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
137 lines
4.8 KiB
Plaintext
137 lines
4.8 KiB
Plaintext
#include <thrust/extrema.h>
|
|
#include <thrust/iterator/retag.h>
|
|
|
|
#include <cuda/iterator>
|
|
|
|
#include <unittest/unittest.h>
|
|
|
|
template <class Vector>
|
|
void TestMinMaxElementSimple()
|
|
{
|
|
Vector data{3, 5, 1, 2, 5, 1};
|
|
|
|
ASSERT_EQUAL(*thrust::minmax_element(data.begin(), data.end()).first, 1);
|
|
ASSERT_EQUAL(*thrust::minmax_element(data.begin(), data.end()).second, 5);
|
|
ASSERT_EQUAL(thrust::minmax_element(data.begin(), data.end()).first - data.begin(), 2);
|
|
ASSERT_EQUAL(thrust::minmax_element(data.begin(), data.end()).second - data.begin(), 1);
|
|
}
|
|
DECLARE_VECTOR_UNITTEST(TestMinMaxElementSimple);
|
|
|
|
template <class Vector>
|
|
void TestMinMaxElementWithTransform()
|
|
{
|
|
using T = typename Vector::value_type;
|
|
|
|
Vector data{3, 5, 1, 2, 5, 1};
|
|
|
|
ASSERT_EQUAL(*thrust::minmax_element(thrust::make_transform_iterator(data.begin(), ::cuda::std::negate<T>()),
|
|
thrust::make_transform_iterator(data.end(), ::cuda::std::negate<T>()))
|
|
.first,
|
|
-5);
|
|
ASSERT_EQUAL(*thrust::minmax_element(thrust::make_transform_iterator(data.begin(), ::cuda::std::negate<T>()),
|
|
thrust::make_transform_iterator(data.end(), ::cuda::std::negate<T>()))
|
|
.second,
|
|
-1);
|
|
}
|
|
DECLARE_VECTOR_UNITTEST(TestMinMaxElementWithTransform);
|
|
|
|
template <typename T>
|
|
void TestMinMaxElement(const size_t n)
|
|
{
|
|
thrust::host_vector<T> h_data = unittest::random_samples<T>(n);
|
|
thrust::device_vector<T> d_data = h_data;
|
|
|
|
typename thrust::host_vector<T>::iterator h_min;
|
|
typename thrust::host_vector<T>::iterator h_max;
|
|
typename thrust::device_vector<T>::iterator d_min;
|
|
typename thrust::device_vector<T>::iterator d_max;
|
|
|
|
h_min = thrust::minmax_element(h_data.begin(), h_data.end()).first;
|
|
d_min = thrust::minmax_element(d_data.begin(), d_data.end()).first;
|
|
h_max = thrust::minmax_element(h_data.begin(), h_data.end()).second;
|
|
d_max = thrust::minmax_element(d_data.begin(), d_data.end()).second;
|
|
|
|
ASSERT_EQUAL(h_min - h_data.begin(), d_min - d_data.begin());
|
|
ASSERT_EQUAL(h_max - h_data.begin(), d_max - d_data.begin());
|
|
|
|
h_max = thrust::minmax_element(h_data.begin(), h_data.end(), ::cuda::std::greater<T>()).first;
|
|
d_max = thrust::minmax_element(d_data.begin(), d_data.end(), ::cuda::std::greater<T>()).first;
|
|
h_min = thrust::minmax_element(h_data.begin(), h_data.end(), ::cuda::std::greater<T>()).second;
|
|
d_min = thrust::minmax_element(d_data.begin(), d_data.end(), ::cuda::std::greater<T>()).second;
|
|
|
|
ASSERT_EQUAL(h_min - h_data.begin(), d_min - d_data.begin());
|
|
ASSERT_EQUAL(h_max - h_data.begin(), d_max - d_data.begin());
|
|
}
|
|
DECLARE_VARIABLE_UNITTEST(TestMinMaxElement);
|
|
|
|
template <typename ForwardIterator>
|
|
cuda::std::pair<ForwardIterator, ForwardIterator>
|
|
minmax_element(my_system& system, ForwardIterator first, ForwardIterator)
|
|
{
|
|
system.validate_dispatch();
|
|
return cuda::std::make_pair(first, first);
|
|
}
|
|
|
|
void TestMinMaxElementDispatchExplicit()
|
|
{
|
|
thrust::device_vector<int> vec(1);
|
|
|
|
my_system sys(0);
|
|
thrust::minmax_element(sys, vec.begin(), vec.end());
|
|
|
|
ASSERT_EQUAL(true, sys.is_valid());
|
|
}
|
|
DECLARE_UNITTEST(TestMinMaxElementDispatchExplicit);
|
|
|
|
template <typename ForwardIterator>
|
|
cuda::std::pair<ForwardIterator, ForwardIterator> minmax_element(my_tag, ForwardIterator first, ForwardIterator)
|
|
{
|
|
*first = 13;
|
|
return cuda::std::make_pair(first, first);
|
|
}
|
|
|
|
void TestMinMaxElementDispatchImplicit()
|
|
{
|
|
thrust::device_vector<int> vec(1);
|
|
|
|
thrust::minmax_element(thrust::retag<my_tag>(vec.begin()), thrust::retag<my_tag>(vec.end()));
|
|
|
|
ASSERT_EQUAL(13, vec.front());
|
|
}
|
|
DECLARE_UNITTEST(TestMinMaxElementDispatchImplicit);
|
|
|
|
void TestMinMaxElementWithBigIndexesHelper(int magnitude)
|
|
{
|
|
using Iter = thrust::counting_iterator<long long>;
|
|
Iter begin(1);
|
|
Iter end = begin + (1ll << magnitude);
|
|
ASSERT_EQUAL(::cuda::std::distance(begin, end), 1ll << magnitude);
|
|
|
|
cuda::std::pair<Iter, Iter> result = thrust::minmax_element(thrust::device, begin, end);
|
|
ASSERT_EQUAL(*result.first, 1);
|
|
ASSERT_EQUAL(*result.second, (1ll << magnitude));
|
|
|
|
result = thrust::minmax_element(thrust::device, begin, end, ::cuda::std::greater<long long>());
|
|
ASSERT_EQUAL(*result.second, 1);
|
|
ASSERT_EQUAL(*result.first, (1ll << magnitude));
|
|
}
|
|
|
|
void TestMinMaxElementWithBigIndexes()
|
|
{
|
|
TestMinMaxElementWithBigIndexesHelper(30);
|
|
#ifndef THRUST_FORCE_32_BIT_OFFSET_TYPE
|
|
TestMinMaxElementWithBigIndexesHelper(31);
|
|
TestMinMaxElementWithBigIndexesHelper(32);
|
|
TestMinMaxElementWithBigIndexesHelper(33);
|
|
#endif
|
|
}
|
|
DECLARE_UNITTEST(TestMinMaxElementWithBigIndexes);
|
|
|
|
void TestMinElementCudaIterator()
|
|
{
|
|
auto result = thrust::minmax_element(thrust::device, cuda::counting_iterator{0}, cuda::counting_iterator{0} + 100);
|
|
ASSERT_EQUAL(*result.first, 0);
|
|
ASSERT_EQUAL(*result.second, 99);
|
|
}
|
|
DECLARE_UNITTEST(TestMinElementCudaIterator);
|