Skip to content

GitLab

Commit 80c996e5 authored by danyao12's avatar danyao12 Committed by wunhuang
Browse files

set grouped bwd occupancy to 1

parent 3129c942
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 2 deletions
include/ck/tensor_operation/gpu/device/impl/device_grouped_multihead_attention_backward_xdl_cshuffle_v1.hpp
View file @ 80c996e5
...@@ -37,7 +37,7 @@ template <typename GridwiseGemm, ...@@ -37,7 +37,7 @@ template <typename GridwiseGemm,
bool HasMainKBlockLoop> bool HasMainKBlockLoop>
__global__ void __global__ void
#if CK_USE_LAUNCH_BOUNDS #if CK_USE_LAUNCH_BOUNDS
__launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU) __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, /*CK_MIN_BLOCK_PER_CU*/ 1)
#endif #endif
kernel_grouped_multihead_attention_backward_xdl_cshuffle_v1( kernel_grouped_multihead_attention_backward_xdl_cshuffle_v1(
const void CK_CONSTANT_ADDRESS_SPACE* group_kernel_args, const void CK_CONSTANT_ADDRESS_SPACE* group_kernel_args,
......
include/ck/tensor_operation/gpu/device/impl/device_grouped_multihead_attention_backward_xdl_cshuffle_v2.hpp
View file @ 80c996e5
...@@ -37,7 +37,7 @@ template <typename GridwiseGemm, ...@@ -37,7 +37,7 @@ template <typename GridwiseGemm,
bool HasMainKBlockLoop> bool HasMainKBlockLoop>
__global__ void __global__ void
#if CK_USE_LAUNCH_BOUNDS #if CK_USE_LAUNCH_BOUNDS
__launch_bounds__(CK_MAX_THREAD_PER_BLOCK, CK_MIN_BLOCK_PER_CU) __launch_bounds__(CK_MAX_THREAD_PER_BLOCK, /*CK_MIN_BLOCK_PER_CU*/ 1)
#endif #endif
kernel_grouped_multihead_attention_backward_xdl_cshuffle_v2( kernel_grouped_multihead_attention_backward_xdl_cshuffle_v2(
const void CK_CONSTANT_ADDRESS_SPACE* group_kernel_args, const void CK_CONSTANT_ADDRESS_SPACE* group_kernel_args,
......
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