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
94 lines
2.1 KiB
Plaintext
94 lines
2.1 KiB
Plaintext
//===----------------------------------------------------------------------===//
|
|
//
|
|
// Part of CUDASTF 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) 2022-2024 NVIDIA CORPORATION & AFFILIATES.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
/**
|
|
* @file
|
|
* @brief This test ensures that read() can be called on const logical_data
|
|
* (typed and untyped).
|
|
*/
|
|
|
|
#include <cuda/experimental/stf.cuh>
|
|
|
|
using namespace cuda::experimental::stf;
|
|
|
|
struct foo
|
|
{
|
|
// Intentionally choose an odd (actually prime) size
|
|
foo(context& ctx)
|
|
{
|
|
l = ctx.logical_data(shape_of<slice<int>>(50867));
|
|
}
|
|
|
|
void set(context& ctx, int val)
|
|
{
|
|
ctx.parallel_for(l.shape(), l.write())->*[=] _CCCL_DEVICE(size_t i, auto dl) {
|
|
dl(i) = val;
|
|
};
|
|
}
|
|
|
|
void copy_from(context& ctx, const foo& other)
|
|
{
|
|
ctx.parallel_for(l.shape(), l.write(), other.l.read())->*[=] _CCCL_DEVICE(size_t i, auto dl, auto dotherl) {
|
|
dl(i) = dotherl(i);
|
|
};
|
|
}
|
|
|
|
void ensure(context& ctx, int val)
|
|
{
|
|
std::ignore = val;
|
|
ctx.parallel_for(l.shape(), l.read())->*[=] _CCCL_DEVICE(size_t i, auto dl) {
|
|
assert(dl(i) == val);
|
|
};
|
|
}
|
|
|
|
auto& get_l() const
|
|
{
|
|
return l;
|
|
}
|
|
|
|
logical_data<slice<int>> l;
|
|
};
|
|
|
|
void read_only_access(context& ctx, const foo& f)
|
|
{
|
|
ctx.parallel_for(f.l.shape(), f.l.read())->*[] _CCCL_DEVICE(size_t i, auto dl) {
|
|
// no-op
|
|
};
|
|
|
|
ctx.parallel_for(f.get_l().shape(), f.get_l().read())->*[] _CCCL_DEVICE(size_t i, auto dl) {
|
|
// no-op
|
|
};
|
|
}
|
|
|
|
void read_only_access_untyped(const logical_data_untyped& ld)
|
|
{
|
|
auto dep = ld.read();
|
|
(void) dep;
|
|
}
|
|
|
|
int main()
|
|
{
|
|
context ctx;
|
|
|
|
foo A(ctx);
|
|
A.set(ctx, 42);
|
|
A.ensure(ctx, 42);
|
|
foo B(ctx);
|
|
B.copy_from(ctx, A);
|
|
B.ensure(ctx, 42);
|
|
|
|
read_only_access(ctx, A);
|
|
|
|
const logical_data_untyped& ld_untyped = A.l;
|
|
read_only_access_untyped(ld_untyped);
|
|
|
|
ctx.finalize();
|
|
}
|