[compiler-rt] Adding Separate OpenMP Offloading Backend to `libcxx/include/__algorithm/pstl_backends` (PR #66968)

[Louis Dionne via llvm-commits]

================
@@ -0,0 +1,74 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef _LIBCPP___ALGORITHM_PSTL_BACKENDS_OPENMP_BACKEND_FOR_EACH_H
+#define _LIBCPP___ALGORITHM_PSTL_BACKENDS_OPENMP_BACKEND_FOR_EACH_H
+
+#include <__algorithm/for_each.h>
+#include <__algorithm/pstl_backends/cpu_backends/backend.h>
+#include <__algorithm/pstl_backends/openmp/backend.h>
+#include <__config>
+#include <__type_traits/is_execution_policy.h>
+#include <__utility/empty.h>
+#include <optional>
+
+#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
+#  pragma GCC system_header
+#endif
+
+#if !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
+
+_LIBCPP_BEGIN_NAMESPACE_STD
+
+namespace __par_backend {
+inline namespace __omp_gpu_backend {
+
+template <class _Tp, class _DifferenceType, class _Function>
+_LIBCPP_HIDE_FROM_ABI optional<__empty>
+__omp_parallel_for_simd(_Tp* __first, _DifferenceType __n, _Function __f) noexcept {
+  __omp_gpu_backend::__omp_map_to(__first, __n);
+#  pragma omp target teams distribute parallel for simd
+  for (_DifferenceType __i = 0; __i < __n; ++__i)
+    __f(*(__first + __i));
+  __omp_gpu_backend::__omp_map_from(__first, __n);
+  return __empty{};
+}
+
+// Extracting the underlying pointer
+
+template <class _ForwardIterator, class _DifferenceType, class _Function>
+_LIBCPP_HIDE_FROM_ABI optional<__empty>
+__parallel_for_simd_1(_ForwardIterator __first, _DifferenceType __n, _Function __f) noexcept {
+  return __omp_gpu_backend::__omp_parallel_for_simd(__omp_gpu_backend::__omp_extract_base_ptr(__first), __n, __f);
+}
+
+} // namespace __omp_gpu_backend
+} // namespace __par_backend
+
+template <class _ExecutionPolicy, class _ForwardIterator, class _Functor>
+_LIBCPP_HIDE_FROM_ABI optional<__empty>
+__pstl_for_each(__omp_backend_tag, _ForwardIterator __first, _ForwardIterator __last, _Functor __func) {
+  // It is only safe to execute for_each on the GPU, it the execution policy is
+  // parallel unsequenced, as it is the only execution policy prohibiting throwing
+  // exceptions and allowing SIMD instructions
----------------
ldionne wrote:

This comment is not technically correct, since exceptions can be thrown from user code even when `par_unseq`  is used.

And `unseq` also allows for SIMD. So I think we need to explain that we don't have sequencing between the teams (thanks @jdoerfert !).

Also, I think it might be worth promoting this comment to `openmp.h` since it is not specific for `for_each`.

https://github.com/llvm/llvm-project/pull/66968