Merge branch 'main' into patches
This commit is contained in:
@@ -146,7 +146,6 @@ if TYPE_CHECKING:
|
||||
VLLM_ROCM_AITER_MLA_ASM_PADDING: Literal["auto", "gluon", "asm"] = "auto"
|
||||
VLLM_ROCM_AITER_MLA_DCP_VERIFY: Literal["auto", "asm", "segmented"] = "auto"
|
||||
VLLM_ROCM_USE_AITER_MHA: bool = True
|
||||
VLLM_ROCM_USE_AITER_FP4_ASM_GEMM: bool = False
|
||||
VLLM_ROCM_USE_AITER_TRITON_SPARSE_MLA: bool = False
|
||||
VLLM_ROCM_USE_AITER_TRITON_ROPE: bool = False
|
||||
VLLM_ROCM_USE_AITER_FP8BMM: bool = False
|
||||
@@ -1346,11 +1345,6 @@ environment_variables: dict[str, Callable[[], Any]] = {
|
||||
"VLLM_ROCM_USE_AITER_MHA": lambda: (
|
||||
os.getenv("VLLM_ROCM_USE_AITER_MHA", "True").lower() in ("true", "1")
|
||||
),
|
||||
# Whether to use aiter fp4 gemm asm.
|
||||
# By default is disabled.
|
||||
"VLLM_ROCM_USE_AITER_FP4_ASM_GEMM": lambda: (
|
||||
os.getenv("VLLM_ROCM_USE_AITER_FP4_ASM_GEMM", "False").lower() in ("true", "1")
|
||||
),
|
||||
# Whether sparse MLA prefill and decode run on aiter's Triton kernel, which
|
||||
# reads the KV cache as stored (bf16 or fp8, paged or flat). Used by the
|
||||
# ROCM_AITER_MLA_SPARSE backend (DeepSeek V3.2, GLM-5.x) and DeepSeek
|
||||
|
||||
Reference in New Issue
Block a user