notes-ing2/gpu/tp1/c/src/ex4.cu
2024-03-10 21:31:32 +01:00

96 lines
2.4 KiB
Text

#include <iostream>
#define RANGE(I, FROM, TO) size_t I = FROM; I < TO; I += 1
//
// example: CUDA_CHECK( cudaMalloc(dx, x, N*sizeof(int) );
//
#define CUDA_CHECK(code) { cuda_check((code), __FILE__, __LINE__); }
inline void cuda_check(cudaError_t code, const char *file, int line) {
if(code != cudaSuccess) {
std::cout << file << ':' << line << ": [CUDA ERROR] " << cudaGetErrorString(code) << std::endl;
std::abort();
}
}
// step 06
__global__ void add_strided(int N, const int* dx, int* dy) {
size_t threads = blockDim.x * gridDim.x;
size_t items_per_threads = (N / threads) + 1;
size_t base_index = (blockIdx.x * blockDim.x + threadIdx.x) * items_per_threads;
for (RANGE(i, 0, items_per_threads)) {
size_t index = base_index + i;
if (index > N) continue;
dy[index] += dx[index];
}
}
int main()
{
constexpr int N = 1000;
int* x = (int*)malloc(N*sizeof(int));
int* y = (int*)malloc(N*sizeof(int));
for(int i = 0; i < N; ++i) {
x[i] = i;
y[i] = i*i;
}
// step 07
int* dx;
int* dy;
// 1. allocate on device
size_t size = N * sizeof(int);
cudaMalloc(&dx, size);
cudaMalloc(&dy, size);
// 2. copy from host to device
cudaMemcpy(dx, x, size, cudaMemcpyHostToDevice);
cudaMemcpy(dy, y, size, cudaMemcpyHostToDevice);
// 3. launch CUDA kernel
const int threads_per_bloc = 32;
const int blocs = 8;
add_strided<<<blocs, threads_per_bloc>>>(N, dx, dy);
cudaDeviceSynchronize();
// 4. copy result from device to host
cudaMemcpy(y, dy, size, cudaMemcpyDeviceToHost);
// 5. free device memory
cudaFree(dx);
cudaFree(dy);
// checking results
bool ok = true;
for(int i = 0; i < N; ++i) {
const int expected_result = i + i*i;
if(y[i] != expected_result) {
std::cout << "Failure" << std::endl;
std::cout << "Result at index i="
<< i << ": expected "
<< i << '+' << i*i << '=' << expected_result << ", got " << y[i] << std::endl;
ok = false;
break;
}
}
if(ok) std::cout << "Success" << std::endl;
free(x);
free(y);
return 0;
}
// # Question 4
// Pour N tâches, X threads en tout,
// - nous devons faire en moyenne N / X tâches par threads
// - un stride valable est ceil(N / X)