Skip to content

blog: three dies, one verdict -- R3-3 makes a silicon check something you run once (Closes #1568) - #1569

Merged
gHashTag merged 1 commit into
mainfrom
claude/blog-three-dies-one-verdict
Oct 9, 2026
Merged

gHashTag merged 1 commit into
mainfrom
claude/blog-three-dies-one-verdict

Conversation

@gHashTag

@gHashTag gHashTag commented Oct 9, 2026

Copy link
Copy Markdown
Owner

Closes #1568

A new post for t27.ai, EN and RU: Three dies, one verdict: a silicon check you run once.

Files: the post body, its index entry, and its registration in posts.ts.

{
  "version": 1,
  "head_sha": "647ead892c84a7e76ac5e351da7cb357c5c83736",
  "summary": "A new t27.ai post, in English and Russian, on R3-3: one t27 spec checked on three distinct FPGA dies, and what a citable silicon verdict changes for build cost.",
  "changes": [
    "src/data/blog/bodies/three-dies-one-verdict.ts: the post body, EN and RU, 14 blocks each, every number from the R3-3 receipts, run-record or a spec.",
    "src/data/blog/index.ts: the post's entry with summary, tags, eight receipts, four open questions and the Russian meta.",
    "src/data/blog/posts.ts: the body registered under its slug, as every post is."
  ],
  "tests": [
    {
      "command": "node --experimental-strip-types --check on the three changed files",
      "status": "passed",
      "result": "all three parse as TypeScript with exit code 0",
      "evidence": "local run on the three files of this head, 2026-10-09"
    },
    {
      "command": "t27c run-record --challenge e61798b78cbb6b287b8ff8523cea71a6 specs/fpga/ternary_link.t27 on t27 master 8eb0ff41c",
      "status": "passed",
      "result": "3 receipts, run complete, FRESH, Die NAMED, INDEP_DIES, citable",
      "evidence": "local run with the t27c-v0.5.1 tag build, 2026-10-09"
    },
    {
      "command": "npm run typecheck in apps/website",
      "status": "not_run",
      "result": "the full website typecheck was not run locally",
      "evidence": "website-checks CI on this pull request runs it"
    }
  ],
  "limitations": [
    "The speed claim is about what a citable verdict allows under reuse.t27; CI does not yet skip silicon checks by receipt, and the post says so.",
    "The site deploy is separate; the post appears on t27.ai only after the apex publish succeeds."
  ],
  "tags": [
    "Blog",
    "FPGA",
    "Verification"
  ],
  "blog": {
    "title": "Three dies, one verdict",
    "summary": "How one t27 spec was checked on three distinct FPGA dies, and why a citable hardware verdict lets a silicon check run once instead of on every build.",
    "outline": [
      "ternary_link ran on three distinct dies on three machines, one of them driven over USB/IP from 410 ms away, and run-record calls the run citable.",
      "The second die found a bug the first could not: the DNA tail differs per die, fixed in t27c 0.5.1.",
      "With a citable verdict, the five-part reuse rule lets a silicon check run once per spec version and toolchain; CI wiring is the next step."
    ]
  }
}

🤖 Generated with Claude Code

… you run once (Closes #1568)

The owner asked (2026-10-09, translated): an X post and a blog post about R3-3,
and what it gives for build speed for everyone.

The post, EN and RU: specs/fpga/ternary_link.t27 checked on three distinct dies
(A 050d58218fd9854, B 0389c0c2d85e85c, C 050a5824d85e85c) on three machines,
die C driven over USB/IP from 410 ms away; run-record INDEP_DIES, citable; die B
found the per-die DNA tail bug (t27#7761, t27c 0.5.1). The speed section ties a
citable silicon verdict to the five-part reuse rule of specs/verified/reuse.t27,
and says plainly what is not done: one spec so far, CI does not yet skip a
silicon check by receipt, INDEP_DIES not INDEP_OPERATORS, DNA is not a secret.

Files: bodies/three-dies-one-verdict.ts, its index entry, its registration in
posts.ts.

Closes #1568

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@gHashTag
gHashTag enabled auto-merge (squash) October 9, 2026 07:54
@t27-bees

t27-bees Bot commented Oct 9, 2026

Copy link
Copy Markdown

Summary by t27-bees

This PR adds a new blog post "Three dies, one verdict" in English and Russian, documenting the first t27 spec (ternary_link) verified on three distinct FPGA dies with a citable run-record verdict (INDEP_DIES). The post explains how this enables reusing silicon checks via reuse.t27's five-part rule instead of repeating them per build.

Worth a look

  • apps/website/src/data/blog/posts.ts:1, +104 — The new import and body registration follow the existing pattern exactly. No type mismatch visible; the slug matches the index entry.
  • apps/website/src/data/blog/index.ts:3–40 — The PostMeta entry is complete: slug, title, summary, date, readingMinutes, tags, eight receipts, four open questions, and Russian meta. All URLs point to existing GitHub issues/PRs and repo files. The published: true flag means it will appear on t27.ai once the site publishes.
  • apps/website/src/data/blog/bodies/three-dies-one-verdict.ts — Both body and ruBody are well-formed Block[] arrays (14 blocks each). All cited numbers (die DNAs, challenge prefix, 60.75 s place-and-route, 410 ms RTT, 0x4F vs 0x37 DNA tail) trace to the receipts and issues listed in the index. No Markdown or HTML in the blocks—only the expected p and h kinds.
  • apps/website/src/data/blog/index.ts:3 — The new entry is prepended to postsIndex, making it the first post. If the site renders posts in array order, this is intentional (latest first). If the site sorts by date, the "2026-10-09" date will place it correctly regardless.
  • apps/website/src/data/blog/posts.ts:1 — The import uses named exports body and ruBody from the new file. The file exports exactly those names, so the binding is correct.

Read: 3 files, +132 -0, the whole diff. Head 647ead8. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.

@gHashTag
gHashTag merged commit ce7aef7 into main Oct 9, 2026
20 checks passed
@github-actions github-actions Bot added the status:in-progress 🔵 Agent working label Oct 9, 2026
@github-actions

github-actions Bot commented Oct 9, 2026

Copy link
Copy Markdown
Contributor

X Brain Health Check

Score: 100.0/100
Status: 🟢 HEALTHY
Threshold: 80/100

X Brain is above merge threshold

@github-actions github-actions Bot added status:completed Done and removed status:in-progress 🔵 Agent working labels Oct 9, 2026
github-actions Bot added a commit that referenced this pull request Oct 9, 2026
blog: three dies, one verdict -- R3-3 makes a silicon check something you run once (Closes #1568) (#1569)

The owner asked (2026-10-09, translated): an X post and a blog post about R3-3,
and what it gives for build speed for everyone.

The post, EN and RU: specs/fpga/ternary_link.t27 checked on three distinct dies
(A 050d58218fd9854, B 0389c0c2d85e85c, C 050a5824d85e85c) on three machines,
die C driven over USB/IP from 410 ms away; run-record INDEP_DIES, citable; die B
found the per-die DNA tail bug (t27#7761, t27c 0.5.1). The speed section ties a
citable silicon verdict to the five-part reuse rule of specs/verified/reuse.t27,
and says plainly what is not done: one spec so far, CI does not yet skip a
silicon check by receipt, INDEP_DIES not INDEP_OPERATORS, DNA is not a secret.

Files: bodies/three-dies-one-verdict.ts, its index entry, its registration in
posts.ts.

Closes #1568

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Oct 9, 2026

Copy link
Copy Markdown
Contributor

🧪 Brain Stress Test Results

Tests: 261/261 passed, 0 leaked
Status: 🟢 PASS
Gate: every test in zig build test-brain-stress passes and none leaks

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

blog: three dies, one verdict -- R3-3 makes a silicon check something you run once

1 participant