Skip to content
This repository was archived by the owner on Mar 28, 2023. It is now read-only.

Update more tests to account for optionality of double type #1242

Merged
merged 6 commits into from
Sep 13, 2022
Merged

Update more tests to account for optionality of double type #1242

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
bba398d
Update more tests to account for optionality of double type
AlexeySachkov Sep 9, 2022
1d84cb6
A couple of fixes
AlexeySachkov Sep 9, 2022
4424a47
Revert spec constant changes
AlexeySachkov Sep 12, 2022
4514445
Merge remote-tracking branch 'origin/intel' into private/asachkov/upd…
AlexeySachkov Sep 12, 2022
b68ac93
Update copy.cpp
AlexeySachkov Sep 12, 2022
a5e1dc5
Apply code review comments
AlexeySachkov Sep 13, 2022
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
8 changes: 5 additions & 3 deletions SYCL/GroupAlgorithm/SYCL2020/sort.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -I . -o %t.out
// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -fsycl-device-code-split=per_kernel %s -I . -o %t.out
// RUN: %CPU_RUN_PLACEHOLDER %t.out
// RUN: %GPU_RUN_PLACEHOLDER %t.out
// RUN: %ACC_RUN_PLACEHOLDER %t.out
Expand Down Expand Up @@ -357,8 +357,10 @@ int main(int argc, char *argv[]) {
test_sort_by_type<std::int32_t>(q, sizes[i]);
test_sort_by_type<std::uint32_t>(q, sizes[i]);
test_sort_by_type<float>(q, sizes[i]);
test_sort_by_type<sycl::half>(q, sizes[i]);
test_sort_by_type<double>(q, sizes[i]);
if (q.get_device().has(sycl::aspect::fp16))
test_sort_by_type<sycl::half>(q, sizes[i]);
if (q.get_device().has(sycl::aspect::fp64))
test_sort_by_type<double>(q, sizes[i]);
test_sort_by_type<std::size_t>(q, sizes[i]);

test_custom_type(q, sizes[i]);
Expand Down
5 changes: 1 addition & 4 deletions SYCL/SubGroup/info.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,3 @@
// See https://github.com/intel/llvm/issues/2922 for more info
// UNSUPPORTED: cuda || hip

// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
// RUN: %CPU_RUN_PLACEHOLDER %t.out
// RUN: %GPU_RUN_PLACEHOLDER %t.out
Expand Down Expand Up @@ -40,7 +37,7 @@ int main() {
auto Kernel = KB.get_kernel(KernelID);
range<2> GlobalRange{50, 40};

buffer<double, 2> ABuf{GlobalRange}, BBuf{GlobalRange}, CBuf{GlobalRange};
buffer<float, 2> ABuf{GlobalRange}, BBuf{GlobalRange}, CBuf{GlobalRange};
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm also touching it in #1246, but I prefer your change. I'll exclude the file from my PR.


Queue.submit([&](sycl::handler &cgh) {
auto A = ABuf.get_access<access::mode::read_write>(cgh);
Expand Down
10 changes: 6 additions & 4 deletions SYCL/USM/copy.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
//
//===----------------------------------------------------------------------===//
//
// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t1.out
// RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -fsycl-device-code-split=per_kernel %s -o %t1.out
// RUN: %CPU_RUN_PLACEHOLDER %t1.out
// RUN: %GPU_RUN_PLACEHOLDER %t1.out
// RUN: %ACC_RUN_PLACEHOLDER %t1.out
Expand All @@ -27,12 +27,11 @@ struct test_struct {
long long d;
half e;
float f;
double g;
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could we have another struct with the double and give it the same treatment as just double?

Copy link

@aelovikov-intel aelovikov-intel Sep 13, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this can be addressed post-commit in another PR to fix the currently failing tests in CI asap.

@intel/llvm-gatekeepers would you agree with that? If so, please merge it in - the tests modified here are passing in pre-commit CI and the failures are on other tests and thus unrelated.

};

bool operator==(const test_struct &lhs, const test_struct &rhs) {
return lhs.a == rhs.a && lhs.b == rhs.b && lhs.c == rhs.c && lhs.d == rhs.d &&
lhs.e == rhs.e && lhs.f == rhs.f && lhs.g == rhs.g;
lhs.e == rhs.e && lhs.f == rhs.f;
}

template <typename T> T *regular(queue q, alloc kind) {
Expand All @@ -44,6 +43,9 @@ template <typename T> T *aligned(queue q, alloc kind) {
}

template <typename T> void test(queue q, T val, T *src, T *dst, bool dev_dst) {
if (std::is_same_v<T, double> && !q.get_device().has(aspect::fp64))
return;

q.fill(src, val, N).wait();

// Use queue::copy for the first half and handler::copy for the second
Expand Down Expand Up @@ -87,7 +89,7 @@ int main() {
queue q;
auto dev = q.get_device();

test_struct test_obj{4, 42, 424, 4242, 4.2f, 4.242f, 4.24242};
test_struct test_obj{4, 42, 424, 4242, 4.2f, 4.242f};

if (dev.has(aspect::usm_host_allocations)) {
runTests<short>(q, 4, alloc::host, alloc::host);
Expand Down