我在CUDA中编写了一个小程序,计算C数组中有多少3个并打印出来。
#include <stdio.h>
#include <assert.h>
#include <cuda.h>
#include <cstdlib>
__global__ void incrementArrayOnDevice(int *a, int N, int *count)
{
int id = blockIdx.x * blockDim.x + threadIdx.x;
//__shared__ int s_a[512]; // one for each thread
//s_a[threadIdx.x] = a[id];
if( id < N )
{
//if( s_a[threadIdx.x] == 3 )
if( a[id] == 3 )
{
atomicAdd(count, 1);
}
}
}
int main(void)
{
int *a_h; // host memory
int *a_d; // device memory
int N = 16777216;
// allocate array on host
a_h = (int*)malloc(sizeof(int) * N);
for(int i = 0; i < N; ++i)
a_h[i] = (i % 3 == 0 ? 3 : 1);
// allocate arrays on device
cudaMalloc(&a_d, sizeof(int) * N);
// copy data from host to device
cudaMemcpy(a_d, a_h, sizeof(int) * N, cudaMemcpyHostToDevice);
// do calculation on device
int blockSize = 512;
int nBlocks = N / blockSize + (N % blockSize == 0 ? 0 : 1);
printf("number of blocks: %d\n", nBlocks);
int count;
int *devCount;
cudaMalloc(&devCount, sizeof(int));
cudaMemset(devCount, 0, sizeof(int));
incrementArrayOnDevice<<<nBlocks, blockSize>>> (a_d, N, devCount);
// retrieve result from device
cudaMemcpy(&count, devCount, sizeof(int), cudaMemcpyDeviceToHost);
printf("%d\n", count);
free(a_h);
cudaFree(a_d);
cudaFree(devCount);
}
我得到的结果是: 真正的0m3.025s 用户0m2.989s sys 0m0.029s
当我在具有4个线程的CPU上运行时,我得到: 真正的0m0.101s 用户0m0.100s sys 0m0.024s
请注意,GPU是旧的 - 我不知道确切的模型,因为我没有root访问权限,但它使用MESA驱动程序运行的OpenGL版本为1.2。
我做错了吗?我该怎么做才能让它跑得更快?
注意:我已经尝试过为每个块使用存储桶(因此每个存储块的atomicAdd()都会减少)但是我得到了完全相同的性能。 我还尝试将分配给该块的512个整数复制到共享内存块(您可以在注释中看到它),时间也是一样的。
答案 0 :(得分:0)
这是对你的问题“我能做些什么才能让它跑得更快?”的回答。正如我在评论中提到的,时间方法存在问题(可能),我对速度改进的主要建议是使用“经典并行缩减”算法。以下代码实现了更好的(在我看来)时序测量,并将内核转换为简化样式内核:
#include <stdio.h>
#include <assert.h>
#include <cstdlib>
#define N (1<<24)
#define nTPB 512
#define NBLOCKS 32
__global__ void incrementArrayOnDevice(int *a, int n, int *count)
{
__shared__ int lcnt[nTPB];
int id = blockIdx.x * blockDim.x + threadIdx.x;
int lcount = 0;
while (id < n) {
if (a[id] == 3) lcount++;
id += gridDim.x * blockDim.x;
}
lcnt[threadIdx.x] = lcount;
__syncthreads();
int stride = blockDim.x;
while(stride > 1) {
// assume blockDim.x is a power of 2
stride >>= 1;
if (threadIdx.x < stride) lcnt[threadIdx.x] += lcnt[threadIdx.x + stride];
__syncthreads();
}
if (threadIdx.x == 0) atomicAdd(count, lcnt[0]);
}
int main(void)
{
int *a_h; // host memory
int *a_d; // device memory
cudaEvent_t gstart1,gstart2,gstop1,gstop2,cstart,cstop;
float etg1, etg2, etc;
cudaEventCreate(&gstart1);
cudaEventCreate(&gstart2);
cudaEventCreate(&gstop1);
cudaEventCreate(&gstop2);
cudaEventCreate(&cstart);
cudaEventCreate(&cstop);
// allocate array on host
a_h = (int*)malloc(sizeof(int) * N);
for(int i = 0; i < N; ++i)
a_h[i] = (i % 3 == 0 ? 3 : 1);
// allocate arrays on device
cudaMalloc(&a_d, sizeof(int) * N);
int blockSize = nTPB;
int nBlocks = NBLOCKS;
printf("number of blocks: %d\n", nBlocks);
int count;
int *devCount;
cudaMalloc(&devCount, sizeof(int));
cudaMemset(devCount, 0, sizeof(int));
// copy data from host to device
cudaEventRecord(gstart1);
cudaMemcpy(a_d, a_h, sizeof(int) * N, cudaMemcpyHostToDevice);
cudaMemset(devCount, 0, sizeof(int));
cudaEventRecord(gstart2);
// do calculation on device
incrementArrayOnDevice<<<nBlocks, blockSize>>> (a_d, N, devCount);
cudaEventRecord(gstop2);
// retrieve result from device
cudaMemcpy(&count, devCount, sizeof(int), cudaMemcpyDeviceToHost);
cudaEventRecord(gstop1);
printf("GPU count = %d\n", count);
int hostCount = 0;
cudaEventRecord(cstart);
for (int i=0; i < N; i++)
if (a_h[i] == 3) hostCount++;
cudaEventRecord(cstop);
printf("CPU count = %d\n", hostCount);
cudaEventSynchronize(cstop);
cudaEventElapsedTime(&etg1, gstart1, gstop1);
cudaEventElapsedTime(&etg2, gstart2, gstop2);
cudaEventElapsedTime(&etc, cstart, cstop);
printf("GPU total time = %fs\n", (etg1/(float)1000) );
printf("GPU compute time = %fs\n", (etg2/(float)1000));
printf("CPU time = %fs\n", (etc/(float)1000));
free(a_h);
cudaFree(a_d);
cudaFree(devCount);
}
当我在相当快的GPU(Quadro 5000,比Tesla M2050慢一点)上运行时,我得到以下结果:
number of blocks: 32
GPU count = 5592406
CPU count = 5592406
GPU total time = 0.025714s
GPU compute time = 0.000793s
CPU time = 0.017332s
我们发现GPU比计算部分的这种(天真的,单线程)CPU实现要快得多。当我们添加传输数据的成本时,GPU版本较慢但速度不会慢30倍。
通过比较,当我计算原始算法时,我得到了这样的数字:
GPU total time = 0.118131s
GPU compute time = 0.093213s
我的系统配置是Xeon X5560 CPU,RHEL 5.5,CUDA 5.0,Quadro5000 GPU。