ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-16 02:54:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix for Add the API to load SGPR (#2913)

Browse Source 
* Revert "Revert "[CK-Tile] Add the API to load SGPR  (#2878)" (#2904)"

This reverts commit 5cc40c160f.

* Fix: sgpr minor issue

* cyclic dependency resolved

* clang formatted

* removing unused variable

* clang formatted

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

[ROCm/composable_kernel commit: b56e5d1d79]
This commit is contained in:
Khushbu Agarwal
2025-09-25 10:32:42 -07:00
committed by GitHub
parent 5a39b14c52
commit bb5eeef2af
41 changed files with 224 additions and 173 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
include/ck_tile/ops/reduce/kernel/reduce2d_kernel.hpp
View File

@@ -156,7 +156,7 @@ struct Reduce
const auto merged_reduce_len =
transformed_x_tensor.get_tensor_descriptor().get_lengths().at(number<1>{});
index_t num_n_tile_iteration =
__builtin_amdgcn_readfirstlane(integer_divide_ceil(merged_reduce_len, S::Block_N));
amd_wave_read_first_lane(integer_divide_ceil(merged_reduce_len, S::Block_N));
auto block_reduce2d = Policy::template GetBlockReduce2d<Problem>();
auto block_reduce2d_sync = Policy::template GetBlockReduce2dSync<Problem>();
@@ -167,7 +167,7 @@ struct Reduce
auto y_compute = block_reduce2d.template MakeYBlockTile<XTensorType>();
set_tile(y_compute, reduce_func.template GetIdentityValue<ComputeDataType>());
for(int iN = __builtin_amdgcn_readfirstlane(0); iN < num_n_tile_iteration; ++iN)
for(int iN = amd_wave_read_first_lane(0); iN < num_n_tile_iteration; ++iN)
{
const auto x = load_tile(x_window);
block_reduce2d(x, y_compute, reduce_func);