[INFRA] Import NVIDIA/CCCL upstream as optimization reference library

CCCL (CUDA C++ Core Libraries) provides:
- CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk)
- Thrust: high-level parallel algorithms (transform_reduce, sort, scan)
- libcudacxx: CUDA C++ standard library (atomics, barriers, memory)
- cudax: experimental features (memory resources, allocators)
- Tuning policies: per-SM hardware-specific algorithm parameters

Competition optimization vectors mapped to CCCL:
- Output TPS (83% weight): warp_reduce, block_reduce, device_topk
- Input TPS (14% weight): device_scan, block_load, prefetch
- Cache TPS (3% weight): prefix caching strategy patterns
- Memory (0.9 util): pooled/cached/buddy allocators

Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only)
License: Apache-2.0

cccl_upstream/cudax/test/execution/policies/get_execution_policy.cu

@@ -0,0 +1,142 @@
+//===----------------------------------------------------------------------===//
+//
+// 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) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cuda/std/type_traits>
+
+#include <cuda/experimental/execution.cuh>
+
+#include <testing.cuh>
+
+namespace execution = cuda::experimental::execution;
+
+struct with_get_execution_policy_const_lvalue
+{
+  execution::any_execution_policy pol_ = execution::seq;
+
+  const execution::any_execution_policy& get_execution_policy() const noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Can call get_execution_policy on a type with a get_execution_policy method that returns a const lvalue",
+         "[execution][policies]")
+{
+  with_get_execution_policy_const_lvalue val{};
+  auto&& res = cuda::experimental::execution::get_execution_policy(val);
+  STATIC_REQUIRE(cuda::std::is_same_v<decltype(res), execution::any_execution_policy&&>);
+  CHECK(val.pol_ == res);
+}
+
+struct with_get_execution_policy_rvalue
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy get_execution_policy() const noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Can call get_execution_policy on a type with a get_execution_policy method returns an rvalue",
+         "[execution][policies]")
+{
+  with_get_execution_policy_rvalue val{};
+  auto&& res = cuda::experimental::execution::get_execution_policy(val);
+  STATIC_REQUIRE(cuda::std::is_same_v<decltype(res), execution::any_execution_policy&&>);
+  CHECK(val.pol_ == res);
+}
+
+struct with_get_execution_policy_non_const
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy get_execution_policy() noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Cannot call get_execution_policy on a type with a non-const get_execution_policy method",
+         "[execution][policies]")
+{
+  STATIC_REQUIRE(!::cuda::std::is_invocable_v<cuda::experimental::execution::get_execution_policy_t,
+                                              const with_get_execution_policy_non_const&>);
+}
+
+struct env_with_query_const_ref
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy query(cuda::experimental::execution::get_execution_policy_t) const noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Can call get_execution_policy on an env with a get_execution_policy query that returns a const lvalue",
+         "[execution][policies]")
+{
+  env_with_query_const_ref val{};
+  auto&& res = cuda::experimental::execution::get_execution_policy(val);
+  STATIC_REQUIRE(cuda::std::is_same_v<decltype(res), execution::any_execution_policy&&>);
+  CHECK(val.pol_ == res);
+}
+
+struct env_with_query_rvalue
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy query(cuda::experimental::execution::get_execution_policy_t) const noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Can call get_execution_policy on an env with a get_execution_policy query that returns an rvalue",
+         "[execution][policies]")
+{
+  env_with_query_rvalue val{};
+  auto&& res = cuda::experimental::execution::get_execution_policy(val);
+  STATIC_REQUIRE(cuda::std::is_same_v<decltype(res), execution::any_execution_policy&&>);
+  CHECK(val.pol_ == res);
+}
+
+struct env_with_query_non_const
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy query(cuda::experimental::execution::get_execution_policy_t) noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Cannot call get_execution_policy on an env with a non-const query", "[execution][policies]")
+{
+  STATIC_REQUIRE(
+    !::cuda::std::is_invocable_v<cuda::experimental::execution::get_execution_policy_t, const env_with_query_non_const&>);
+}
+
+struct env_with_query_and_method
+{
+  execution::any_execution_policy pol_{};
+
+  execution::any_execution_policy get_execution_policy() const noexcept
+  {
+    return pol_;
+  }
+
+  execution::any_execution_policy query(cuda::experimental::execution::get_execution_policy_t) const noexcept
+  {
+    return pol_;
+  }
+};
+C2H_TEST("Can call get_execution_policy on a type with both get_execution_policy and query", "[execution][policies]")
+{
+  env_with_query_and_method val{};
+  auto&& res = cuda::experimental::execution::get_execution_policy(val);
+  STATIC_REQUIRE(cuda::std::is_same_v<decltype(res), execution::any_execution_policy&&>);
+  CHECK(val.pol_ == res);
+}

cccl_upstream/cudax/test/execution/policies/policies.cu

@@ -0,0 +1,34 @@
+//===----------------------------------------------------------------------===//
+//
+// 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) 2024 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cuda/std/execution>
+#include <cuda/std/type_traits>
+
+#include <cuda/experimental/execution.cuh>
+
+#include <testing.cuh>
+
+namespace cudax = cuda::experimental;
+
+template <class T, class U>
+using is_same = cuda::std::is_same<cuda::std::remove_cvref_t<T>, U>;
+
+C2H_TEST("Execution policies", "[execution][policies]")
+{
+  namespace execution = cuda::std::execution;
+  SECTION("Individual options")
+  {
+    cudax::execution::any_execution_policy pol = execution::seq;
+    pol                                        = execution::par;
+    pol                                        = execution::par_unseq;
+    pol                                        = execution::unseq;
+    CHECK(pol == execution::unseq);
+  }
+}