dylanyunlong/project_6
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[CCCL] Add missing CCCL components: c2h, nvbench_helper, cmake, cudax, AGENTS.md

Browse Source 
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
This commit is contained in:
muh-bot
2026-08-06 02:14:18 +00:00
parent b0d597363a
commit dedf08166a
864 changed files with 174321 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
cccl_upstream/cudax/test/stf/loop_dispatch/dispatch_on_streams.cu Normal file
View File

@@ -0,0 +1,45 @@
//===----------------------------------------------------------------------===//
//
// 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/__places/place_partition.cuh>
#include <cuda/experimental/__stf/internal/stf_places_partition_into_stf.cuh>
#include <cuda/experimental/stf.cuh>
using namespace cuda::experimental::stf;
int main()
{
#if _CCCL_CTK_BELOW(12, 4)
fprintf(stderr, "Green contexts are not supported by this version of CUDA: skipping test.\n");
return 0;
#else // ^^^ _CCCL_CTK_BELOW(12, 4) ^^^ / vvv _CCCL_CTK_AT_LEAST(12, 4) vvv
context ctx;
auto lX = ctx.logical_data<int>(size_t(32 * 1024 * 1024));
ctx.parallel_for(lX.shape(), lX.write())->*[] __device__(size_t i, auto x) {
x(i) = 3 * i - 7;
};
for (auto& sub_place :
place_partition(exec_place::current_device(), ctx.async_resources(), place_partition_scope::green_context))
{
for (size_t i = 0; i < 4; i++)
{
ctx.parallel_for(sub_place, lX.shape(), lX.rw())->*[] __device__(size_t i, auto x) {
x(i) += 1;
};
}
}
ctx.finalize();
#endif // ^^^ _CCCL_CKT_AT_LEAST(12, 4) ^^^
}

41
cccl_upstream/cudax/test/stf/loop_dispatch/loop_dispatch.cu Normal file
View File

@@ -0,0 +1,41 @@
//===----------------------------------------------------------------------===//
//
// 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/internal/loop_dispatch.cuh>
#include <cuda/experimental/stf.cuh>
using namespace cuda::experimental::stf;
int main()
{
context ctx;
// Loop count
int n = 1024;
auto lB = ctx.logical_data<int>(size_t(1024 * 1024));
ctx.parallel_for(lB.shape(), lB.write())->*[] __device__(size_t i, auto b) {
b(i) = 42;
};
loop_dispatch(ctx, exec_place::all_devices(), 0, n, [&](size_t) {
auto lA = ctx.logical_data<int>(size_t(1024 * 1024));
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.write())->*[] __device__(size_t i, auto a) {
a(i) = (int) (10.0 * cos((double) i));
};
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.rw(), lB.read())->*[] __device__(size_t i, auto a, auto b) {
a(i) += b(i);
};
});
ctx.finalize();
}

52
cccl_upstream/cudax/test/stf/loop_dispatch/nested_loop_dispatch.cu Normal file
View File

@@ -0,0 +1,52 @@
//===----------------------------------------------------------------------===//
//
// 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/std/cmath>
#include <cuda/experimental/__stf/internal/loop_dispatch.cuh>
#include <cuda/experimental/stf.cuh>
using namespace cuda::experimental::stf;
int main()
{
context ctx;
auto lB = ctx.logical_data<int>(size_t(1024 * 1024));
ctx.parallel_for(lB.shape(), lB.write())->*[] __device__(size_t i, auto b) {
b(i) = 42;
};
// A fake grid which should work regardless of the underlying machine
auto grid = exec_place::repeat(exec_place::current_device(), 8);
// Split the affinity into 4 parts
loop_dispatch(ctx, grid, place_partition_scope::cuda_device, 0, 4, [&](size_t) {
// We should have 2 places per subplace
EXPECT(ctx.current_affinity().size() == 2);
// This should use ctx.current_affinity() implicitly
loop_dispatch(ctx, 0, 4, [&](size_t) {
auto lA = ctx.logical_data<int>(size_t(1024 * 1024));
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.write())->*[] __device__(size_t i, auto a) {
a(i) = (int) (10.0 * cuda::std::cos((double) i));
};
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.rw(), lB.read())
->*[] __device__(size_t i, auto a, auto b) {
a(i) += b(i);
};
});
});
ctx.finalize();
}

50
cccl_upstream/cudax/test/stf/loop_dispatch/stackable_loop_dispatch.cu Normal file
View File

@@ -0,0 +1,50 @@
//===----------------------------------------------------------------------===//
//
// 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/internal/loop_dispatch.cuh>
#include <cuda/experimental/stf.cuh>
using namespace cuda::experimental::stf;
int main()
{
stackable_ctx ctx;
_CCCL_ASSERT(ctx.has_head_set(), "ctx construction must set head for current thread");
// Loop count
int n = 1024;
auto lB = ctx.logical_data<int>(size_t(1024 * 1024)).set_symbol("B");
ctx.parallel_for(lB.shape(), lB.write())->*[] __device__(size_t i, auto b) {
b(i) = 42;
};
lB.set_read_only(true);
for (size_t iter = 0; iter < 4; iter++)
{
loop_dispatch(ctx, exec_place::all_devices(), place_partition_scope::green_context, 0, n, [&](size_t iter) {
auto lA = ctx.logical_data<int>(size_t(1024 * 1024)).set_symbol(::std::string("A") + ::std::to_string(iter));
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.write()).set_symbol("pfor1" + ::std::to_string(iter))
->*[] __device__(size_t i, auto a) {
a(i) = (int) (10.0 * cos((double) i));
};
ctx.parallel_for(ctx.current_exec_place(), lA.shape(), lA.rw(), lB.read())
.set_symbol("pfor2" + ::std::to_string(iter))
->*[] __device__(size_t i, auto a, auto b) {
a(i) += b(i);
};
});
}
ctx.finalize();
}