mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-12 09:16:52 +00:00
f63a23acb1
* update online kernel wrapper bundle all descriptors in a tuple * change __CONSTANT__ to CONSTANT * rename * adding tuning * added IsValidCompileParameter * reorginze * adding tunable for fp16 and int8 * fix kernel compile warning and bug fixes * suppress warning about cast CONSTANT (address space 4) pointer * fix building issue
22 lines
324 B
C++
22 lines
324 B
C++
#ifndef CK_SYNCHRONIZATION_AMD_HPP
|
|
#define CK_SYNCHRONIZATION_AMD_HPP
|
|
|
|
#include "config.hpp"
|
|
|
|
namespace ck {
|
|
|
|
__device__ void block_sync_lds()
|
|
{
|
|
#if CK_BLOCK_SYNC_LDS_WITHOUT_SYNC_VMEM
|
|
asm volatile("\
|
|
s_waitcnt lgkmcnt(0) \n \
|
|
s_barrier \
|
|
" ::);
|
|
#else
|
|
__syncthreads();
|
|
#endif
|
|
}
|
|
|
|
} // namespace ck
|
|
#endif
|