Atomics Reduce Counting Optimization for SplitK Skinny GEMMs. (#29843)
Signed-off-by: Hashem Hashemi <hashem.hashemi@amd.com>
This commit is contained in:
@@ -2072,6 +2072,12 @@ def wvSplitK(
|
||||
return torch.ops._rocm_C.wvSplitK(a, b, bias, cu_count)
|
||||
|
||||
|
||||
def wvSplitKrc(
|
||||
a: torch.Tensor, b: torch.Tensor, cu_count: int, bias: torch.Tensor = None
|
||||
) -> torch.Tensor:
|
||||
return torch.ops._rocm_C.wvSplitKrc(a, b, bias, cu_count)
|
||||
|
||||
|
||||
def wvSplitKQ(
|
||||
a: torch.Tensor,
|
||||
b: torch.Tensor,
|
||||
|
||||
Reference in New Issue
Block a user