mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-06-30 03:37:38 +00:00
add vmcnt guard for async copy
This commit is contained in:
@@ -50,14 +50,14 @@ using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMX_Xdl_CShuffle
|
||||
GemmSpec, // GemmSpec
|
||||
ScaleBlockSize, // ScaleBlockSize: Scaling block size
|
||||
256, // BlockSize: Thread block size
|
||||
128, // MPerBlock
|
||||
256, // MPerBlock
|
||||
256, // NPerBlock
|
||||
KPerBlock, // KPerBlock
|
||||
16, // AK1
|
||||
16, // BK1
|
||||
16, // MPerXDL
|
||||
16, // NPerXDL
|
||||
4, // MXdlPerWave
|
||||
8, // MXdlPerWave
|
||||
8, // NXdlPerWave
|
||||
S<8, 32, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
|
||||
S<1, 0, 2>, // ABlockTransferThreadClusterArrangeOrder
|
||||
|
||||
Reference in New Issue
Block a user