Files
project_6/cccl_upstream/thrust/examples/device_ptr.cu
EngineX CI 56fd68e7dd [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
2026-07-30 09:35:51 +00:00

45 lines
1.3 KiB
Plaintext

#include <thrust/device_free.h>
#include <thrust/device_malloc.h>
#include <thrust/device_ptr.h>
#include <thrust/reduce.h>
#include <thrust/sequence.h>
#include <cassert>
#include <iostream>
int main()
{
// allocate memory buffer to store 10 integers on the device
thrust::device_ptr<int> d_ptr = thrust::device_malloc<int>(10);
// device_ptr supports pointer arithmetic
thrust::device_ptr<int> first = d_ptr;
thrust::device_ptr<int> last = d_ptr + 10;
std::cout << "device array contains " << cuda::std::distance(first, last) << " values\n";
// algorithms work as expected
thrust::sequence(first, last);
std::cout << "sum of values is " << thrust::reduce(first, last) << "\n";
// device memory can be read and written transparently
d_ptr[0] = 10;
d_ptr[1] = 11;
d_ptr[2] = d_ptr[0] + d_ptr[1];
// device_ptr can be converted to a "raw" pointer for use in other APIs and kernels, etc.
int* raw_ptr = thrust::raw_pointer_cast(d_ptr);
// note: raw_ptr cannot necessarily be accessed by the host!
// conversely, raw pointers can be wrapped
[[maybe_unused]] thrust::device_ptr<int> wrapped_ptr = thrust::device_pointer_cast(raw_ptr);
// back to where we started
assert(wrapped_ptr == d_ptr);
// deallocate device memory
thrust::device_free(d_ptr);
return 0;
}