Signed-off-by: Roger Wang <ywang@roblox.com>
This commit is contained in:
@@ -31,7 +31,8 @@ void cutlass_scaled_fp4_mm(torch::Tensor& D, torch::Tensor const& A,
|
||||
#if defined ENABLE_NVFP4 && ENABLE_NVFP4
|
||||
return cutlass_scaled_fp4_mm_sm100a(D, A, B, A_sf, B_sf, alpha);
|
||||
#endif
|
||||
TORCH_CHECK_NOT_IMPLEMENTED(false, "No compiled nvfp4 mm kernel, vLLM should "
|
||||
"be compiled using CUDA 12.8 and target "
|
||||
"compute capability 100 or above.");
|
||||
TORCH_CHECK_NOT_IMPLEMENTED(false,
|
||||
"No compiled nvfp4 mm kernel, vLLM should "
|
||||
"be compiled using CUDA 12.8 and target "
|
||||
"compute capability 100 or above.");
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user