dedf08166a
Added 863 files from NVIDIA/cccl sparse checkout: - c2h/ (27 files): Catch2 test helpers — generators, validators, runner - nvbench_helper/ (10 files): Benchmark harness utilities - cmake/ (29 files): CMake presets and build helpers - cudax/ (794 files): Experimental CUDA extensions - AGENTS.md: NVIDIA's official AI agent instructions for CCCL - CMakePresets.json: Standardized build configurations - cccl-version.json: Version tracking Also added CCCL_ASSET_MAP.md mapping all 4295 CCCL files to competition value and PRD items. cccl_upstream now covers 100% of competition-critical assets: - 27 tuning headers (SM80/90/100 benchmark data) - 32 dispatch headers (algorithm implementations) - 60 Thrust examples (correctness verification) - 217 CUB Catch2 tests (regression matrix) - 153 CUB benchmarks (parameter space search) - 18 CUB examples (API verification) - 27 test helpers + benchmark harness - 794 cudax experimental extensions
99 lines
2.2 KiB
Plaintext
99 lines
2.2 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.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include <cuda/experimental/stf.cuh>
|
|
|
|
/*
|
|
* The goal of this test is to ensure that using read access modes actually
|
|
* results in concurrent tasks
|
|
*/
|
|
|
|
using namespace cuda::experimental::stf;
|
|
|
|
/**
|
|
* @brief Call `__nanosleep` (potentially repeatedly) to sleep `nanoseconds` nanoseconds. Supports sleep times longer
|
|
* than 4 billion nanoseconds (i.e. 4 seconds).
|
|
*
|
|
* @param nanoseconds how many nanoseconds to sleep
|
|
* @return void
|
|
*/
|
|
__global__ void nano_sleep(unsigned long long nanoseconds)
|
|
{
|
|
#if defined(__CUDA_ARCH__) && (__CUDA_ARCH__ >= 700)
|
|
static constexpr auto m = std::numeric_limits<unsigned int>::max();
|
|
for (;;)
|
|
{
|
|
if (nanoseconds > m)
|
|
{
|
|
__nanosleep(m);
|
|
nanoseconds -= m;
|
|
}
|
|
else
|
|
{
|
|
__nanosleep(static_cast<unsigned int>(nanoseconds));
|
|
break;
|
|
}
|
|
}
|
|
#else
|
|
const clock_t end = clock() + nanoseconds / (1000000000ULL / CLOCKS_PER_SEC);
|
|
while (clock() < end)
|
|
{
|
|
// busy wait
|
|
}
|
|
#endif
|
|
}
|
|
|
|
void run(context& ctx, int NTASKS, int ms)
|
|
{
|
|
int dummy[1];
|
|
auto handle = ctx.logical_data(dummy);
|
|
|
|
ctx.task().add_deps(handle.rw())->*[](cudaStream_t stream) {
|
|
nano_sleep<<<1, 1, 0, stream>>>(0);
|
|
};
|
|
|
|
for (int iter = 0; iter < 10; iter++)
|
|
{
|
|
for (int k = 0; k < NTASKS; k++)
|
|
{
|
|
ctx.task().add_deps(handle.read())->*[&](cudaStream_t stream) {
|
|
nano_sleep<<<1, 1, 0, stream>>>(ms * 1000ULL * 1000ULL);
|
|
};
|
|
}
|
|
|
|
ctx.task().add_deps(handle.rw())->*[&](cudaStream_t stream) {
|
|
nano_sleep<<<1, 1, 0, stream>>>(0);
|
|
};
|
|
}
|
|
|
|
ctx.finalize();
|
|
}
|
|
|
|
int main(int argc, char** argv)
|
|
{
|
|
int NTASKS = 256;
|
|
int ms = 40;
|
|
|
|
if (argc > 1)
|
|
{
|
|
NTASKS = atoi(argv[1]);
|
|
}
|
|
|
|
if (argc > 2)
|
|
{
|
|
ms = atoi(argv[2]);
|
|
}
|
|
|
|
context ctx;
|
|
run(ctx, NTASKS, ms);
|
|
ctx = graph_ctx();
|
|
run(ctx, NTASKS, ms);
|
|
}
|