Sergey Kopienko

Results 60 issues of Sergey Kopienko

https://github.com/oneapi-src/oneDPL/blob/a60fa807b1baaad492d86b331548320bf3a15f09/test/support/sycl_sanity.h#L34 ```C++ template using new_kernel_name = unique_kernel_name; ```

https://github.com/oneapi-src/oneDPL/blob/a6d3f7f2421ae5a2cd445b6894fa1a2297b0c75c/src/pstl_offload.cpp#L115 This function declared as `noexcept` but call function which is not declared as `noexcept` I mean https://github.com/oneapi-src/oneDPL/blob/a6d3f7f2421ae5a2cd445b6894fa1a2297b0c75c/src/pstl_offload.cpp#L70

https://github.com/oneapi-src/oneDPL/blob/180f18ad25fbc39fa13bad43f1df7d54ee8f5609/include/oneapi/dpl/internal/exclusive_scan_by_segment_impl.h#L126 The all computation must be done on a device, we must not getting the host access. Yes, it is a different(from this PR) story... But, lets don't forget to...

https://github.com/oneapi-src/oneDPL/blob/fec85a0169b3c62ce8e67663bf8630385f37e7df/include/oneapi/dpl/pstl/execution_impl.h#L99 The next two aliases looks like not used: ```C++ template using __collector_t = typename __internal::__policy_traits::__collector_type; template using __allow_vector = typename __internal::__policy_traits::__allow_vector; ```

In this PR we have 3 big changes: 1) In the file `test/support/test_config.h` we add inclusion of some files from standard library to have correct definitions of used defines: ```C++...

In this PR we simplify constructor calls in oneDPL code: - before: ```C++ if constexpr (std::is_trivial_v) __target = std::forward(__source); else ::new (std::addressof(__target)) _TargetValueType(std::forward(__source)); ``` - after: ```C++ ::new (std::addressof(__target)) _TargetValueType(std::forward(__source));...

This PR is only prototype for descussions. ## The goal of this PR The goal of this PR - redesign our host backends (`SERIAL`, `TBB`, `OMP`) : - to use...

In this PR we redefine `timing_t` as `std::chrono::milliseconds` and remove `count()` function call at all, so we always operates by value with physical sense. Of course it's not mandatory improvement,...

During the review of https://github.com/oneapi-src/oneDPL/pull/1511 @rarutyun and @kboyarinov found some incorrect implementations in our code. To clarify this moment I extend tests for the set of `oneDPL` algorithms: - `uninitialized_default_construct`;...

The goal of this PR - rename template params and function params of `transform_reduce` staff to be closer to https://en.cppreference.com/w/cpp/algorithm/transform_reduce template param type names and add semantic info to template...