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

Bug 5034785: Update all non-ctx nppi APIs to ctx APIs as per latest change on NPP

Browse Source
This commit is contained in:
Shawn Zeng
2025-02-27 03:01:47 -08:00
parent acd3a015c8
commit 7f0f63f311
6 changed files with 279 additions and 173 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
Samples/4_CUDA_Libraries/cannyEdgeDetectorNPP/cannyEdgeDetectorNPP.cpp
View File

@@ -68,26 +68,6 @@ inline int cudaDeviceInit(int argc, const char **argv) {
return dev;
}
bool printfNPPinfo(int argc, char *argv[]) {
const NppLibraryVersion *libVer = nppGetLibVersion();
printf("NPP Library Version %d.%d.%d\n", libVer->major, libVer->minor,
libVer->build);
int driverVersion, runtimeVersion;
cudaDriverGetVersion(&driverVersion);
cudaRuntimeGetVersion(&runtimeVersion);
printf(" CUDA Driver Version: %d.%d\n", driverVersion / 1000,
(driverVersion % 100) / 10);
printf(" CUDA Runtime Version: %d.%d\n", runtimeVersion / 1000,
(runtimeVersion % 100) / 10);
// Min spec is SM 1.0 devices
bool bVal = checkCudaCapabilities(1, 0);
return bVal;
}
int main(int argc, char *argv[]) {
printf("%s Starting...\n\n", argv[0]);
@@ -97,10 +77,50 @@ int main(int argc, char *argv[]) {
cudaDeviceInit(argc, (const char **)argv);
if (printfNPPinfo(argc, argv) == false) {
exit(EXIT_SUCCESS);
NppStreamContext nppStreamCtx;
nppStreamCtx.hStream = 0; // The NULL stream by default, set this to whatever your stream ID is if not the NULL stream.
cudaError_t cudaError = cudaGetDevice(&nppStreamCtx.nCudaDeviceId);
if (cudaError != cudaSuccess)
{
printf("CUDA error: no devices supporting CUDA.\n");
return NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY;
}
const NppLibraryVersion *libVer = nppGetLibVersion();
printf("NPP Library Version %d.%d.%d\n", libVer->major, libVer->minor, libVer->build);
int driverVersion, runtimeVersion;
cudaDriverGetVersion(&driverVersion);
cudaRuntimeGetVersion(&runtimeVersion);
printf("CUDA Driver Version: %d.%d\n", driverVersion/1000, (driverVersion%100)/10);
printf("CUDA Runtime Version: %d.%d\n\n", runtimeVersion/1000, (runtimeVersion%100)/10);
cudaError = cudaDeviceGetAttribute(&nppStreamCtx.nCudaDevAttrComputeCapabilityMajor,
cudaDevAttrComputeCapabilityMajor,
nppStreamCtx.nCudaDeviceId);
if (cudaError != cudaSuccess)
return NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY;
cudaError = cudaDeviceGetAttribute(&nppStreamCtx.nCudaDevAttrComputeCapabilityMinor,
cudaDevAttrComputeCapabilityMinor,
nppStreamCtx.nCudaDeviceId);
if (cudaError != cudaSuccess)
return NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY;
cudaError = cudaStreamGetFlags(nppStreamCtx.hStream, &nppStreamCtx.nStreamFlags);
cudaDeviceProp oDeviceProperties;
cudaError = cudaGetDeviceProperties(&oDeviceProperties, nppStreamCtx.nCudaDeviceId);
nppStreamCtx.nMultiProcessorCount = oDeviceProperties.multiProcessorCount;
nppStreamCtx.nMaxThreadsPerMultiProcessor = oDeviceProperties.maxThreadsPerMultiProcessor;
nppStreamCtx.nMaxThreadsPerBlock = oDeviceProperties.maxThreadsPerBlock;
nppStreamCtx.nSharedMemPerBlock = oDeviceProperties.sharedMemPerBlock;
if (checkCmdLineFlag(argc, (const char **)argv, "input")) {
getCmdLineArgumentString(argc, (const char **)argv, "input", &filePath);
} else {
@@ -190,11 +210,11 @@ int main(int argc, char *argv[]) {
Npp16s nHighThreshold = 256;
if ((nBufferSize > 0) && (pScratchBufferNPP != 0)) {
NPP_CHECK_NPP(nppiFilterCannyBorder_8u_C1R(
NPP_CHECK_NPP(nppiFilterCannyBorder_8u_C1R_Ctx(
oDeviceSrc.data(), oDeviceSrc.pitch(), oSrcSize, oSrcOffset,
oDeviceDst.data(), oDeviceDst.pitch(), oSizeROI, NPP_FILTER_SOBEL,
NPP_MASK_SIZE_3_X_3, nLowThreshold, nHighThreshold, nppiNormL2,
NPP_BORDER_REPLICATE, pScratchBufferNPP));
NPP_BORDER_REPLICATE, pScratchBufferNPP, nppStreamCtx));
}
// free scratch buffer memory