Skip to content

docs(benchmarks): gate-971 section 8 points the embedding stall at the shared slot (#975) - #1781

Merged
MakazhanAlpamys merged 1 commit into
mainfrom
docs/975-gate-971-section-8-note
Oct 12, 2026
Merged

MakazhanAlpamys merged 1 commit into
mainfrom
docs/975-gate-971-section-8-note

Conversation

@MakazhanAlpamys

Copy link
Copy Markdown
Owner

Adds the dated correction note promised on #975 (2026-10-04) to benchmarks/gate-971-async-nvme-source.md, section 8.

Section 8 still explained the embedding stall with the _plan_queue single-member-group sentence (opening paragraph, and again in Finding 8a) and did not point at #975. Two inline notes in the record's own style (like its 2026-09-14 16:45 corrections) now say that for the embedding entry the shared large-layer slot described in #975 supersedes that explanation (embed_tokens and lm_head share one device slot, so each step's _prime() re-copies the whole matrix host to device; fixed by #1178), and that the record's other findings, including the 0.07-0.12% materiality ruling, stand. The original sentences are left in place.

Docs only: one file, 11 added lines, no code, no changelog fragment (the record is not user-facing documentation). Cyrillic check clean; pre-commit hooks pass.

Closes #975.

@MakazhanAlpamys MakazhanAlpamys added the ci:full Run the full CI matrix on this PR (a maintainer adds it at approval) label Oct 12, 2026
@MakazhanAlpamys
MakazhanAlpamys merged commit ca8a139 into main Oct 12, 2026
23 checks passed
@MakazhanAlpamys
MakazhanAlpamys deleted the docs/975-gate-971-section-8-note branch October 12, 2026 16:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ci:full Run the full CI matrix on this PR (a maintainer adds it at approval)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

The single shared large-layer slot forces a full embedding-matrix H2D every step on an untied checkpoint (~33 ms for 500 MiB, measured at 16 GB/s)

1 participant