Commit 782f6092 authored by yuguo's avatar yuguo
Browse files

[DCU] fix compile

parent eac75188
...@@ -5,8 +5,8 @@ ...@@ -5,8 +5,8 @@
************************************************************************/ ************************************************************************/
#include <assert.h> #include <assert.h>
#include <cooperative_groups.h>
#include <cuda_runtime.h> #include <cuda_runtime.h>
#include <cooperative_groups.h>
#include <transformer_engine/fused_router.h> #include <transformer_engine/fused_router.h>
#include "../common.h" #include "../common.h"
......
...@@ -9,6 +9,10 @@ ...@@ -9,6 +9,10 @@
#include "transformer_engine/transformer_engine.h" #include "transformer_engine/transformer_engine.h"
#ifdef __HIP_PLATFORM_AMD__
#define __syncwarp __syncthreads
#endif
namespace transformer_engine { namespace transformer_engine {
constexpr size_t kThreadsPerWarp = 32; constexpr size_t kThreadsPerWarp = 32;
......
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