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

[Louis Dionne via libcxx-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_TRANSFORM_H
+#define _LIBCPP___ALGORITHM_PSTL_BACKENDS_OPENMP_BACKEND_TRANSFORM_H
+
+#include <__algorithm/pstl_backends/cpu_backends/backend.h>
+#include <__algorithm/pstl_backends/openmp/backend.h>
+#include <__algorithm/transform.h>
+#include <__config>
+#include <__iterator/concepts.h>
+#include <__type_traits/is_execution_policy.h>
+
+#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
+
+template <class _ExecutionPolicy, class _ForwardIterator, class _ForwardOutIterator, class _UnaryOperation>
+_LIBCPP_HIDE_FROM_ABI _ForwardOutIterator __pstl_transform(
+    __omp_backend_tag,
+    _ForwardIterator __first,
+    _ForwardIterator __last,
+    _ForwardOutIterator __result,
+    _UnaryOperation __op) {
+  // The interface for the function switched between C++17 and C++20
+  if constexpr (__is_unsequenced_execution_policy_v<_ExecutionPolicy> &&
+                __has_random_access_iterator_category_or_concept<_ForwardIterator>::value &&
----------------
ldionne wrote:

For now, I would stick to `is_contiguous` for the input iterators as well. However, as we discussed, we can (and should) introduce a notion of a "self-contained" iterator in a future patch. Once we have that, `iota_view` and others can specialize that libc++-internal trait and we can run those on the GPU as well.

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