Do not use warpSize as compile time constant as it is removed (#2320)

* Do not use warpSize as compile time constant as it is removed

* Update tile_image_to_column_shape.hpp

update warpSize usage.

* clean-up all use of warpSize, make sure code builds

* fix

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: illsilin <Illia.Silin@amd.com>
Co-authored-by: Bartlomiej Kocot <barkocot@amd.com>
This commit is contained in:
Satyanvesh Dittakavi
2025-06-18 00:24:30 +05:30
committed by GitHub
parent 3af66e99ab
commit 4c57157d50
31 changed files with 213 additions and 206 deletions

View File

@@ -210,7 +210,7 @@ struct BlockReduce2dCrossWarpSync
// | w0 | w1 | w2 | w3 | -----> | w0123 |
//
// -> also store data from every wave into LDS
constexpr index_t num_warps = BlockShape::BlockSize / warpSize;
constexpr index_t num_warps = BlockShape::BlockSize / get_warp_size();
return num_warps * thread_buf_size * sizeof(DataType);
}
@@ -226,7 +226,7 @@ struct BlockReduce2dCrossWarpSync
const index_t lane_id = get_lane_id();
const index_t warp_id = get_warp_id();
constexpr auto num_reduce_warps = GetReduceWarps<YDistributedTensor_>();
constexpr index_t num_warps = BlockShape::BlockSize / warpSize;
constexpr index_t num_warps = BlockShape::BlockSize / get_warp_size();
const index_t smem_offset = warp_id;
// skip if nonthing to do