[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/utility/unstable_unique.cu

@@ -0,0 +1,101 @@
+//===----------------------------------------------------------------------===//
+//
+// 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) 2022-2025 NVIDIA CORPORATION & AFFILIATES.
+//
+//===----------------------------------------------------------------------===//
+
+#include <cuda/std/functional>
+
+#include <cuda/experimental/__utility/unstable_unique.cuh>
+
+#include <algorithm>
+#include <iterator>
+#include <list>
+#include <vector>
+
+#include <c2h/catch2_test_helper.h>
+
+namespace cudax = cuda::experimental;
+
+TEST_CASE("unstable_unique empty range", "[utility]")
+{
+  std::vector<int> v;
+  auto new_end = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(v.end() == new_end);
+}
+
+TEST_CASE("unstable_unique no duplicates", "[utility]")
+{
+  std::vector<int> v = {1, 2, 3, 4, 5};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(v.end() == new_end);
+  REQUIRE(std::vector<int>({1, 2, 3, 4, 5}) == v);
+}
+
+TEST_CASE("unstable_unique leading duplicates", "[utility]")
+{
+  std::vector<int> v = {1, 1, 2, 3, 4, 5};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(v.begin() + 5 == new_end);
+  REQUIRE(std::vector<int>({1, 5, 2, 3, 4, 5}) == v);
+}
+
+TEST_CASE("unstable_unique interleaved duplicates", "[utility]")
+{
+  std::vector<int> v = {1, 1, 2, 2, 3, 3, 4, 4, 5, 5};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(v.begin() + 5 == new_end);
+  REQUIRE(std::vector<int>({1, 5, 2, 4, 3, 3, 4, 4, 5, 5}) == v);
+}
+
+TEST_CASE("unstable_unique all same", "[utility]")
+{
+  std::vector<int> v = {1, 1, 1, 1, 1};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(1 + v.begin() == new_end);
+  REQUIRE(std::vector<int>({1, 1, 1, 1, 1}) == v);
+}
+
+TEST_CASE("unstable_unique trailing unique", "[utility]")
+{
+  std::vector<int> v = {1, 1, 1, 1, 1, 2};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end());
+  REQUIRE(v.begin() + 2 == new_end);
+  REQUIRE(std::vector<int>({1, 2, 1, 1, 1, 2}) == v);
+}
+
+TEST_CASE("unstable_unique with custom predicate", "[utility]")
+{
+  std::vector<int> v = {1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 3, 4, 5};
+  auto new_end       = cudax::unstable_unique(v.begin(), v.end(), cuda::std::equal_to<int>{});
+  REQUIRE(v.begin() + 5 == new_end);
+  REQUIRE(std::vector<int>{1, 5, 4, 3, 2, 1, 1, 1, 1, 2, 2, 2, 3, 4, 5} == v);
+}
+
+TEST_CASE("unstable_unique on bidirectional iterators (std::list)", "[utility]")
+{
+  // std::list has bidirectional (not random-access) iterators -- exercises the
+  // !=-based loop termination path.
+  std::list<int> l = {1, 1, 2, 2, 3, 3, 4, 4, 5, 5};
+  auto new_end     = cudax::unstable_unique(l.begin(), l.end());
+  REQUIRE(std::distance(l.begin(), new_end) == 5);
+
+  std::vector<int> deduped(l.begin(), new_end);
+  std::sort(deduped.begin(), deduped.end());
+  REQUIRE(std::vector<int>({1, 2, 3, 4, 5}) == deduped);
+}
+
+TEST_CASE("unstable_unique on bidirectional iterators with custom predicate", "[utility]")
+{
+  std::list<int> l = {1, 1, 1, 1, 1, 2, 2, 3};
+  auto new_end     = cudax::unstable_unique(l.begin(), l.end(), cuda::std::equal_to<int>{});
+  REQUIRE(std::distance(l.begin(), new_end) == 3);
+
+  std::vector<int> deduped(l.begin(), new_end);
+  std::sort(deduped.begin(), deduped.end());
+  REQUIRE(std::vector<int>({1, 2, 3}) == deduped);
+}