56fd68e7dd
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
39 lines
1.1 KiB
Plaintext
39 lines
1.1 KiB
Plaintext
#include <thrust/reduce.h>
|
|
|
|
#include <unittest/unittest.h>
|
|
|
|
template <typename T, unsigned int N>
|
|
void _TestReduceWithLargeTypes()
|
|
{
|
|
size_t n = (64 * 1024) / sizeof(FixedVector<T, N>);
|
|
|
|
thrust::host_vector<FixedVector<T, N>> h_data(n);
|
|
|
|
for (size_t i = 0; i < h_data.size(); i++)
|
|
{
|
|
h_data[i] = FixedVector<T, N>(static_cast<T>(i));
|
|
}
|
|
|
|
thrust::device_vector<FixedVector<T, N>> d_data = h_data;
|
|
|
|
FixedVector<T, N> h_result = thrust::reduce(h_data.begin(), h_data.end(), FixedVector<T, N>(T{0}));
|
|
FixedVector<T, N> d_result = thrust::reduce(d_data.begin(), d_data.end(), FixedVector<T, N>(T{0}));
|
|
|
|
ASSERT_EQUAL_QUIET(h_result, d_result);
|
|
}
|
|
|
|
void TestReduceWithLargeTypes()
|
|
{
|
|
_TestReduceWithLargeTypes<int, 4>();
|
|
_TestReduceWithLargeTypes<int, 8>();
|
|
_TestReduceWithLargeTypes<int, 16>();
|
|
|
|
// XXX these take too long to compile
|
|
// _TestReduceWithLargeTypes<int, 32>();
|
|
// _TestReduceWithLargeTypes<int, 64>();
|
|
// _TestReduceWithLargeTypes<int, 128>();
|
|
// _TestReduceWithLargeTypes<int, 256>();
|
|
// _TestReduceWithLargeTypes<int, 512>();
|
|
}
|
|
DECLARE_UNITTEST(TestReduceWithLargeTypes);
|