Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 6 additions & 3 deletions exllamav2/exllamav2_ext/cuda/layer_norm.cu
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,12 @@
#include "compat.cuh"

#if defined(USE_ROCM)
#define NUM_WARPS (1024 / warpSize)
#define WARP_SIZE (warpSize)
#if defined(__GFX8__) || defined(__GFX9__)
#define WARP_SIZE 64
#else
#define WARP_SIZE 32
#endif
#define NUM_WARPS (1024 / WARP_SIZE)
#else
#define NUM_WARPS 32
#define WARP_SIZE 32
Expand Down Expand Up @@ -230,4 +234,3 @@ void layer_norm_cuda_update_y
{
graph->update_param_ptr(label, 0, 3, y);
}

8 changes: 6 additions & 2 deletions exllamav2/exllamav2_ext/cuda/rms_norm.cu
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,12 @@
#include "compat.cuh"

#if defined(USE_ROCM)
#define NUM_WARPS (1024 / warpSize)
#define WARP_SIZE (warpSize)
#if defined(__GFX8__) || defined(__GFX9__)
#define WARP_SIZE 64
#else
#define WARP_SIZE 32
#endif
#define NUM_WARPS (1024 / WARP_SIZE)
#else
#define NUM_WARPS 32
#define WARP_SIZE 32
Expand Down