[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/thrust/testing/dereference.cu

@@ -0,0 +1,130 @@
+#include <thrust/device_ptr.h>
+#include <thrust/device_vector.h>
+#include <thrust/functional.h>
+#include <thrust/iterator/counting_iterator.h>
+#include <thrust/iterator/transform_iterator.h>
+
+#include <unittest/unittest.h>
+
+_CCCL_DIAG_PUSH
+_CCCL_DIAG_SUPPRESS_MSVC(4244 4267) // possible loss of data
+
+template <typename Iterator1, typename Iterator2>
+#if THRUST_DEVICE_SYSTEM == THRUST_DEVICE_SYSTEM_CUDA
+__global__
+#endif
+  void
+  simple_copy_on_device(Iterator1 first1, Iterator1 last1, Iterator2 first2)
+{
+  while (first1 != last1)
+  {
+    *(first2++) = *(first1++);
+  }
+}
+
+template <typename Iterator1, typename Iterator2>
+void simple_copy(Iterator1 first1, Iterator1 last1, Iterator2 first2)
+{
+#if THRUST_DEVICE_SYSTEM == THRUST_DEVICE_SYSTEM_CUDA
+  simple_copy_on_device<<<1, 1>>>(first1, last1, first2);
+#else
+  simple_copy_on_device(first1, last1, first2);
+#endif
+}
+
+void TestDeviceDereferenceDeviceVectorIterator()
+{
+  thrust::device_vector<int> input = unittest::random_integers<int>(100);
+  thrust::device_vector<int> output(input.size(), 0);
+
+  simple_copy(input.begin(), input.end(), output.begin());
+
+  ASSERT_EQUAL(input, output);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceDeviceVectorIterator);
+
+void TestDeviceDereferenceDevicePtr()
+{
+  thrust::device_vector<int> input = unittest::random_integers<int>(100);
+  thrust::device_vector<int> output(input.size(), 0);
+
+  thrust::device_ptr<int> _first1 = &input[0];
+  thrust::device_ptr<int> _last1  = _first1 + static_cast<std::ptrdiff_t>(input.size());
+  thrust::device_ptr<int> _first2 = &output[0];
+
+  simple_copy(_first1, _last1, _first2);
+
+  ASSERT_EQUAL(input, output);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceDevicePtr);
+
+void TestDeviceDereferenceTransformIterator()
+{
+  thrust::device_vector<int> input = unittest::random_integers<int>(100);
+  thrust::device_vector<int> output(input.size(), 0);
+
+  simple_copy(thrust::make_transform_iterator(input.begin(), ::cuda::std::identity{}),
+              thrust::make_transform_iterator(input.end(), ::cuda::std::identity{}),
+              output.begin());
+
+  ASSERT_EQUAL(input, output);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceTransformIterator);
+
+void TestDeviceDereferenceTransformIteratorInputConversion()
+{
+  thrust::device_vector<int> input = unittest::random_integers<int>(100);
+  thrust::device_vector<double> output(input.size(), 0);
+
+  simple_copy(thrust::make_transform_iterator(input.begin(), ::cuda::std::identity{}),
+              thrust::make_transform_iterator(input.end(), ::cuda::std::identity{}),
+              output.begin());
+
+  ASSERT_EQUAL(input == output, true);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceTransformIteratorInputConversion);
+
+void TestDeviceDereferenceTransformIteratorOutputConversion()
+{
+  thrust::device_vector<int> input = unittest::random_integers<int>(100);
+  thrust::device_vector<double> output(input.size(), 0);
+
+  simple_copy(thrust::make_transform_iterator(input.begin(), ::cuda::std::identity{}),
+              thrust::make_transform_iterator(input.end(), ::cuda::std::identity{}),
+              output.begin());
+
+  ASSERT_EQUAL(input == output, true);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceTransformIteratorOutputConversion);
+
+void TestDeviceDereferenceCountingIterator()
+{
+  thrust::counting_iterator<int> first(1);
+  thrust::counting_iterator<int> last(6);
+
+  thrust::device_vector<int> output(5);
+
+  simple_copy(first, last, output.begin());
+
+  thrust::device_vector<int> ref{1, 2, 3, 4, 5};
+  ASSERT_EQUAL(output, ref);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceCountingIterator);
+
+void TestDeviceDereferenceTransformedCountingIterator()
+{
+  thrust::counting_iterator<int> first(1);
+  thrust::counting_iterator<int> last(6);
+
+  thrust::device_vector<int> output(5);
+
+  simple_copy(thrust::make_transform_iterator(first, ::cuda::std::negate<int>()),
+              thrust::make_transform_iterator(last, ::cuda::std::negate<int>()),
+              output.begin());
+
+  thrust::device_vector<int> ref{-1, -2, -3, -4, -5};
+  ASSERT_EQUAL(output, ref);
+}
+DECLARE_UNITTEST(TestDeviceDereferenceTransformedCountingIterator);
+
+_CCCL_DIAG_POP