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
66 lines
2.2 KiB
Plaintext
66 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/stream/interfaces/hashtable_linearprobing.cuh>
|
|
#include <cuda/experimental/__stf/stream/stream_ctx.cuh>
|
|
|
|
using namespace cuda::experimental::stf;
|
|
|
|
// Insert the key/values in kvs into the hashtable
|
|
__global__ void gpu_hashtable_insert_kernel(hashtable h, const reserved::KeyValue* kvs, unsigned int numkvs)
|
|
{
|
|
unsigned int threadid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
if (threadid < numkvs)
|
|
{
|
|
h.insert(kvs[threadid]);
|
|
}
|
|
}
|
|
|
|
void gpu_hashtable_insert(hashtable d_h, const reserved::KeyValue* device_kvs, unsigned int num_kvs, cudaStream_t stream)
|
|
{
|
|
// Have CUDA calculate the thread block size
|
|
const auto occ_res = reserved::compute_occupancy(gpu_hashtable_insert_kernel);
|
|
int threadblocksize = occ_res.block_size;
|
|
int gridsize = ((uint32_t) num_kvs + threadblocksize - 1) / threadblocksize;
|
|
|
|
// Insert all the keys into the hash table
|
|
gpu_hashtable_insert_kernel<<<gridsize, threadblocksize, 0, stream>>>(d_h, device_kvs, (uint32_t) num_kvs);
|
|
}
|
|
|
|
int main()
|
|
{
|
|
stream_ctx ctx;
|
|
|
|
// This constructor automatically initializes an empty hashtable on the host
|
|
hashtable h;
|
|
auto lh = ctx.logical_data(h);
|
|
|
|
// Create an array of values on the host
|
|
reserved::KeyValue kvs_array[16];
|
|
for (uint32_t i = 0; i < 16; i++)
|
|
{
|
|
kvs_array[i].key = i * 10;
|
|
kvs_array[i].value = 17 + i * 14;
|
|
}
|
|
auto h_kvs_array = ctx.logical_data(make_slice(&kvs_array[0], 16));
|
|
|
|
ctx.task(lh.rw(), h_kvs_array.read())->*[](auto stream, auto h, auto a) {
|
|
gpu_hashtable_insert(h, a.data_handle(), static_cast<uint32_t>(a.extent(0)), stream);
|
|
};
|
|
|
|
ctx.finalize();
|
|
|
|
// Thanks to the write-back mechanism on lh, h has been updated
|
|
for (size_t i = 0; i < 16; i++)
|
|
{
|
|
assert(h.get(i * 10) == 17 + i * 14);
|
|
}
|
|
}
|