Skip to content

GitLab

Unverified Commit 3be8d312 authored by Tyler Michael Smith's avatar Tyler Michael Smith Committed by GitHub
Browse files

[Kernel][Bugfix] Fixup some warnings in nvfp4_blockwise_moe when CUDA < 12.8 (#20324) 


Signed-off-by: default avatarTyler Michael Smith <tyler@neuralmagic.com>
parent 3abfe221
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 0 deletions
csrc/quantization/fp4/nvfp4_blockwise_moe_kernel.cu
View file @ 3be8d312
...@@ -335,8 +335,10 @@ void run_fp4_blockwise_scaled_group_mm( ...@@ -335,8 +335,10 @@ void run_fp4_blockwise_scaled_group_mm(
TORCH_CHECK(status == cutlass::Status::kSuccess, "Failed to run GEMM"); TORCH_CHECK(status == cutlass::Status::kSuccess, "Failed to run GEMM");
} }
#if defined ENABLE_NVFP4 && ENABLE_NVFP4
constexpr auto FLOAT4_E2M1X2 = at::ScalarType::Byte; constexpr auto FLOAT4_E2M1X2 = at::ScalarType::Byte;
constexpr auto SF_DTYPE = at::ScalarType::Float8_e4m3fn; constexpr auto SF_DTYPE = at::ScalarType::Float8_e4m3fn;
#endif
#define CHECK_TYPE(x, st, m) \ #define CHECK_TYPE(x, st, m) \
TORCH_CHECK(x.scalar_type() == st, ": Inconsistency of Tensor type:", m) TORCH_CHECK(x.scalar_type() == st, ": Inconsistency of Tensor type:", m)
......
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