NASP/cuda-samples
85
0
Fork 0
mirror of https://github.com/NVIDIA/cuda-samples.git synced 2026-10-25 22:28:17 +08:00
Code Issues Packages Projects Releases Wiki Activity

Apply consistent code formatting across the repo. Add clang-format and pre-commit hooks.

Browse Source
This commit is contained in:
Rob Armstrong
2025-03-27 10:30:07 -07:00
parent 2cd58fbc9a
commit ceab6e8bcc
782 changed files with 107230 additions and 106548 deletions
Download Patch File Download Diff File Expand all files Collapse all files

276
Samples/0_Introduction/simpleDrvRuntime/simpleDrvRuntime.cpp
View File

@@ -33,12 +33,12 @@
*/
// Includes
#include <cstring>
#include <cuda.h>
#include <cuda_runtime.h>
#include <iostream>
#include <stdio.h>
#include <string.h>
#include <cstring>
#include <iostream>
// includes, project
#include <helper_cuda.h>
@@ -62,165 +62,165 @@ float *d_B;
float *d_C;
// Functions
int CleanupNoFailure(CUcontext &cuContext);
int CleanupNoFailure(CUcontext &cuContext);
void RandomInit(float *, int);
bool findModulePath(const char *, string &, char **, ostringstream &);
static void check(CUresult result, char const *const func,
const char *const file, int const line) {
if (result) {
fprintf(stderr, "CUDA error at %s:%d code=%d \"%s\" \n", file, line,
static_cast<unsigned int>(result), func);
exit(EXIT_FAILURE);
}
static void check(CUresult result, char const *const func, const char *const file, int const line)
{
if (result) {
fprintf(stderr, "CUDA error at %s:%d code=%d \"%s\" \n", file, line, static_cast<unsigned int>(result), func);
exit(EXIT_FAILURE);
}
}
#define checkCudaDrvErrors(val) check((val), #val, __FILE__, __LINE__)
// Host code
int main(int argc, char **argv) {
printf("simpleDrvRuntime..\n");
int N = 50000, devID = 0;
size_t size = N * sizeof(float);
CUdevice cuDevice;
CUfunction vecAdd_kernel;
CUmodule cuModule = 0;
CUcontext cuContext;
int main(int argc, char **argv)
{
printf("simpleDrvRuntime..\n");
int N = 50000, devID = 0;
size_t size = N * sizeof(float);
CUdevice cuDevice;
CUfunction vecAdd_kernel;
CUmodule cuModule = 0;
CUcontext cuContext;
// Initialize
checkCudaDrvErrors(cuInit(0));
// Initialize
checkCudaDrvErrors(cuInit(0));
cuDevice = findCudaDevice(argc, (const char **)argv);
// Create context
checkCudaDrvErrors(cuCtxCreate(&cuContext, 0, cuDevice));
cuDevice = findCudaDevice(argc, (const char **)argv);
// Create context
checkCudaDrvErrors(cuCtxCreate(&cuContext, 0, cuDevice));
// first search for the module path before we load the results
string module_path;
ostringstream fatbin;
// first search for the module path before we load the results
string module_path;
ostringstream fatbin;
if (!findModulePath(FATBIN_FILE, module_path, argv, fatbin)) {
exit(EXIT_FAILURE);
} else {
printf("> initCUDA loading module: <%s>\n", module_path.c_str());
}
if (!fatbin.str().size()) {
printf("fatbin file empty. exiting..\n");
exit(EXIT_FAILURE);
}
// Create module from binary file (FATBIN)
checkCudaDrvErrors(cuModuleLoadData(&cuModule, fatbin.str().c_str()));
// Get function handle from module
checkCudaDrvErrors(
cuModuleGetFunction(&vecAdd_kernel, cuModule, "VecAdd_kernel"));
// Allocate input vectors h_A and h_B in host memory
checkCudaErrors(cudaMallocHost(&h_A, size));
checkCudaErrors(cudaMallocHost(&h_B, size));
checkCudaErrors(cudaMallocHost(&h_C, size));
// Initialize input vectors
RandomInit(h_A, N);
RandomInit(h_B, N);
// Allocate vectors in device memory
checkCudaErrors(cudaMalloc((void **)(&d_A), size));
checkCudaErrors(cudaMalloc((void **)(&d_B), size));
checkCudaErrors(cudaMalloc((void **)(&d_C), size));
cudaStream_t stream;
checkCudaErrors(cudaStreamCreateWithFlags(&stream, cudaStreamNonBlocking));
// Copy vectors from host memory to device memory
checkCudaErrors(
cudaMemcpyAsync(d_A, h_A, size, cudaMemcpyHostToDevice, stream));
checkCudaErrors(
cudaMemcpyAsync(d_B, h_B, size, cudaMemcpyHostToDevice, stream));
int threadsPerBlock = 256;
int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;
void *args[] = {&d_A, &d_B, &d_C, &N};
// Launch the CUDA kernel
checkCudaDrvErrors(cuLaunchKernel(vecAdd_kernel, blocksPerGrid, 1, 1,
threadsPerBlock, 1, 1, 0, stream, args,
NULL));
// Copy result from device memory to host memory
// h_C contains the result in host memory
checkCudaErrors(
cudaMemcpyAsync(h_C, d_C, size, cudaMemcpyDeviceToHost, stream));
checkCudaErrors(cudaStreamSynchronize(stream));
// Verify result
int i;
for (i = 0; i < N; ++i) {
float sum = h_A[i] + h_B[i];
if (fabs(h_C[i] - sum) > 1e-7f) {
break;
if (!findModulePath(FATBIN_FILE, module_path, argv, fatbin)) {
exit(EXIT_FAILURE);
}
else {
printf("> initCUDA loading module: <%s>\n", module_path.c_str());
}
}
checkCudaDrvErrors(cuModuleUnload(cuModule));
CleanupNoFailure(cuContext);
printf("%s\n", (i == N) ? "Result = PASS" : "Result = FAIL");
if (!fatbin.str().size()) {
printf("fatbin file empty. exiting..\n");
exit(EXIT_FAILURE);
}
exit((i == N) ? EXIT_SUCCESS : EXIT_FAILURE);
// Create module from binary file (FATBIN)
checkCudaDrvErrors(cuModuleLoadData(&cuModule, fatbin.str().c_str()));
// Get function handle from module
checkCudaDrvErrors(cuModuleGetFunction(&vecAdd_kernel, cuModule, "VecAdd_kernel"));
// Allocate input vectors h_A and h_B in host memory
checkCudaErrors(cudaMallocHost(&h_A, size));
checkCudaErrors(cudaMallocHost(&h_B, size));
checkCudaErrors(cudaMallocHost(&h_C, size));
// Initialize input vectors
RandomInit(h_A, N);
RandomInit(h_B, N);
// Allocate vectors in device memory
checkCudaErrors(cudaMalloc((void **)(&d_A), size));
checkCudaErrors(cudaMalloc((void **)(&d_B), size));
checkCudaErrors(cudaMalloc((void **)(&d_C), size));
cudaStream_t stream;
checkCudaErrors(cudaStreamCreateWithFlags(&stream, cudaStreamNonBlocking));
// Copy vectors from host memory to device memory
checkCudaErrors(cudaMemcpyAsync(d_A, h_A, size, cudaMemcpyHostToDevice, stream));
checkCudaErrors(cudaMemcpyAsync(d_B, h_B, size, cudaMemcpyHostToDevice, stream));
int threadsPerBlock = 256;
int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;
void *args[] = {&d_A, &d_B, &d_C, &N};
// Launch the CUDA kernel
checkCudaDrvErrors(
cuLaunchKernel(vecAdd_kernel, blocksPerGrid, 1, 1, threadsPerBlock, 1, 1, 0, stream, args, NULL));
// Copy result from device memory to host memory
// h_C contains the result in host memory
checkCudaErrors(cudaMemcpyAsync(h_C, d_C, size, cudaMemcpyDeviceToHost, stream));
checkCudaErrors(cudaStreamSynchronize(stream));
// Verify result
int i;
for (i = 0; i < N; ++i) {
float sum = h_A[i] + h_B[i];
if (fabs(h_C[i] - sum) > 1e-7f) {
break;
}
}
checkCudaDrvErrors(cuModuleUnload(cuModule));
CleanupNoFailure(cuContext);
printf("%s\n", (i == N) ? "Result = PASS" : "Result = FAIL");
exit((i == N) ? EXIT_SUCCESS : EXIT_FAILURE);
}
int CleanupNoFailure(CUcontext &cuContext) {
// Free device memory
checkCudaErrors(cudaFree(d_A));
checkCudaErrors(cudaFree(d_B));
checkCudaErrors(cudaFree(d_C));
int CleanupNoFailure(CUcontext &cuContext)
{
// Free device memory
checkCudaErrors(cudaFree(d_A));
checkCudaErrors(cudaFree(d_B));
checkCudaErrors(cudaFree(d_C));
// Free host memory
if (h_A) {
checkCudaErrors(cudaFreeHost(h_A));
}
// Free host memory
if (h_A) {
checkCudaErrors(cudaFreeHost(h_A));
}
if (h_B) {
checkCudaErrors(cudaFreeHost(h_B));
}
if (h_B) {
checkCudaErrors(cudaFreeHost(h_B));
}
if (h_C) {
checkCudaErrors(cudaFreeHost(h_C));
}
if (h_C) {
checkCudaErrors(cudaFreeHost(h_C));
}
checkCudaDrvErrors(cuCtxDestroy(cuContext));
checkCudaDrvErrors(cuCtxDestroy(cuContext));
return EXIT_SUCCESS;
return EXIT_SUCCESS;
}
// Allocates an array with random float entries.
void RandomInit(float *data, int n) {
for (int i = 0; i < n; ++i) {
data[i] = rand() / (float)RAND_MAX;
}
}
bool inline findModulePath(const char *module_file, string &module_path,
char **argv, ostringstream &ostrm) {
char *actual_path = sdkFindFilePath(module_file, argv[0]);
if (actual_path) {
module_path = actual_path;
} else {
printf("> findModulePath file not found: <%s> \n", module_file);
return false;
}
if (module_path.empty()) {
printf("> findModulePath could not find file: <%s> \n", module_file);
return false;
} else {
printf("> findModulePath found file at <%s>\n", module_path.c_str());
if (module_path.rfind("fatbin") != string::npos) {
ifstream fileIn(module_path.c_str(), ios::binary);
ostrm << fileIn.rdbuf();
void RandomInit(float *data, int n)
{
for (int i = 0; i < n; ++i) {
data[i] = rand() / (float)RAND_MAX;
}
}
bool inline findModulePath(const char *module_file, string &module_path, char **argv, ostringstream &ostrm)
{
char *actual_path = sdkFindFilePath(module_file, argv[0]);
if (actual_path) {
module_path = actual_path;
}
else {
printf("> findModulePath file not found: <%s> \n", module_file);
return false;
}
if (module_path.empty()) {
printf("> findModulePath could not find file: <%s> \n", module_file);
return false;
}
else {
printf("> findModulePath found file at <%s>\n", module_path.c_str());
if (module_path.rfind("fatbin") != string::npos) {
ifstream fileIn(module_path.c_str(), ios::binary);
ostrm << fileIn.rdbuf();
}
return true;
}
return true;
}
}

9
Samples/0_Introduction/simpleDrvRuntime/vectorAdd_kernel.cu
View File

@@ -34,9 +34,10 @@
*/
// Device code
extern "C" __global__ void VecAdd_kernel(const float *A, const float *B,
float *C, int N) {
int i = blockDim.x * blockIdx.x + threadIdx.x;
extern "C" __global__ void VecAdd_kernel(const float *A, const float *B, float *C, int N)
{
int i = blockDim.x * blockIdx.x + threadIdx.x;
if (i < N) C[i] = A[i] + B[i];
if (i < N)
C[i] = A[i] + B[i];
}