Unverified Commit ade81f17 authored by Kevin McKay's avatar Kevin McKay Committed by GitHub
Browse files

[Bugfix][Hardware][AMD] Gate FP4 ops on gfx950 to prevent MI300X crash (#35250)


Signed-off-by: default avatarc0de128 <kevin.mckay@outlook.com>
parent 6042e66c
...@@ -1052,12 +1052,16 @@ class rocm_aiter_ops: ...@@ -1052,12 +1052,16 @@ class rocm_aiter_ops:
@classmethod @classmethod
@if_aiter_supported @if_aiter_supported
def is_fp4bmm_enabled(cls) -> bool: def is_fp4bmm_enabled(cls) -> bool:
return cls._AITER_ENABLED and cls._FP4BMM_ENABLED from vllm.platforms.rocm import on_gfx950
return cls._AITER_ENABLED and cls._FP4BMM_ENABLED and on_gfx950()
@classmethod @classmethod
@if_aiter_supported @if_aiter_supported
def is_asm_fp4_gemm_dynamic_quant_enabled(cls) -> bool: def is_asm_fp4_gemm_dynamic_quant_enabled(cls) -> bool:
return cls._AITER_ENABLED and cls._FP4_GEMM_DYNAMIC_QUANT_ASM from vllm.platforms.rocm import on_gfx950
return cls._AITER_ENABLED and cls._FP4_GEMM_DYNAMIC_QUANT_ASM and on_gfx950()
@classmethod @classmethod
@if_aiter_supported @if_aiter_supported
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment