[SYCL] Add support for radix_sorter #7639
Conversation
Co-authored-by: Andrei Fedorov <andrey.fedorov@intel.com>
|
Test is here: intel/llvm-test-suite#1435 |
|
/verify with intel/llvm-test-suite#1435 |
|
/verify with intel/llvm-test-suite#1435 |
This reverts commit ce8b972.
|
/verify with intel/llvm-test-suite#1435 |
1 similar comment
|
/verify with intel/llvm-test-suite#1435 |
|
/verify with intel/llvm-test-suite#1435 |
|
@v-klochkov Could you please take a look? |
sycl/include/sycl/ext/oneapi/experimental/group_helpers_sorters.hpp
Outdated
Show resolved
Hide resolved
sycl/include/sycl/ext/oneapi/experimental/group_helpers_sorters.hpp
Outdated
Show resolved
Hide resolved
v-klochkov
dismissed
their stale review
After additional commit, need to get answer to a review comment.
|
/verify with intel/llvm-test-suite#1435 |
|
/verify with intel/llvm-test-suite#1435 |
1 similar comment
|
/verify with intel/llvm-test-suite#1435 |
|
The failures are unrelated. |
| uint32_t new_offset_idx = private_scan_memory[bucket_val]++ + | ||
| scan_memory[bucket_val * wgsize + idx]; | ||
| if (val_idx < n) { | ||
| keys_output[new_offset_idx] = keys_input[val_idx]; |
There was a problem hiding this comment.
It's better to use placement new for 1st initialisation of this memory: https://en.cppreference.com/w/cpp/language/new#Placement_new
There was a problem hiding this comment.
Could you please clarify why it's required here since we have trivially copyable types?
There was a problem hiding this comment.
Sorry, we use just arithmetic types for radix. It's ok if we work with trivially copyable types
|
/verify with intel/llvm-test-suite#1435 |
|
@romanovvlad, post-commit issue on windows https://github.com/intel/llvm/actions/runs/3687647938/jobs/6241495307 |
This seemed familiar, so I prepared a PR: #7768 |
Co-authored-by: Andrei Fedorov andrey.fedorov@intel.com