
cat(sprintf("\n- Fastest convergence: %.1f ticks", golden$mean_convergence_rate))
cat(sprintf("\n- Settling time: %.1f ticks", golden$mean_settling_time))
cat(sprintf("\n- Recommendation: Use '%s' as golden configuration\n", golden$configuration))
```
---
## Execution Plan
### Phase 1: Implementation (3-4 hours)
1. Extend `DoeMetrics` struct with heartbeat columns
2. Implement `analyze_heartbeat_trajectory()` in `doe_metrics.c`
3. Capture snapshot data during heartbeat ticks
4. Update CSV schema
### Phase 2: Experiment Execution (2-4 hours)
1. Create `run_factorial_doe_with_heartbeat.sh` script
2. Run 5 elite configurations × 50 runs each = 250 total runs
3. Collect heartbeat + performance data
### Phase 3: Analysis (2-3 hours)
1. Load CSV in R
2. Compute stability rankings
3. Generate visualizations
4. Identify golden configuration
---
## Expected Outcomes
**Primary Question:** Which of the 5 elite configurations exhibits:
- Lowest heartbeat jitter (CV)?
- Fastest convergence to steady state?
- Strongest load-response correlation?
- Cleanest temporal signature?
**Expected Winner:** Configuration with balanced F-factors, likely:
- F1=1 (heat tracking required)
- F3=1 or F5=1 (decay or inference, but not both)
- F6=0 or F6=1 (decay inference optional)
**Golden Configuration:** Becomes baseline physics for MamaForth & PapaForth downstream work.
---
## Success Criteria
✓ All 5 configurations build and run without segfault
✓ Heartbeat data collected for all 250 runs
✓ Stability rankings computed and visualized
✓ Golden configuration identified with p < 0.05 significance
✓ R analysis report generated with 5+ plots
---
**Version:** 1.0
**Status:** Ready for implementation
**Next Step:** Implement heartbeat metrics capture and extended DoE script
docs/working/archive/superseded/heartbeat-rate-logging-plan.md
+545
View File
@@ -0,0 +1,545 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09.
> Instead: its own header. Moved here from `docs/working/architecture/03-architecture/heartbeat-system/implementation-plan.md`.
<!-- Moved from docs/03-architecture/heartbeat-system/implementation-plan.md to docs/working/archive/superseded/heartbeat-rate-logging-plan.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** Superseded by `instrumentation-plan.md`. Proposes the
> earlier `HeartbeatRateSample` design; the live target is `HeartbeatTickSnapshot`
> (`include/vm.h`) with a `tick_buffer` and `heartbeat_export_csv()`. Kept for historical
> record.
# Implementation Plan: Heartbeat Rate Logging
**Date:** November 20, 2025
**Priority:** BLOCKING — Must implement before running Phase 2 DoE
**Complexity:** Medium (requires code additions, CSV schema update)
---
## Overview
We need to capture **tick_ns trajectory** (the actual heartbeat rate changes) during execution, then analyze how each configuration modulates heartrate in response to workload.
---
## Step 1: Add Heartbeat Rate Sample Struct
### File: `include/vm.h`
Add after `HeartbeatSnapshot` definition:
```c
/* === Heartbeat Rate Trajectory Collection === */
#define HEARTBEAT_RATE_SAMPLE_MAX 100000 /* Maximum samples per run */
typedef struct {
uint64_t tick_number; /* Absolute tick count */
uint64_t tick_ns; /* ← HEARTBEAT RATE (nanoseconds) */
uint64_t workload_ops_this_tick; /* Dictionary lookups this tick */
uint64_t timestamp_ns; /* Wall-clock timestamp */
uint64_t hot_word_count; /* Physics state: hot words */
uint64_t total_heat; /* Physics state: total execution heat */
} HeartbeatRateSample;
```
### File: `include/vm.h` (in VM struct)
Add to the HeartbeatState section:
```c
/** @name Heartbeat Rate Trajectory (Phase 2 DoE)
* @{
*/
HeartbeatRateSample* heartbeat_rate_samples; /* Circular buffer of rate samples */
uint64_t heartbeat_rate_sample_count; /* Number of samples collected */
/** @} */
```
---
## Step 2: Initialize Heartbeat Rate Collection
### File: `src/vm.c` in `vm_init()`
Add after heartbeat initialization:
```c
/* Initialize heartbeat rate sampling for Phase 2 DoE */
vm->heartbeat_rate_samples = calloc(HEARTBEAT_RATE_SAMPLE_MAX, sizeof(HeartbeatRateSample));
vm->heartbeat_rate_sample_count = 0;
```
### File: `src/vm.c` in `vm_cleanup()`
Add cleanup:
```c
if (vm->heartbeat_rate_samples) {
free(vm->heartbeat_rate_samples);
vm->heartbeat_rate_samples = NULL;
}
```
---
## Step 3: Capture Heartbeat Rate Per Tick
### File: `src/vm.c` - Add helper function
```c
/**
* @brief Capture current heartbeat rate sample for analysis
*
* Called by heartbeat_thread_main() to log the tick_ns (heartbeat rate)
* along with contemporaneous workload metrics.
*
* @param vm Pointer to VM
* @param tick_ns Current heartbeat interval (nanoseconds)
* @param workload_ops Dictionary lookups executed this tick
*/
static void capture_heartbeat_rate_sample(VM *vm, uint64_t tick_ns, uint64_t workload_ops)
{
if (!vm || !vm->heartbeat_rate_samples)
return;
if (vm->heartbeat_rate_sample_count >= HEARTBEAT_RATE_SAMPLE_MAX)
return; /* Buffer full, stop sampling */
HeartbeatRateSample *sample = &vm->heartbeat_rate_samples[vm->heartbeat_rate_sample_count];
sample->tick_number = vm->heartbeat.tick_count;
sample->tick_ns = tick_ns;
sample->workload_ops_this_tick = workload_ops;
sample->timestamp_ns = platform_monotonic_ns();
/* Capture current physics state */
sample->hot_word_count = vm->hot_word_count_at_check;
sample->total_heat = vm->total_heat_at_last_check;
vm->heartbeat_rate_sample_count++;
}
```
---
## Step 4: Log Workload Per Tick
### File: `src/vm.c` - Track lookups per tick
Add to VM struct (in `include/vm.h`):
```c
/** @name Per-Tick Workload Metrics
* @{
*/
uint64_t workload_ops_current_tick; /* Dictionary lookups in current tick */
/** @} */
```
### File: `src/vm.c` - In `vm_heartbeat_run_cycle()`
**Before calling inference engine:**
```c
void vm_heartbeat_run_cycle(VM *vm)
{
if (!vm || !vm->heartbeat.heartbeat_enabled)
return;
vm->heartbeat.tick_count++;
/* Reset per-tick workload counter */
uint64_t lookups_this_tick = vm->workload_ops_current_tick;
vm->workload_ops_current_tick = 0; /* Reset for next tick */
/* ... existing inference logic ... */
}
```
### File: `src/vm.c` - Increment lookup counter
In the main word execution loop (in `vm_interpret()` or inner interpreter):
```c
/* After successful dictionary lookup */
vm->workload_ops_current_tick++;
```
---
## Step 5: Capture in Heartbeat Thread
### File: `src/vm.c` in `heartbeat_thread_main()`
Replace the existing sleep logic:
```c
static void* heartbeat_thread_main(void *arg)
{
VM *vm = (VM*)arg;
if (!vm || !vm->heartbeat.worker)
return NULL;
HeartbeatWorker *worker = vm->heartbeat.worker;
worker->running = 1;
while (!worker->stop_requested && vm->heartbeat.heartbeat_enabled)
{
vm_heartbeat_run_cycle(vm);
/* GET THE CURRENT HEARTBEAT RATE */
uint64_t tick_ns = worker->tick_ns ? worker->tick_ns : HEARTBEAT_TICK_NS;
uint64_t workload_ops = vm->workload_ops_current_tick;
/* ↓ ADD THIS ↓ CAPTURE HEARTBEAT RATE SAMPLE */
capture_heartbeat_rate_sample(vm, tick_ns, workload_ops);
/* ↑ ADD THIS ↑ */
struct timespec req;
req.tv_sec = (time_t)(tick_ns / 1000000000ULL);
req.tv_nsec = (long)(tick_ns % 1000000000ULL);
struct timespec rem;
while (!worker->stop_requested && vm->heartbeat.heartbeat_enabled &&
nanosleep(&req, &rem) == -1 && errno == EINTR)
{
req = rem;
}
}
worker->running = 0;
return NULL;
}
```
---
## Step 6: Extend DoeMetrics
### File: `include/doe_metrics.h`
Add to `DoeMetrics` struct:
```c
/* === HEARTBEAT RATE MODULATION METRICS (NEW) === */
/* Per-run heartbeat rate statistics */
uint64_t total_heartbeat_ticks;
double mean_heartbeat_rate_hz; /* Average frequency (Hz) */
double heartbeat_rate_stddev_hz; /* Variation in Hz */
double heartbeat_rate_cv; /* CV of heartbeat rate */
uint64_t heartbeat_rate_min_ns; /* Slowest tick observed */
uint64_t heartbeat_rate_max_ns; /* Fastest tick observed */
double heartbeat_rate_modulation_range_hz; /* max_hz - min_hz */
/* Load-Heartrate Response */
double load_to_heartrate_correlation; /* Correlation(workload, heartrate) */
int load_response_direction; /* +1 if slower under load, -1 if faster */
double heartrate_adaptation_latency_ticks; /* Ticks until rate responds to load */
double heartrate_settling_time_ticks; /* Ticks to settle after adaptation */
/* Heartrate Convergence */
double heartrate_convergence_time_ms; /* Time to reach stable rate */
int heartrate_converged; /* 1 = converged, 0 = oscillating */
double heartrate_oscillation_amplitude_hz; /* Peak-to-peak variation in final state */
/* Overall Heartrate Quality Score */
double heartrate_responsiveness_score; /* 0-100: how well does it respond? */
double heartrate_stability_score; /* 0-100: how smooth is adaptation? */
```
---
## Step 7: Implement Metrics Extraction
### File: `src/doe_metrics.c`
Add new function:
```c
/**
* @brief Analyze heartbeat rate trajectory and compute metrics
*
* @param vm Pointer to VM with heartbeat_rate_samples
* @param metrics Pointer to DoeMetrics to fill
*/
static void analyze_heartbeat_rate_trajectory(VM *vm, DoeMetrics *metrics)
{
if (!vm || !vm->heartbeat_rate_samples || vm->heartbeat_rate_sample_count == 0) {
metrics->mean_heartbeat_rate_hz = 0;
metrics->load_to_heartrate_correlation = 0;
return;
}
uint64_t count = vm->heartbeat_rate_sample_count;
HeartbeatRateSample *samples = vm->heartbeat_rate_samples;
/* 1. Compute heartbeat rate statistics (tick_ns → Hz) */
double *heartrate_hz = malloc(count * sizeof(double));
double *workload_intensity = malloc(count * sizeof(double));
double min_ns = samples[0].tick_ns;
double max_ns = samples[0].tick_ns;
double sum_hz = 0;
double sum_workload = 0;
for (uint64_t i = 0; i < count; i++) {
double tick_ns = samples[i].tick_ns;
heartrate_hz[i] = 1e9 / tick_ns;
workload_intensity[i] = samples[i].workload_ops_this_tick;
sum_hz += heartrate_hz[i];
sum_workload += workload_intensity[i];
if (tick_ns < min_ns) min_ns = tick_ns;
if (tick_ns > max_ns) max_ns = tick_ns;
}
/* Mean heartbeat rate */
metrics->mean_heartbeat_rate_hz = sum_hz / count;
metrics->heartbeat_rate_min_ns = (uint64_t)min_ns;
metrics->heartbeat_rate_max_ns = (uint64_t)max_ns;
metrics->heartbeat_rate_modulation_range_hz = (1e9 / min_ns) - (1e9 / max_ns);
/* 2. Compute standard deviation */
double variance = 0;
for (uint64_t i = 0; i < count; i++) {
double delta = heartrate_hz[i] - metrics->mean_heartbeat_rate_hz;
variance += delta * delta;
}
metrics->heartbeat_rate_stddev_hz = sqrt(variance / count);
metrics->heartbeat_rate_cv = metrics->heartbeat_rate_stddev_hz / metrics->mean_heartbeat_rate_hz;
/* 3. Compute correlation(workload, heartrate) — THE KEY METRIC */
double workload_mean = sum_workload / count;
double cov = 0;
double workload_var = 0;
for (uint64_t i = 0; i < count; i++) {
double workload_delta = workload_intensity[i] - workload_mean;
double heartrate_delta = heartrate_hz[i] - metrics->mean_heartbeat_rate_hz;
cov += workload_delta * heartrate_delta;
workload_var += workload_delta * workload_delta;
}
if (workload_var > 0 && metrics->heartbeat_rate_stddev_hz > 0) {
metrics->load_to_heartrate_correlation = cov / sqrt(workload_var *
(metrics->heartbeat_rate_stddev_hz * metrics->heartbeat_rate_stddev_hz * count));
}
/* 4. Determine if heartrate responds correctly to load */
metrics->load_response_direction = (metrics->load_to_heartrate_correlation > 0.3) ? 1 : -1;
/* 5. Estimate convergence time (when heartrate stabilizes) */
double threshold = 0.15; /* 15% of final value */
metrics->heartrate_converged = 1;
for (uint64_t i = count / 2; i < count; i++) {
double current_cv = 0;
/* Compute CV of last N samples */
for (uint64_t j = i; j < (i + 100 < count ? i + 100 : count); j++) {
double delta = heartrate_hz[j] - metrics->mean_heartbeat_rate_hz;
current_cv += delta * delta;
}
if (sqrt(current_cv) > threshold * metrics->mean_heartbeat_rate_hz) {
metrics->heartrate_converged = 0;
metrics->heartrate_convergence_time_ms = i; /* Approximate */
break;
}
}
/* 6. Compute overall responsiveness score */
/* Higher correlation + faster convergence + lower CV = higher score */
metrics->heartrate_responsiveness_score =
(metrics->load_to_heartrate_correlation + 1.0) * 50 *
(1.0 / (1.0 + metrics->heartbeat_rate_cv));
metrics->heartrate_stability_score =
(100.0 * (1.0 - metrics->heartbeat_rate_cv)) *
(metrics->heartrate_converged ? 1.0 : 0.5);
free(heartrate_hz);
free(workload_intensity);
}
```
### Update `metrics_from_vm()`
Call the new analysis function:
```c
DoeMetrics metrics_from_vm(VM *vm, ...) {
/* ... existing metrics extraction ... */
/* NEW: Analyze heartbeat rate trajectory */
analyze_heartbeat_rate_trajectory(vm, &metrics);
return metrics;
}
```
---
## Step 8: Update CSV Schema
### File: `src/doe_metrics.c` in `metrics_write_csv_header()`
Add new columns:
```c
void metrics_write_csv_header(FILE *out) {
{
printf '[...existing columns...]');
printf('total_heartbeat_ticks,mean_heartbeat_rate_hz,heartbeat_rate_stddev_hz,heartbeat_rate_cv,'
'heartbeat_rate_min_ns,heartbeat_rate_max_ns,heartbeat_rate_modulation_range_hz,'
'load_to_heartrate_correlation,load_response_direction,'
'heartrate_convergence_time_ms,heartrate_converged,'
'heartrate_responsiveness_score,heartrate_stability_score\n');
} > output_csv;
}
```
### File: `src/doe_metrics.c` in `metrics_write_csv_row()`
Add the values:
```c
void metrics_write_csv_row(FILE *out, const DoeMetrics *metrics) {
fprintf(out, "[...existing fields...],"
"%lu,%lf,%lf,%lf,%lu,%lu,%lf,%lf,%d,%lu,%d,%lf,%lf\n",
metrics->total_heartbeat_ticks,
metrics->mean_heartbeat_rate_hz,
metrics->heartbeat_rate_stddev_hz,
metrics->heartbeat_rate_cv,
metrics->heartbeat_rate_min_ns,
metrics->heartbeat_rate_max_ns,
metrics->heartbeat_rate_modulation_range_hz,
metrics->load_to_heartrate_correlation,
metrics->load_response_direction,
metrics->heartrate_convergence_time_ms,
metrics->heartrate_converged,
metrics->heartrate_responsiveness_score,
metrics->heartrate_stability_score);
}
```
---
## Step 9: Updated R Analysis
### File: `scripts/analyze_heartbeat_stability_REVISED.R`
```r
library(tidyverse)
# Load data
doe_data <- read.csv("experiment_results_heartbeat.csv") %>%
mutate(
configuration = as.factor(configuration),
# New metrics available now
load_response = ifelse(load_to_heartrate_correlation > 0.5, "Responsive", "Sluggish")
)
# PRIMARY METRIC: Load-Heartrate Correlation
# This tells us which config best responds to workload changes
ranking <- doe_data %>%
group_by(configuration) %>%
summarize(
mean_correlation = mean(load_to_heartrate_correlation, na.rm = TRUE),
mean_responsiveness = mean(heartrate_responsiveness_score, na.rm = TRUE),
mean_stability = mean(heartrate_stability_score, na.rm = TRUE),
mean_convergence_time = mean(heartrate_convergence_time_ms, na.rm = TRUE),
golden_score = (mean(load_to_heartrate_correlation) * 0.5) +
(mean(heartrate_responsiveness_score) * 0.3) +
(mean(heartrate_stability_score) * 0.2)
) %>%
arrange(-golden_score)
print("=== HEARTBEAT RATE RESPONSIVENESS RANKING ===")
print(ranking)
# Plot 1: Load-Heartrate Correlation (KEY METRIC)
ggplot(doe_data, aes(x = reorder(configuration, load_to_heartrate_correlation, FUN=median),
y = load_to_heartrate_correlation,
fill = configuration)) +
geom_boxplot() +
coord_flip() +
labs(title = "Load-Heartrate Correlation (Higher = Better)",
y = "Correlation(workload, heartrate)")
ggsave("01_load_heartrate_correlation.png")
# Plot 2: Heartrate Responsiveness Score
ggplot(doe_data, aes(x = reorder(configuration, heartrate_responsiveness_score, FUN=median),
y = heartrate_responsiveness_score,
fill = configuration)) +
geom_boxplot() +
coord_flip() +
labs(title = "Heartrate Responsiveness Score (0-100)")
ggsave("02_heartrate_responsiveness.png")
```
---
## Compilation Steps
1. **Update headers:** Add structs and fields to `include/vm.h`, `include/doe_metrics.h`
2. **Implement collection:** Add capture function in `src/vm.c`
3. **Implement analysis:** Add analysis function in `src/doe_metrics.c`
4. **Update CSV:** Modify CSV header/row writers
5. **Compile:** `make clean && make fastest`
6. **Test:** `./build/starforth -c "1 2 + . BYE"`
---
## Verification
After implementation, verify:
✓ Binary compiles without errors
✓ Heartbeat rate samples are captured (check sample_count > 0 after run)
✓ CSV contains new heartrate columns
✓ Correlation metrics are computed
✓ R analysis generates ranking by load-response
---
## Expected Results
When properly implemented, you'll see:
```
=== HEARTBEAT RATE RESPONSIVENESS RANKING ===
configuration mean_correlation mean_responsiveness golden_score
1_0_1_1_1_0 0.82 78.5 0.795
1_0_1_1_1_1 0.78 75.2 0.761
1_1_0_1_1_1 0.65 62.3 0.641
1_0_1_0_1_0 0.45 48.1 0.451
0_1_1_0_1_1 0.12 15.8 0.148
```
**Golden config:** 1_0_1_1_1_0 (strongest load-heartrate coupling)
---
## Timeline
- **Implementation:** 2-3 hours
- **Testing:** 1 hour
- **Re-run Phase 2:** 2-4 hours
- **Analysis:** 30 min
- **Total:** ~6-9 hours
---
## Files Modified
- `include/vm.h` — Add HeartbeatRateSample struct and VM fields
- `include/doe_metrics.h` — Add heartrate metrics to DoeMetrics
- `src/vm.c` — Add sample capture, logging, initialization
- `src/doe_metrics.c` — Add analysis function, CSV updates
- `scripts/analyze_heartbeat_stability_REVISED.R` — Rewritten for load-response focus
---
**This implementation captures the actual heartbeat rate modulation and measures how well each config responds to load.**
This is the correct physics story. 🚀
docs/working/archive/superseded/loop-5-sketch.md
+278
View File
@@ -0,0 +1,278 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged superseded 2026-10-09 (an early design sketch of Loop #5).
> Instead: its own header; Loop #5 as built is v3/src/inference_engine.c. Moved here from `docs/working/architecture/03-architecture/adaptive-systems/loop-5-sketch.md`.
<!-- Moved from docs/03-architecture/adaptive-systems/loop-5-sketch.md to docs/working/archive/superseded/loop-5-sketch.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ SUPERSEDED (flagged 2026-10-09).** Superseded by `window-inference-redesign.md`
> (Levene's-test window inference — the shipped Loop #5 mechanism). This undated sketch ended
> in an open question and was never approved as designed. Kept for historical record.
# Loop #5: Context-Aware Window Tuning - Design Sketch
## The Challenge
**Current state:** Window size shrinks based on pattern diversity, but we don't know if shrinking improves or hurts **pipelining prediction accuracy**.
**Goal:** Use actual prefetch hit rate to guide window size tuning via binary chop search.
---
## What Loop #5 Should Do
### High-Level Flow
1. **Measure**: Track global pipelining metrics across ALL words
- Total speculative prefetch attempts
- Successful prefetch hits
- Current accuracy = hits / attempts
2. **Trigger**: Periodically (e.g., every 1000 executions when warm)
- Calculate prediction accuracy at current effective_window_size
- Call binary chop to suggest next size to try
3. **Adapt**: Shrink or grow effective_window_size based on accuracy trend
- If smaller window → better accuracy: keep shrinking
- If smaller window → worse accuracy: grow back up
- Converge to sweet spot
4. **Measure**: Record the accuracy at each window size tried
- Build history of (window_size → accuracy) pairs
- Use this history to accelerate convergence
---
## Architecture Needed
### 1. VM-Level Metrics (New)
**File:** `include/vm.h` (add to VM struct)
```c
typedef struct {
uint64_t prefetch_attempts; /* Total speculative prefetch calls */
uint64_t prefetch_hits; /* Successful hits (word was looked up next) */
uint64_t window_tuning_checks; /* How many times we've checked window size */
uint32_t last_checked_window_size;
double last_checked_accuracy;
uint32_t suggested_next_size; /* What binary chop recommends trying */
} PipelineGlobalMetrics;
```
**Add to VM struct:**
```c
struct VM {
// ... existing fields ...
PipelineGlobalMetrics pipeline_metrics; /* NEW: aggregated metrics */
};
```
### 2. Update Prefetch Feedback (in vm.c)
**When speculative prefetch succeeds:**
```c
// In execute_colon_word(), after promotion:
if (spec_entry && vm->hotwords_cache && ENABLE_HOTWORDS_CACHE)
{
spec_entry->execution_heat = HOTWORDS_EXECUTION_HEAT_THRESHOLD + 1;
hotwords_cache_promote(vm->hotwords_cache, spec_entry);
// NEW: Record the prefetch attempt globally
vm->pipeline_metrics.prefetch_attempts++;
prev_word->transition_metrics->prefetch_attempts++; // Per-word too
}
```
**When word is actually looked up (check if it was speculated):**
```c
// In vm_interpret_word(), during dictionary lookup:
if (entry && ENABLE_PIPELINING && vm->hotwords_cache)
{
// NEW: Check if this word was in speculated cache (dirty bit or timestamp)
if (word_was_speculatively_promoted_recently(entry))
{
vm->pipeline_metrics.prefetch_hits++; // Global metric
// per_word_metrics->prefetch_hits++; // Already done in prefetch logic
}
}
```
### 3. Window Tuning Decision Loop
**New function:** `physics_pipelining_metrics.c`
```c
/**
* Suggest window size via binary chop based on prefetch accuracy trend.
* Called periodically (every 1000 executions when pipelining is enabled).
*/
uint32_t loop_5_binary_chop_suggest_window(
const PipelineGlobalMetrics *global,
uint32_t current_window_size,
uint32_t min_window,
uint32_t max_window)
{
if (global->prefetch_attempts == 0)
return current_window_size; /* Not enough data yet */
double current_accuracy = (double)global->prefetch_hits / (double)global->prefetch_attempts;
/* Binary chop: try smaller window first (aggressive shrinking)
* If accuracy drops, back off and try larger windows */
if (global->window_tuning_checks == 0)
{
/* First check: try shrinking by 25% */
return (current_window_size * 75) / 100;
}
/* Compare current accuracy to last check */
double accuracy_delta = current_accuracy - global->last_checked_accuracy;
if (accuracy_delta > 0.01) /* Improvement threshold: 1% */
{
/* Accuracy improved! Try shrinking more */
uint32_t smaller = (current_window_size * 75) / 100;
return (smaller > min_window) ? smaller : min_window;
}
else if (accuracy_delta < -0.01)
{
/* Accuracy degraded. Try growing instead */
uint32_t larger = (current_window_size * 133) / 100; /* Grow by ~33% */
return (larger < max_window) ? larger : max_window;
}
else
{
/* Plateau: stick with current size */
return current_window_size;
}
}
```
### 4. Integration Point
**In rolling_window_of_truth.c: `rolling_window_check_adaptive_shrink()`**
```c
void rolling_window_check_adaptive_shrink(RollingWindowOfTruth* window)
{
// ... existing adaptive shrinking code ...
/* Loop #5: Binary chop window tuning (Phase 2 IMPLEMENTATION) */
if (ENABLE_PIPELINING && window->is_warm)
{
window->adaptive_check_count++;
/* Check every 1000 executions for tuning opportunity */
if (window->adaptive_check_count % 1000 == 0)
{
PipelineGlobalMetrics *metrics = &vm->pipeline_metrics;
/* Calculate current accuracy */
double current_accuracy = (metrics->prefetch_attempts > 0)
? (double)metrics->prefetch_hits / (double)metrics->prefetch_attempts
: 0.0;
/* Ask binary chop for next window size to try */
uint32_t suggested_size = loop_5_binary_chop_suggest_window(
metrics,
window->effective_window_size,
ADAPTIVE_MIN_WINDOW_SIZE,
ROLLING_WINDOW_SIZE
);
/* Apply if different */
if (suggested_size != window->effective_window_size)
{
log_message(LOG_INFO,
"Loop #5: Binary chop tuning: %u → %u (accuracy %.2f%%, %lu/%lu hits)",
window->effective_window_size,
suggested_size,
current_accuracy * 100.0,
metrics->prefetch_hits,
metrics->prefetch_attempts);
window->effective_window_size = suggested_size;
}
/* Record this check for next iteration */
metrics->last_checked_window_size = window->effective_window_size;
metrics->last_checked_accuracy = current_accuracy;
metrics->window_tuning_checks++;
}
}
}
```
---
## Key Design Decisions
### A. Accuracy Measurement
- **Count**: Successful speculative promotions that were actually used
- **Challenge**: How to know if lookup was "predicted" vs "natural"?
- Option 1: Mark speculatively-promoted entries with a timestamp/flag
- Option 2: Track word IDs predicted and check if they match next lookup
- Option 3: Use per-word prefetch metrics already collected
### B. Tuning Frequency
- **Every 1000 executions** (when window is warm)
- Could be tunable: `make WINDOW_TUNING_FREQUENCY=1000`
### C. Search Strategy
- **Greedy shrinking first**: Try smaller windows aggressively
- **Backoff on degradation**: If accuracy drops, grow back up
- **Convergence**: Stop when oscillating around sweet spot
### D. Knobs Needed (Makefile)
```bash
make ENABLE_PIPELINING=1 # Must be enabled for Loop #5
make WINDOW_TUNING_FREQUENCY=1000 # How often to check
make WINDOW_TUNING_ACCURACY_THRESHOLD=1 # % improvement to continue shrinking
```
---
## Validation Approach
Once Loop #5 is wired, DoE would test:
- **A_BASELINE**: No optimizations
- **A_B_CACHE**: Cache only
- **A_C_FULL**: Window tuning with pipelining only
- **A_B_C_FULL**: Cache + window tuning + pipelining
**Metrics to collect:**
- Final effective_window_size (what did it converge to?)
- Final prefetch accuracy (%)
- Throughput improvement vs baseline
---
## Risks & Unknowns
1. **Overhead**: Checking accuracy every 1000 executions adds latency
2. **Noise**: If accuracy is noisy, binary chop might oscillate
3. **Workload-dependent**: Optimal window size might vary by workload
4. **False positives**: Speculative lookup might not correlate with prediction
---
## Sketch Summary
**Loop #5 would:**
1. Track global prefetch accuracy (hits/attempts)
2. Every 1000 executions, ask binary chop: "Is current window size optimal?"
3. Binary chop suggests shrinking, holding, or growing based on accuracy trend
4. Apply suggestion and record for next iteration
5. Over time, converge to window size that maximizes prediction accuracy
**Code locations if implemented:**
- VM struct: Add PipelineGlobalMetrics field
- vm.c: Update global metrics on prefetch attempts/hits
- physics_pipelining_metrics.c: `loop_5_binary_chop_suggest_window()`
- rolling_window_of_truth.c: Call binary chop from adaptive shrink check
**Effort**: ~150 LOC across 3-4 files
---
**Question for user:** Does this architecture align with what you had in mind for Loop #5?
docs/working/archive/superseded/steady-state-machine.md
+223
View File
@@ -0,0 +1,223 @@
> **SUPERSEDED (docs triage 2026-10-14).** Flagged obsolete 2026-10-09.
> Instead: its own header. Moved here from `docs/working/architecture/03-architecture/physics-engine/steady-state-machine.md`.
<!-- Moved from docs/03-architecture/physics-engine/steady-state-machine.md to docs/working/archive/superseded/steady-state-machine.md on 2026-06-16 (docs reorg Phase 2) -->
> **⚠ OBSOLETE (flagged 2026-10-09).** Conceptual precursor to the L8 Jacquard mode-selector
> system; it never mentions Jacquard/L8, the attractor buckets, or the six named modes, and
> was never reconciled with the implemented `ssm_jacquard.c` system. The canonical current
> reference for the physics/feedback model is `feedback-loops.md`. Kept for historical record.
# 1. Introduction
A Steady State Machine (SSM) is a novel computational model in which the system’s operational state is defined not by discrete symbolic transitions, but by convergence to an optimal runtime equilibrium.
Unlike traditional models such as Finite State Machines (FSMs), Pushdown Automata, or Turing Machines — where “state” is explicit and symbolic — a Steady State Machine derives its behavior from continuous thermodynamic variables such as:
* execution_heat
* entropy (diversity)
* pipeline readiness
* heat decay
* steady-state slope
* localized temperature gradients
* adaptive strategy selection
An SSM observes itself, adapts, and settles into stable performance basins called steady states.
This makes the Steady State Machine the first computational model where the attractor defines the machine, not the symbol.
# 2. Core Definition
Definition 1 — Steady State Machine
> A Steady State Machine is a computational system defined by:
A set of internal physical metrics
* 𝐻 = execution_heats
* 𝐸 = entropy/diversity measures
* 𝑃 = pipeline activation thresholds
* Θ = heat decay profile
* 𝑊 = sliding execution window
A set of adaptive policies
* lookup strategies
* caching regimes
* bucket reorganizations
* pipeline decisions
* inference-based mode switches
* A state-space composed of attractor basins, not symbolic states
* steady attractor
* quasi-steady attractor
* transient attractor
* chaotic attractor
* A governing dynamic
### 𝑀𝑡+1=𝑓(𝑀𝑡,Δ𝐻,Δ𝑊,Δ𝐸) where 𝑓 pushes the machine toward a stable basin.
The machine is defined by its convergence behavior, not its instruction sequence.
# 3. The Fundamental Insight
In an SSM, performance is the state.
The system is constantly measuring:
* how hot words are
* how often patterns recur
* how pipeline-able a sequence is
* how diverse the window is
* how much the dictionary should reorganize
And as these stabilize, the machine settles into a steady state — a self-maintaining optimal regime.
This regime is preserved until the environment changes (e.g., new workloads, diversity spikes), at which point the SSM moves into a transient or chaotic regime and eventually finds a new equilibrium.
# 4. Phases of a Steady State Machine
## 1. Transient Phase
- System warms up
- Execution patterns are sporadic
- Heat gradients are noisy
- Pipeline inactive or unstable
## 2. Quasi-Steady Phase
- Some patterns emerge
- Lookups begin to bias
- Hot words form local attractors
- Dictionary reorders stabilize
## 3. True Steady State
- Heat surface smooth
- Pipeline active
- Diversity window predictable
- Lookups converge to optimal
- Runtime > baseline VM
- System remains stable unless perturbed
## 4. Chaotic/Disruption Phase
### Triggered by:
- major workload shift
- dictionary mutation
- extreme entropy spike
- cold-cache shock
- SSM falls out of equilibrium temporarily.
## 5. Axioms of Steady State Machines
>Axiom 1 — Convergence
Every sustained workload induces the SSM to converge toward a stable attractor unless external entropy forces divergence.
>Axiom 2 — Locality of Heat
Execution heat reflects both locality and temporal relevance. Locality produces stability.
>Axiom 3 — Adaptive Reordering
Reordering is not symbolic mutation; it is thermodynamic self-optimization.
>Axiom 4 — Stability Maximizes Throughput
The steady state is always faster than the cold state, and usually faster than the naive baseline VM.
>Axiom 5 — Perturbation Response
When disrupted, the SSM seeks a new steady state; it does not thrash indefinitely.
## 6. Relationship to Other Computational Models
>FSM vs SSM
FSM: discrete symbolic states
SSM: emergent attractors in thermal space
>Turing Machine vs SSM
TM: behavior defined by tape symbols and transitions
SSM: behavior defined by adaptation toward equilibrium
>Markov Chain vs SSM
MC: probabilistic transitions between explicit states
SSM: deterministic drift toward stable basins, driven by observation
>Neural Network vs SSM
NN: learns weights
SSM: reorganizes runtime structures, not parameters
## 7. Why This Breaks New Ground
### A Steady State Machine is the first:
- runtime-thermodynamic computing model
- adaptive VM with convergence properties
- load-dependent reordering machine
- self-optimizing dictionary machine
- phase-shifting execution system
- state-machine whose state is emergent, not explicit
- This is a fundamental new category of computation.
## 8. Canonical Example: StarForth
### StarForth implements the first real SSM with:
- execution_heat
- rolling diversity windows
- phase-based lookup strategies
- adaptive dictionary ordering
- pipeline activation
- DOE-calibrated inference loop
- background physics monitor
- It is, quite literally, the reference implementation.
## 9. Formal Summary Statement
>A Steady State Machine (SSM) is a computational system in which the operational state is defined by attractor convergence in thermodynamic runtime space rather than by discrete symbolic transitions. An SSM continuously measures workload characteristics, adjusts internal structures, and stabilizes into performance-optimized steady states. When perturbed, it transitions through transient or chaotic phases before finding a new equilibrium. This model defines a new class of adaptive computational systems with properties distinct from finite state machines, pushdown automata, and Turing Machines.