mudler / mudler/vllm.cpp

GFX1100-TG200: GDN b/a GEMV merge (VT_GDN_MERGED_BA_ROCM) — closed negative

Open
#2,888 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
423
Forks
53
Avg merge
20h 26m
Merged PRs (30d)
310

Description

Row: GFX1100-TG200

Summary

GDN b/a same-input GEMV merge: merge the b and a projections of the GDN block when they share the same input tensor. Adjudicated bit-identical, closed below the bar in the original campaign.

Lever

VT_GDN_MERGED_BA_ROCM=1 (default OFF).

Benchmark

A/B interleaved, 5 pairs, Qwen3.5-4B Q4_K_M, 256 tokens, temp 0, seed 0:

A (T25 chain) B (T25+T35r3) Delta
Median 32.40 32.40 -0.05%

Noise — confirms the original campaign's negative disposition.

Token identity

PASS — identical output to T25 chain.

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.

Research direction

Start with the issue's benchmark and token-identity results for GFX1100-TG200, then inspect the VT_GDN_MERGED_BA_ROCM=1 lever if an implementation target is later identified. The payload names no files, tests, or requested change; the documented outcome is a negative result, so completion criteria are not defined.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
18/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.