ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 13:11:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

[Ck_tile] hot fix, fix rpcf param setting err (#1657)

Browse Source 
Co-authored-by: dummycoderfe <noplydummmycoder@163.com>
This commit is contained in:
feli
2024-11-14 14:06:36 +08:00
committed by GitHub
parent efd9261545
commit c1f8d53ce8
3 changed files with 21 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
include/ck_tile/ops/welford/block/block_welford.hpp
View File

@@ -47,8 +47,11 @@ struct BlockWelford
auto x = ck_tile::type_convert<ComputeDataType>(x_tensor[in_dstr_idx]);
welford_update(
mean_tensor(out_dstr_idx), var_tensor(out_dstr_idx), x, cur_count_);
welford_update(mean_tensor(out_dstr_idx),
var_tensor(out_dstr_idx),
x,
cur_count_,
constant<kFastFDiv>{});
});
}
});
@@ -159,7 +162,8 @@ struct BlockWelfordSync
v_local_count,
v_remote_mean,
v_remote_var,
v_remote_count);
v_remote_count,
constant<kFastFDiv>{});
});
}
});
@@ -307,7 +311,8 @@ struct BlockWelfordCrossWarpSync
v_local_count,
v_remote_mean,
v_remote_var,
v_remote_count);
v_remote_count,
constant<kFastFDiv>{});
});
mean_tensor.get_thread_buffer()(i_0) = v_local_mean;