Skip to content

[Issue]:mi350 f4gemm kernel 256x256 corner case hang #964

@junxiaguo

Description

@junxiaguo

Problem Description

  • set git head back to commit d4ea3cf (HEAD -> main, tag: v0.1.5)

  • run python3 op_tests/test_gemm_a4w4.py for kernel _ZN5aiter42f4gemm_bf16_per1x32Fp4_BpreShuffle_256x256E with ksplit=0, M,N,K = (32,7168,256)

  • case core dumped for Memory access fault

Operating System

docker image: rocm/aigmodels-private:mi350-sglang-dsr1-250621-rc1

CPU

general

GPU

mi350

ROCm Version

general

ROCm Component

No response

Steps to Reproduce

No response

(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support

No response

Additional Information

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions