diff options
| author | Cedric Nugteren <web@cedricnugteren.nl> | 2019-05-07 20:47:31 +0200 |
|---|---|---|
| committer | Cedric Nugteren <web@cedricnugteren.nl> | 2019-05-07 20:47:31 +0200 |
| commit | c6ba86cdc3da8ab9599fff5483cedbe5b1422687 (patch) | |
| tree | ec9bff002c8270302691505eabe79d34e82b68ed /src | |
| parent | 774cebaa404080ea725cc57b7a433d891e85d8a2 (diff) | |
Enabled avc_motion_estimation extension for Intel subgroup shuffling
Diffstat (limited to 'src')
| -rw-r--r-- | src/kernels/level3/xgemm_part1.opencl | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/src/kernels/level3/xgemm_part1.opencl b/src/kernels/level3/xgemm_part1.opencl index cc03696e..306280bc 100644 --- a/src/kernels/level3/xgemm_part1.opencl +++ b/src/kernels/level3/xgemm_part1.opencl @@ -125,8 +125,9 @@ R"( #define USE_SUBGROUP_SHUFFLING 0 // Optionally enables subgroup shuffling for Intel GPUs #endif -// Intel subgroups (https://www.khronos.org/registry/OpenCL/extensions/intel/cl_intel_subgroups.txt) +// Intel subgroups (https://www.khronos.org/registry/OpenCL/extensions/intel/cl_intel_subgroups.html) #if USE_SUBGROUP_SHUFFLING == 1 && SUBGROUP_SHUFFLING_INTEL == 1 + #pragma OPENCL EXTENSION cl_intel_device_side_avc_motion_estimation: enable #define SUBGROUP_SIZE 8 // Assumes subgroup size is always 8 on Intel GPUs #endif |