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
83 lines
3.3 KiB
Plaintext
83 lines
3.3 KiB
Plaintext
//===----------------------------------------------------------------------===//
|
|
//
|
|
// Part of libcu++, the C++ Standard Library for your entire system,
|
|
// 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) 2026 NVIDIA CORPORATION & AFFILIATES.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include <thrust/device_vector.h>
|
|
|
|
#include <cuda/iterator>
|
|
#include <cuda/memory_pool>
|
|
#include <cuda/std/execution>
|
|
#include <cuda/stream>
|
|
|
|
#include "nvbench_helper.cuh"
|
|
|
|
template <typename T>
|
|
static void range_iter(nvbench::state& state, nvbench::type_list<T>)
|
|
{
|
|
T val = 1;
|
|
// set up input
|
|
const auto elements = static_cast<std::size_t>(state.get_int64("Elements"));
|
|
const auto common_prefix = state.get_float64("MismatchAt");
|
|
const auto mismatch_point = static_cast<std::size_t>(static_cast<double>(elements) * common_prefix);
|
|
|
|
thrust::device_vector<T> dinput(elements, thrust::no_init);
|
|
cuda::std::fill(cuda::execution::gpu, dinput.begin(), dinput.begin() + mismatch_point, T{0});
|
|
cuda::std::fill(cuda::execution::gpu, dinput.begin() + mismatch_point, dinput.end(), val);
|
|
|
|
state.add_global_memory_reads<T>(mismatch_point + 1);
|
|
state.add_global_memory_writes<size_t>(1);
|
|
|
|
caching_allocator_t alloc{};
|
|
|
|
state.exec(
|
|
nvbench::exec_tag::gpu | nvbench::exec_tag::no_batch | nvbench::exec_tag::sync, [&](nvbench::launch& launch) {
|
|
do_not_optimize(
|
|
cuda::std::equal(cuda_policy(alloc, launch), dinput.begin(), dinput.end(), cuda::constant_iterator<T>{0}));
|
|
});
|
|
}
|
|
|
|
NVBENCH_BENCH_TYPES(range_iter, NVBENCH_TYPE_AXES(fundamental_types))
|
|
.set_name("base_range_iter")
|
|
.add_int64_power_of_two_axis("Elements", nvbench::range(16, 28, 4))
|
|
.add_float64_axis("MismatchAt", std::vector{1.0, 0.5, 0.01});
|
|
|
|
template <typename T>
|
|
static void range_range(nvbench::state& state, nvbench::type_list<T>)
|
|
{
|
|
T val = 1;
|
|
// set up input
|
|
const auto elements = static_cast<std::size_t>(state.get_int64("Elements"));
|
|
const auto common_prefix = state.get_float64("MismatchAt");
|
|
const auto mismatch_point = static_cast<std::size_t>(static_cast<double>(elements) * common_prefix);
|
|
|
|
thrust::device_vector<T> dinput(elements, thrust::no_init);
|
|
cuda::std::fill(cuda::execution::gpu, dinput.begin(), dinput.begin() + mismatch_point, T{0});
|
|
cuda::std::fill(cuda::execution::gpu, dinput.begin() + mismatch_point, dinput.end(), val);
|
|
|
|
state.add_global_memory_reads<T>(mismatch_point + 1);
|
|
state.add_global_memory_writes<size_t>(1);
|
|
|
|
caching_allocator_t alloc{};
|
|
|
|
state.exec(nvbench::exec_tag::gpu | nvbench::exec_tag::no_batch | nvbench::exec_tag::sync,
|
|
[&](nvbench::launch& launch) {
|
|
do_not_optimize(cuda::std::equal(
|
|
cuda_policy(alloc, launch),
|
|
dinput.begin(),
|
|
dinput.end(),
|
|
cuda::constant_iterator<T>{0},
|
|
cuda::constant_iterator<T>{0, elements}));
|
|
});
|
|
}
|
|
|
|
NVBENCH_BENCH_TYPES(range_range, NVBENCH_TYPE_AXES(fundamental_types))
|
|
.set_name("base_range_range")
|
|
.add_int64_power_of_two_axis("Elements", nvbench::range(16, 28, 4))
|
|
.add_float64_axis("MismatchAt", std::vector{1.0, 0.5, 0.01});
|