ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-20 21:09:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix for beta!=0 in reduce (#1440)

Browse Source 
* fix for beta!=0 in reduce

* add reviewers suggestions

[ROCm/composable_kernel commit: b74d4d4d54]
This commit is contained in:
jakpiase
2024-08-06 18:10:39 +02:00
committed by GitHub
parent 458d8bef26
commit ec527c487e
3 changed files with 16 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
example/12_reduce/reduce_blockwise_impl.hpp
View File

@@ -1,5 +1,5 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
@@ -316,7 +316,17 @@ int reduce_blockwise_impl(bool do_verification,
auto invoker_ptr = reduce.MakeInvokerPointer();
float avg_time = invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, time_kernel});
int log_level = 0, cold_niters = 5, nrepeat = 50;
if(beta != 0.0f)
{
std::cerr << "Warning: With beta != 0.0f there must be only one repeat for correct results "
"since out memory is being overwritten."
<< std::endl;
cold_niters = 0;
nrepeat = 1;
}
float avg_time = invoker_ptr->Run(
argument_ptr.get(), StreamConfig{nullptr, time_kernel, log_level, cold_niters, nrepeat});
std::size_t num_bytes = invariant_total_length * reduce_total_length * sizeof(InOutDataType) +
invariant_total_length * sizeof(InOutDataType);