[STF] Extract standalone __places project from __stf/places

CMakePresets.json

@@ -82,6 +82,7 @@
         "cudax_ENABLE_HEADER_TESTING": true,
         "cudax_ENABLE_TESTING": true,
         "cudax_ENABLE_EXAMPLES": true,
+        "cudax_ENABLE_PLACES": true,
         "cudax_ENABLE_CUDASTF": true,
         "cudax_ENABLE_CUDASTF_BOUNDSCHECK": false,
         "cudax_ENABLE_CUDASTF_CODE_GENERATION": true,
@@ -353,6 +354,7 @@
         "cudax_ENABLE_HEADER_TESTING": true,
         "cudax_ENABLE_TESTING": true,
         "cudax_ENABLE_EXAMPLES": true,
+        "cudax_ENABLE_PLACES": true,
         "cudax_ENABLE_CUDASTF": true,
         "cudax_ENABLE_CUDASTF_BOUNDSCHECK": false,
         "cudax_ENABLE_CUDASTF_CODE_GENERATION": true,

c/experimental/stf/src/stf.cu

@@ -10,7 +10,7 @@
 
 #include <cccl/c/experimental/stf/stf.h>
 // #include <cccl/c/parallel/include/cccl/c/extern_c.h>
-#include <cuda/experimental/__stf/places/places.cuh>
+#include <cuda/experimental/places.cuh>
 #include <cuda/experimental/stf.cuh>
 
 #include <cstddef>

cudax/CMakeLists.txt

@@ -13,6 +13,7 @@ option(
 )
 option(cudax_ENABLE_TESTING "Build CUDA Experimental's tests." ON)
 option(cudax_ENABLE_EXAMPLES "Build CUDA Experimental's examples." ON)
+option(cudax_ENABLE_PLACES "Enable standalone Places subproject" ON)
 option(cudax_ENABLE_CUDASTF "Enable CUDASTF subproject" ON)
 option(
   cudax_ENABLE_CUDASTF_CODE_GENERATION
@@ -69,6 +70,9 @@ if (
 endif()
 
 include(cmake/cudaxBuildCompilerTargets.cmake)
+if (cudax_ENABLE_PLACES)
+  include(cmake/cudaxPlacesConfigureTarget.cmake)
+endif()
 if (cudax_ENABLE_CUDASTF)
   include(cmake/cudaxSTFConfigureTarget.cmake)
 endif()

cudax/cmake/cudaxHeaderTesting.cmake

@@ -24,6 +24,9 @@ function(cudax_add_header_test label definitions)
       # cuFile headers are compiled separately:
       "cuda/experimental/cufile.cuh"
       "cuda/experimental/__cufile/*"
+      # Places headers are compiled separately:
+      "cuda/experimental/places.cuh"
+      "cuda/experimental/__places/*"
       # STF headers are compiled separately:
       "cuda/experimental/stf.cuh"
       "cuda/experimental/__stf/*"
@@ -45,6 +48,28 @@ function(cudax_add_header_test label definitions)
     target_link_libraries(${headertest_target} PUBLIC cudax.compiler_interface)
   endif()
 
+  # FIXME: Enable MSVC
+  if (cudax_ENABLE_PLACES AND NOT "MSVC" STREQUAL "${CMAKE_CXX_COMPILER_ID}")
+    ##################
+    # Places headers #
+    set(headertest_target cudax.headers.${label}.places)
+    cccl_generate_header_tests(
+      ${headertest_target}
+      cudax/include
+      GLOBS #
+        "cuda/experimental/places.cuh"
+        "cuda/experimental/__places/*.cuh"
+      HEADER_TEMPLATE "${cudax_SOURCE_DIR}/cmake/header_test.in.cu"
+    )
+    target_link_libraries(${headertest_target} PUBLIC cudax.compiler_interface)
+    target_compile_options(
+      ${headertest_target}
+      PRIVATE
+        $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>:--extended-lambda>
+        $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>:--expt-relaxed-constexpr>
+    )
+  endif()
+
   # FIXME: Enable MSVC
   if (cudax_ENABLE_CUDASTF AND NOT "MSVC" STREQUAL "${CMAKE_CXX_COMPILER_ID}")
     ###############

cudax/cmake/cudaxPlacesConfigureTarget.cmake

@@ -0,0 +1,23 @@
+# Configures a target for the Places framework.
+function(cudax_places_configure_target target_name)
+  target_link_libraries(
+    ${target_name}
+    PRIVATE #
+      CUDA::cudart_static
+      CUDA::cuda_driver
+  )
+
+  target_compile_options(
+    ${target_name}
+    PRIVATE
+      $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>:--extended-lambda>
+      $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>:--expt-relaxed-constexpr>
+  )
+
+  set_target_properties(
+    ${target_name}
+    PROPERTIES #
+      CUDA_RUNTIME_LIBRARY Static
+      CUDA_SEPARABLE_COMPILATION ON
+  )
+endfunction()

cudax/cmake/places_header_unittest.in.cu

@@ -0,0 +1,9 @@
+// This file is autogenerated by configuring places_header_unittest.in.cu.
+
+// clang-format off
+#define UNITTESTED_FILE "@source@"
+
+#include <cuda/experimental/__stf/utility/unittest.cuh>
+
+#include <@source@>
+//clang-format on

cudax/include/cuda/experimental/__stf/places/data_place_impl.cuh → cudax/include/cuda/experimental/__places/data_place_impl.cuh

@@ -28,7 +28,7 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cuda/experimental/__stf/places/data_place_interface.cuh>
+#include <cuda/experimental/__places/data_place_interface.cuh>
 #include <cuda/experimental/__stf/utility/cuda_safe_call.cuh>
 #include <cuda/experimental/__stf/utility/scope_guard.cuh>
 

cudax/include/cuda/experimental/__stf/places/exec/cuda_stream.cuh → cudax/include/cuda/experimental/__places/exec/cuda_stream.cuh

@@ -25,7 +25,7 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cuda/experimental/__stf/places/places.cuh>
+#include <cuda/experimental/__places/places.cuh>
 
 namespace cuda::experimental::stf
 {

cudax/include/cuda/experimental/__stf/places/exec/green_context.cuh → cudax/include/cuda/experimental/__places/exec/green_context.cuh

@@ -25,9 +25,9 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cuda/experimental/__stf/places/data_place_interface.cuh>
-#include <cuda/experimental/__stf/places/exec/green_ctx_view.cuh>
-#include <cuda/experimental/__stf/places/places.cuh>
+#include <cuda/experimental/__places/data_place_interface.cuh>
+#include <cuda/experimental/__places/exec/green_ctx_view.cuh>
+#include <cuda/experimental/__places/places.cuh>
 #include <cuda/experimental/__stf/utility/hash.cuh>
 
 // Used only for unit tests, not in the actual implementation

cudax/include/cuda/experimental/__stf/places/exec/green_ctx_view.cuh → cudax/include/cuda/experimental/__places/exec/green_ctx_view.cuh

@@ -25,7 +25,7 @@
  * @brief Implementation of green context views
  */
 
-#include <cuda/experimental/__stf/places/stream_pool.cuh>
+#include <cuda/experimental/__places/stream_pool.cuh>
 #include <cuda/experimental/__stf/utility/hash.cuh>
 
 #if _CCCL_CTK_AT_LEAST(12, 4)

cudax/include/cuda/experimental/__stf/internal/machine.cuh → cudax/include/cuda/experimental/__places/machine.cuh

@@ -1,10 +1,10 @@
 //===----------------------------------------------------------------------===//
 //
-// Part of CUDASTF in CUDA C++ Core Libraries,
+// Part of CUDA Experimental in CUDA C++ Core Libraries,
 // 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
-// SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES.
+// SPDX-FileCopyrightText: Copyright (c) 2022-2026 NVIDIA CORPORATION & AFFILIATES.
 //
 //===----------------------------------------------------------------------===//
 

cudax/include/cuda/experimental/__stf/places/place_partition.cuh → cudax/include/cuda/experimental/__places/place_partition.cuh

@@ -25,10 +25,10 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/exec/cuda_stream.cuh>
+#include <cuda/experimental/__places/exec/green_context.cuh>
+#include <cuda/experimental/__places/places.cuh>
 #include <cuda/experimental/__stf/internal/async_resources_handle.cuh>
-#include <cuda/experimental/__stf/places/exec/cuda_stream.cuh>
-#include <cuda/experimental/__stf/places/exec/green_context.cuh>
-#include <cuda/experimental/__stf/places/places.cuh>
 
 namespace cuda::experimental::stf
 {

cudax/include/cuda/experimental/__stf/places/places.cuh → cudax/include/cuda/experimental/__places/places.cuh

@@ -27,9 +27,8 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cuda/experimental/__stf/internal/interpreted_execution_policy.cuh>
-#include <cuda/experimental/__stf/places/data_place_impl.cuh>
-#include <cuda/experimental/__stf/places/exec/green_ctx_view.cuh>
+#include <cuda/experimental/__places/data_place_impl.cuh>
+#include <cuda/experimental/__places/exec/green_ctx_view.cuh>
 #include <cuda/experimental/__stf/utility/core.cuh>
 
 #include <typeinfo>
@@ -40,7 +39,6 @@
 #endif
 #include <cuda/experimental/__stf/utility/cuda_safe_call.cuh>
 #include <cuda/experimental/__stf/utility/dimensions.cuh>
-#include <cuda/experimental/__stf/utility/occupancy.cuh>
 #include <cuda/experimental/__stf/utility/scope_guard.cuh>
 
 // Sync only will not move data....
@@ -893,9 +891,6 @@ private:
   exec_place prev_; // Previous state to restore
 };
 
-// Deprecated: Use exec_place_scope instead
-using exec_place_guard = exec_place_scope;
-
 inline exec_place_scope exec_place::activate(size_t idx) const
 {
   return exec_place_scope(*this, idx);

cudax/include/cuda/experimental/__stf/places/stream_pool.cuh → cudax/include/cuda/experimental/__places/stream_pool.cuh

@@ -160,7 +160,7 @@ public:
 
   /**
    * @brief Get the next stream in the pool; when a slot is empty, activate the place (RAII guard) and call
-   * place.create_stream(). Defined in places.cuh so the pool can use exec_place_guard and exec_place::create_stream().
+   * place.create_stream(). Defined in places.cuh so the pool can use exec_place_scope and exec_place::create_stream().
    */
   decorated_stream next(const exec_place& place);
 

cudax/include/cuda/experimental/__stf/allocators/block_allocator.cuh

@@ -25,8 +25,8 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/places.cuh>
 #include <cuda/experimental/__stf/internal/async_prereq.cuh>
-#include <cuda/experimental/__stf/places/places.cuh>
 
 #include <mutex>
 

cudax/include/cuda/experimental/__stf/internal/async_resources_handle.cuh

@@ -25,9 +25,9 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/exec/green_context.cuh>
 #include <cuda/experimental/__stf/internal/exec_affinity.cuh>
 #include <cuda/experimental/__stf/internal/executable_graph_cache.cuh>
-#include <cuda/experimental/__stf/places/exec/green_context.cuh>
 #include <cuda/experimental/__stf/utility/core.cuh>
 #include <cuda/experimental/__stf/utility/cuda_safe_call.cuh>
 #include <cuda/experimental/__stf/utility/hash.cuh> // for ::std::hash<::std::pair<::std::ptrdiff_t, ::std::ptrdiff_t>>

cudax/include/cuda/experimental/__stf/internal/backend_ctx.cuh

@@ -27,12 +27,12 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/machine.cuh>
 #include <cuda/experimental/__stf/allocators/block_allocator.cuh>
 #include <cuda/experimental/__stf/internal/async_resources_handle.cuh>
 #include <cuda/experimental/__stf/internal/ctx_resource.cuh>
 #include <cuda/experimental/__stf/internal/execution_policy.cuh> // backend_ctx<T>::launch() uses execution_policy
 #include <cuda/experimental/__stf/internal/interpreted_execution_policy.cuh>
-#include <cuda/experimental/__stf/internal/machine.cuh> // backend_ctx_untyped::impl usese machine
 #include <cuda/experimental/__stf/internal/reorderer.cuh> // backend_ctx_untyped::impl uses reorderer
 #include <cuda/experimental/__stf/internal/repeat.cuh>
 #include <cuda/experimental/__stf/internal/scheduler.cuh> // backend_ctx_untyped::impl uses scheduler

cudax/include/cuda/experimental/__stf/internal/context.cuh

@@ -17,6 +17,7 @@
 
 #include <cuda/std/__exception/exception_macros.h>
 
+#include <cuda/experimental/__places/exec/cuda_stream.cuh>
 #include <cuda/experimental/__stf/allocators/adapters.cuh>
 #include <cuda/experimental/__stf/allocators/buddy_allocator.cuh>
 #include <cuda/experimental/__stf/allocators/cached_allocator.cuh>
@@ -28,7 +29,6 @@
 #include <cuda/experimental/__stf/internal/scalar_interface.cuh>
 #include <cuda/experimental/__stf/internal/task_dep.cuh>
 #include <cuda/experimental/__stf/internal/void_interface.cuh>
-#include <cuda/experimental/__stf/places/exec/cuda_stream.cuh>
 #include <cuda/experimental/__stf/stream/stream_ctx.cuh>
 
 #include <map>

cudax/include/cuda/experimental/__stf/internal/interpreted_execution_policy_impl.cuh

@@ -31,8 +31,8 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/places.cuh>
 #include <cuda/experimental/__stf/internal/interpreted_execution_policy.cuh>
-#include <cuda/experimental/__stf/places/places.cuh>
 #include <cuda/experimental/__stf/utility/occupancy.cuh>
 
 namespace cuda::experimental::stf

cudax/include/cuda/experimental/__stf/internal/loop_dispatch.cuh

@@ -26,7 +26,7 @@
 #  pragma system_header
 #endif // no system header
 
-#include <cuda/experimental/__stf/places/place_partition.cuh>
+#include <cuda/experimental/__places/place_partition.cuh>
 
 #include <thread>
 

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh

@@ -28,6 +28,7 @@
 #include <cuda/experimental/__stf/internal/task_dep.cuh>
 #include <cuda/experimental/__stf/internal/task_statistics.cuh>
 #include <cuda/experimental/__stf/stream/internal/event_types.cuh>
+#include <cuda/experimental/__stf/utility/occupancy.cuh>
 
 namespace cuda::experimental::stf
 {

cudax/include/cuda/experimental/__stf/internal/reduction_base.cuh

@@ -20,8 +20,8 @@
 #  pragma system_header
 #endif // no system header
 
+#include <cuda/experimental/__places/places.cuh>
 #include <cuda/experimental/__stf/internal/async_prereq.cuh>
-#include <cuda/experimental/__stf/places/places.cuh>
 
 namespace cuda::experimental::stf
 {