[SYCL] Add support for sorting using sub-group

sycl/include/sycl/detail/group_sort_impl.hpp

@@ -10,8 +10,9 @@
 
 #pragma once
 
-#if __cplusplus >= 201703L
 #include <sycl/detail/helpers.hpp>
+#include <sycl/group_barrier.hpp>
+#include <sycl/multi_ptr.hpp>
 
 #ifdef __SYCL_DEVICE_ONLY__
 
@@ -204,14 +205,13 @@ template <typename Group, typename Iter, typename Compare>
 void merge_sort(Group group, Iter first, const std::size_t n, Compare comp,
                 std::byte *scratch) {
   using T = typename GetValueType<Iter>::type;
-  auto id = sycl::detail::Builder::getNDItem<Group::dimensions>();
-  const std::size_t idx = id.get_local_linear_id();
+  const std::size_t idx = group.get_local_linear_id();
   const std::size_t local = group.get_local_range().size();
   const std::size_t chunk = (n - 1) / local + 1;
 
   // we need to sort within work item first
   bubble_sort(first, idx * chunk, sycl::min((idx + 1) * chunk, n), comp);
-  id.barrier();
+  sycl::group_barrier(group);
 
   T *temp = reinterpret_cast<T *>(scratch);
   bool data_in_temp = false;
@@ -231,7 +231,7 @@ void merge_sort(Group group, Iter first, const std::size_t n, Compare comp,
       merge(offset, temp, first, start_1, end_1, end_2, start_1, comp, chunk,
             /*is_first*/ false);
     }
-    id.barrier();
+    sycl::group_barrier(group);
 
     data_in_temp = !data_in_temp;
     sorted_size *= 2;
@@ -246,12 +246,11 @@ void merge_sort(Group group, Iter first, const std::size_t n, Compare comp,
         first[idx * chunk + i] = temp[idx * chunk + i];
       }
     }
-    id.barrier();
+    sycl::group_barrier(group);
   }
 }
 
 } // namespace detail
 } // __SYCL_INLINE_VER_NAMESPACE(_V1)
 } // namespace sycl
 #endif
-#endif // __cplusplus >=201703L

sycl/include/sycl/ext/oneapi/experimental/group_helpers_sorters.hpp

@@ -62,8 +62,7 @@ template <typename Compare = std::less<>> class default_sorter {
 #ifdef __SYCL_DEVICE_ONLY__
     auto range_size = g.get_local_range().size();
     if (scratch_size >= memory_required<T>(Group::fence_scope, range_size)) {
-      auto id = sycl::detail::Builder::getNDItem<Group::dimensions>();
-      std::size_t local_id = id.get_local_linear_id();
+      std::size_t local_id = g.get_local_linear_id();
       T *temp = reinterpret_cast<T *>(scratch);
       ::new (temp + local_id) T(val);
       sycl::detail::merge_sort(g, temp, range_size, comp,

sycl/include/sycl/group_algorithm.hpp

@@ -18,6 +18,7 @@
 #include <sycl/ext/oneapi/functional.hpp>
 #include <sycl/functional.hpp>
 #include <sycl/group.hpp>
+#include <sycl/group_barrier.hpp>
 #include <sycl/known_identity.hpp>
 #include <sycl/nd_item.hpp>
 #include <sycl/sub_group.hpp>
@@ -1006,36 +1007,5 @@ joint_inclusive_scan(Group g, InPtr first, InPtr last, OutPtr result,
   return joint_inclusive_scan(g, first, last, result, binary_op, init);
 }
 
-namespace detail {
-template <typename G> struct group_barrier_scope {};
-template <> struct group_barrier_scope<sycl::sub_group> {
-  constexpr static auto Scope = __spv::Scope::Subgroup;
-};
-template <int D> struct group_barrier_scope<sycl::group<D>> {
-  constexpr static auto Scope = __spv::Scope::Workgroup;
-};
-} // namespace detail
-
-template <typename Group>
-typename std::enable_if<is_group_v<Group>>::type
-group_barrier(Group, memory_scope FenceScope = Group::fence_scope) {
-  (void)FenceScope;
-#ifdef __SYCL_DEVICE_ONLY__
-  // Per SYCL spec, group_barrier must perform both control barrier and memory
-  // fence operations. All work-items execute a release fence prior to
-  // barrier and acquire fence afterwards. The rest of semantics flags specify
-  // which type of memory this behavior is applied to.
-  __spirv_ControlBarrier(detail::group_barrier_scope<Group>::Scope,
-                         sycl::detail::spirv::getScope(FenceScope),
-                         __spv::MemorySemanticsMask::SequentiallyConsistent |
-                             __spv::MemorySemanticsMask::SubgroupMemory |
-                             __spv::MemorySemanticsMask::WorkgroupMemory |
-                             __spv::MemorySemanticsMask::CrossWorkgroupMemory);
-#else
-  throw sycl::runtime_error("Barriers are not supported on host device",
-                            PI_ERROR_INVALID_DEVICE);
-#endif
-}
-
 } // __SYCL_INLINE_VER_NAMESPACE(_V1)
 } // namespace sycl

sycl/include/sycl/group_barrier.hpp

@@ -0,0 +1,54 @@
+
+//==------------------------- group_barrier.hpp ----------------------------==//
+//
+// 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
+//
+//===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include <CL/__spirv/spirv_ops.hpp>
+#include <CL/__spirv/spirv_types.hpp>
+#include <CL/__spirv/spirv_vars.hpp>
+#include <sycl/detail/spirv.hpp>
+#include <sycl/detail/type_traits.hpp>
+#include <sycl/group.hpp>
+
+namespace sycl {
+__SYCL_INLINE_VER_NAMESPACE(_V1) {
+
+namespace detail {
+template <typename G> struct group_barrier_scope {};
+template <> struct group_barrier_scope<sycl::sub_group> {
+  constexpr static auto Scope = __spv::Scope::Subgroup;
+};
+template <int D> struct group_barrier_scope<sycl::group<D>> {
+  constexpr static auto Scope = __spv::Scope::Workgroup;
+};
+} // namespace detail
+
+template <typename Group>
+typename std::enable_if<is_group_v<Group>>::type
+group_barrier(Group, memory_scope FenceScope = Group::fence_scope) {
+  (void)FenceScope;
+#ifdef __SYCL_DEVICE_ONLY__
+  // Per SYCL spec, group_barrier must perform both control barrier and memory
+  // fence operations. All work-items execute a release fence prior to
+  // barrier and acquire fence afterwards. The rest of semantics flags specify
+  // which type of memory this behavior is applied to.
+  __spirv_ControlBarrier(detail::group_barrier_scope<Group>::Scope,
+                         sycl::detail::spirv::getScope(FenceScope),
+                         __spv::MemorySemanticsMask::SequentiallyConsistent |
+                             __spv::MemorySemanticsMask::SubgroupMemory |
+                             __spv::MemorySemanticsMask::WorkgroupMemory |
+                             __spv::MemorySemanticsMask::CrossWorkgroupMemory);
+#else
+  throw sycl::runtime_error("Barriers are not supported on host device",
+                            PI_ERROR_INVALID_DEVICE);
+#endif
+}
+
+} // __SYCL_INLINE_VER_NAMESPACE(_V1)
+} // namespace sycl