Mirza Halilčević
79 lines
2.8 KiB
C++
79 lines
2.8 KiB
C++
/*
|
|
Copyright (c) 2023 Advanced Micro Devices, Inc. All rights reserved.
|
|
|
|
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
of this software and associated documentation files (the "Software"), to deal
|
|
in the Software without restriction, including without limitation the rights
|
|
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
copies of the Software, and to permit persons to whom the Software is
|
|
furnished to do so, subject to the following conditions:
|
|
|
|
The above copyright notice and this permission notice shall be included in
|
|
all copies or substantial portions of the Software.
|
|
|
|
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
|
|
THE SOFTWARE.
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
enum class SyncthreadsKind { kDefault, kCount, kAnd, kOr };
|
|
|
|
template <SyncthreadsKind kind> __device__ int Syncthreads(int predicate) {
|
|
if constexpr (kind == SyncthreadsKind::kDefault) {
|
|
__syncthreads();
|
|
return 0;
|
|
} else if constexpr (kind == SyncthreadsKind::kCount) {
|
|
return __syncthreads_count(predicate);
|
|
} else if constexpr (kind == SyncthreadsKind::kAnd) {
|
|
return __syncthreads_and(predicate);
|
|
} else if constexpr (kind == SyncthreadsKind::kOr) {
|
|
return __syncthreads_or(predicate);
|
|
}
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsKernel(int* out) {
|
|
extern __shared__ int shared_mem[];
|
|
|
|
shared_mem[threadIdx.x] = threadIdx.x + 1;
|
|
|
|
Syncthreads<kind>(0);
|
|
|
|
if (threadIdx.x == 0) {
|
|
int sum = 0;
|
|
for (int i = 0; i < blockDim.x; ++i) {
|
|
sum += shared_mem[i];
|
|
}
|
|
out[blockIdx.x] = sum;
|
|
}
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsZeroKernel(int* out) {
|
|
int tid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
out[tid] = Syncthreads<kind>(0);
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsOneKernel(int* out) {
|
|
int tid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
out[tid] = Syncthreads<kind>(1);
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsOddEvenKernel(int* out) {
|
|
int tid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
out[tid] = Syncthreads<kind>(threadIdx.x % 2);
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsNegativeKernel(int* out) {
|
|
int tid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
out[tid] = Syncthreads<kind>(-1);
|
|
}
|
|
|
|
template <SyncthreadsKind kind> __global__ void SyncthreadsIdKernel(int* out) {
|
|
int tid = blockIdx.x * blockDim.x + threadIdx.x;
|
|
out[tid] = Syncthreads<kind>(threadIdx.x);
|
|
} |