mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-15 10:37:44 +00:00
Merge commit 'e31a7a4f29b371c32ea9daf9211b6ae1fed2fa40' into develop
This commit is contained in:
@@ -429,8 +429,8 @@ struct GridwiseGemmMX_xdl_cshuffle_v3_bpreshuffle
|
||||
constexpr index_t MWave = MPerBlock / (MXdlPerWave * MPerXdl);
|
||||
constexpr index_t WaveSize = BlockSize / (MWave * NWave);
|
||||
constexpr index_t NkSwizzleNumber = Number<WaveSize * KPack>{};
|
||||
return make_naive_tensor_descriptor_packed(
|
||||
make_tuple(N0 / NWave / NXdlPack, NWave, NXdlPack, K0, NkSwizzleNumber));
|
||||
return make_naive_tensor_descriptor_packed(make_tuple(
|
||||
math::integer_divide_ceil(N0, NWave * NXdlPack), NWave, NXdlPack, K0, NkSwizzleNumber));
|
||||
}
|
||||
|
||||
__host__ __device__ static auto MakeBGridDescriptor_BK0_N_BK1(
|
||||
|
||||
Reference in New Issue
Block a user