docs: TRI-NET positioning package (Closes #627)#693
Merged
Conversation
…COMPETITORS, BENCHMARKS, CLARA_TRACEABILITY
Adds the documentation package for positioning t27 as the fourth product
in the TRI-NET line: the spec-first toolchain and numeric format registry,
feeding the three Tiny Tapeout sibling chips (tt-trinity-phi 1x1,
tt-trinity-euler 8x2, tt-trinity-gamma 8x4).
New docs (all root-level, evidence-based, conservative wording):
- STATUS.md - six readiness levels (SPEC/RTL/SIM/SYNTH/GDS/SILICON);
per-component levels derived from this repo only;
explicit "conservative status decisions" section.
- LINEUP.md - single-page map of the four products and how they
relate; ASCII diagram; no per-chip status (lives in
the chip repos).
- FORMAT_REGISTRY.md - human mirror of conformance/FORMAT-SPEC-001.json:
GF16 primary, GoldenFloat family table, TF3 ternary,
FP8 / NF4 / INT4 / INT8 bridges marked PLANNED.
- COMPETITORS.md - restrained, source-linked notes on Hailo-8, Coral,
Axelera Metis, Qualcomm Cloud AI 100 Ultra, MediaTek
Dimensity 9400+, BitNet b1.58, Tiny Tapeout. No TOPS
parity claimed.
- BENCHMARKS.md - register of what is benchmarked and what is NOT;
explicit policy against "expected" or "projected"
figures; how-to-add-a-benchmark contract.
- CLARA_TRACEABILITY.md - claim->file table mapping public DARPA CLARA
program goals (darpa.mil/research/programs/clara)
to artefacts in this repo; gaps marked honestly.
README.md first screen: inserts a "What this repo is" block linking to the
six new docs, names GF16 as the primary numeric path, and names the three
sibling chip repos plus Tiny Tapeout as submission channel. No other parts
of README touched.
Conservative status decisions (per STATUS.md sect. 3):
- No SILICON or GDS/TAPEOUT claim in t27 (those belong to chip repos).
- GF16 marked SIM, not SYNTH-on-vendor-cells.
- CLARA bridge marked demo/draft, not "submitted".
- Coq surface marked "partial".
Validation run:
- python3 scripts/check_first_party_doc_language.py -> DOC-LANG OK
- python3 -c "json.load(...FORMAT-SPEC-001.json)" -> primary GF16
- Rust suite (./scripts/tri test) not run: cargo not present in this env.
Closes #627
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
NOW Sync Gate (check-now-freshness CI) requires every PR/push to master to include an update to docs/NOW.md (see issue #141). This commit adds the TRI-NET positioning-package entry and bumps "Last updated" to today (2026-05-17). Conservative: NOW.md entry mirrors the PR description verbatim, no new claims; readiness ladder, primary GF16, planned FP8/NF4/INT4/INT8 bridges, and the "no code touched" guarantee are repeated for the dashboard. Local re-validation: - python3 scripts/check_first_party_doc_language.py -> DOC-LANG OK - FORMAT-SPEC-001.json sanity -> primary GF16 Refs #141 (coordination anchor for NOW.md) Closes #627 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
The PR Dashboard workflow's bash block runs under `set -euo pipefail` and referenced $TOTAL and $FAILING in the markdown rendering BEFORE assigning them, producing "TOTAL: unbound variable" and failing the `pr-dashboard` check on every PR (independent of PR content). Symptom seen on PR #693: /home/runner/work/_temp/<id>.sh: line 22: TOTAL: unbound variable Process completed with exit code 1 Fix: move the four jq aggregations (TOTAL, FAILING, READY, PENDING) above the table-rendering block, and use the existing FAILING name consistently (was split between FAILED in the aggregation and FAILING in the table). Pure CI repair, no docs/spec/gen change. Required to unblock the TRI-NET positioning-package PR's merge-ready state. Refs #627 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
This was referenced May 18, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds the documentation package positioning t27 as the fourth product in the TRI-NET line: the spec-first toolchain and numeric format registry that feeds the three Tiny Tapeout sibling chips (
tt-trinity-phi1x1,tt-trinity-euler8x2,tt-trinity-gamma8x4).New root-level docs:
conformance/FORMAT-SPEC-001.json: GF16 primary, GoldenFloat family table, TF3 ternary, FP8 / NF4 / INT4 / INT8 bridges marked PLANNED.darpa.mil/research/programs/clara) to artefacts in this repo; gaps marked honestly.README.mdfirst screen: a small "What this repo is" block was inserted (under the identity line, above System Status) that names GF16 as the primary numeric path, lists the three sibling chip repos with Tiny Tapeout as submission channel, and links to the six new docs. The rest of the README is unchanged.Conservative status decisions (per STATUS.md §3)
clara-bridge/marked demo / draft, not "submitted".External source links used
Test plan
python3 scripts/check_first_party_doc_language.py→ DOC-LANG OKconformance/FORMAT-SPEC-001.json→ primary GF16 confirmed./scripts/tri test— not run in this environment (cargo not installed)check-now-freshness,issue-gate,schema-validation, doc-language gateCloses #627
🤖 Generated with Claude Code