The T34 host-split evidence file the GFX1100-TG200 spec cites was never contributed

Open Beginner friendly
#2,936 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
68/100
Issue type
Documentation
Clarity
Mostly clear
Activity status
Active

Research direction

Read .agents/specs/gfx1100-tg200.md around ## Now and ## Owed, then inspect the existing bench-evidence files and scripts/check-agent-record.py for the expected evidence format. Use the capture details stated in the spec to create docs/bench-evidence/gfx1100-tg200-t34-host-split-20260829.md. Done means the evidence file exists with the cited measurement and the agent-record check no longer reports a dangling link.

Written by the indexing model from the issue text.

Description

Row: GFX1100-TG200

.agents/specs/gfx1100-tg200.md ## Now links its T34 lever ranking to
docs/bench-evidence/gfx1100-tg200-t34-host-split-20260829.md. That file is in
no branch of this campaign: git ls-tree over every open row/GFX1100-TG200-*
pull request head finds nothing matching t34-host-split, and it is not on
main.

scripts/check-agent-record.py reports it as
dangling link ../../docs/bench-evidence/gfx1100-tg200-t34-host-split-20260829.md,
which is why the agent-record job is red on the pull requests carrying that
spec text.

The measurement itself is not lost -- the ## Now prose states the capture in
full (hipGraph replay, hipStreamSynchronize 8.75 ms, the 7.56/4.38 in-kernel
split, ~522 inter-kernel gaps at ~4.1 us, ~523 kernels/token, and the T35/T36/T37
ranking priced on it). Only the evidence file backing it is missing, so the
numbers currently have no capture a reader can check.

Fixed in the same flow only to the extent of the dangling citation: the link is
removed and the prose names this issue instead. Contributing the evidence file
is still owed
, and the spec lists it under ## Owed.

Dominant language
C++
Stars
423
Forks
53
Avg merge
19h 7m
Merged PRs (30d)
337

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from mudler/vllm.cpp

All issues in mudler/vllm.cpp

Similar issues

More C++ issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.