Skip to content

[OpenCL:Chore] Update shared_gather_buf md5 after kernel fix in #4659#4660

Open
Qxinyu wants to merge 1 commit into
masterfrom
feature/opencl-sharedgather-md5
Open

[OpenCL:Chore] Update shared_gather_buf md5 after kernel fix in #4659#4660
Qxinyu wants to merge 1 commit into
masterfrom
feature/opencl-sharedgather-md5

Conversation

@Qxinyu

@Qxinyu Qxinyu commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Follow-up to [OpenCL:Bugfix] Fix maskless attention setArg overflow causing wrong SD UNet output on Adreno (#4653) #4659: the shared_gather_buf.cl fix updated the embedded kernel source (shared_gather_buf_mnn_cl.cpp) but not its entry in OpenCLProgramMd5Map (opencl_source_map.hpp), because opencl_codegen.py writes that header to the script's CWD rather than the cl/ directory.
  • The md5 is used to invalidate stale tuning caches when kernel source changes; this syncs it to the fixed kernel (74b2cbe8...7bfae0b1..., verified by re-running the codegen — the only real md5 change in the regenerated map).

Test plan

  • Re-ran opencl_codegen.py from the cl/ directory and confirmed shared_gather_buf is the only md5 that differs; all other diffs are brace-formatting noise (reverted).

@CLAassistant

Copy link
Copy Markdown

CLA assistant check
Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
You have signed the CLA already but the status is still pending? Let us recheck it.

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants