Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Debug #2459: Fix babelstream for SYCL backends #2460

Closed
wants to merge 2 commits into from
Closed

Debug #2459: Fix babelstream for SYCL backends #2460

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f1b4d3c
Rely on alpaka::isMultiThreadAcc instead of hand-picking the accelera…
fwyzard Jan 24, 2025
bf321d6
CI_FILTER: ^macos_xcode-15.4_release
fwyzard Jan 26, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 9 additions & 3 deletions benchmarks/babelstream/src/babelStreamMainTest.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -176,6 +176,14 @@ struct DotKernel
template<typename TAcc, typename T>
ALPAKA_FN_ACC void operator()(TAcc const& acc, T const* a, T const* b, T* sum, alpaka::Idx<TAcc> arraySize) const
{
// check that sum[arraySize] has enough elements
[[maybe_unused]] auto const numBlocks = alpaka::getWorkDiv<alpaka::Grid, alpaka::Blocks>(acc)[0];
ALPAKA_ASSERT_ACC(numBlocks <= arraySize);

// check that tbSum[blockThreadExtentMain] has enough elements
auto const blockSize = alpaka::getWorkDiv<alpaka::Block, alpaka::Threads>(acc)[0];
ALPAKA_ASSERT_ACC(blockSize <= blockThreadExtentMain);

using Idx = alpaka::Idx<TAcc>;
auto& tbSum = alpaka::declareSharedVar<T[blockThreadExtentMain], __COUNTER__>(acc);

Expand All @@ -188,7 +196,6 @@ struct DotKernel
threadSum += a[i] * b[i];
tbSum[local_i] = threadSum;

auto const blockSize = alpaka::getWorkDiv<alpaka::Block, alpaka::Threads>(acc)[0];
for(Idx offset = blockSize / 2; offset > 0; offset /= 2)
{
alpaka::syncBlockThreads(acc);
Expand Down Expand Up @@ -326,8 +333,7 @@ void testKernels()
auto getWorkDivForDotKernel = [&]<typename AccType>() -> alpaka::WorkDivMembers<Dim, Idx>
{
// Use babelstream standard work division for multi-threaded backends
if constexpr(alpaka::
accMatchesTags<AccType, alpaka::TagGpuCudaRt, alpaka::TagGpuHipRt, alpaka::TagGpuSyclIntel>)
if constexpr(alpaka::isMultiThreadAcc<AccType>)
{
return alpaka::WorkDivMembers{
Vec::all(static_cast<alpaka::Idx<AccType>>(dotGridBlockExtent)),
Expand Down
Loading