NVIDIA
PBA Distance Transform using NPP returns non-exact EDT.
Related to this topic I followed the examples at https://github.com/NVIDIA/CUDALibrarySamples/tree/master/NPP/distanceTransform In these examples, I set the first row a sites.
// nppiDistanceTransformPBA_8u32f_C1R_Ctx
[0]: 0 0 0 0 0 0 0 0 0 0
[1]: 1 1 1 1 1 1 1 1 1 1
[2]: 1 1 1 1 1 1 1 1 1 1
[3]: 3 3 3 3 3 3 3 3 3 3
[4]: 3 3 3 3 3 3 3 3 3 3
[5]: 4 4 4 4 4 4 4 4 4 4
[6]: 5 5 5 5 5 5 5 5 5 5
[7]: 6 6 6 6 6 6 6 6 6 6
[8]: 7 7 7 7 7 7 7 7 7 7
[9]: 8 8 8 8 8 8 8 8 8 8
[10]: 9 9 9 9 9 9 9 9 9 9
// nppiDistanceTransformAbsPBA_8u16u_C1R_Ctx
[0]: 0 0 0 0 0 0 0 0 0 0
[1]: 1 1 1 1 1 1 1 1 1 1
[2]: 1 1 1 1 1 1 1 1 1 1
[3]: 3 3 3 3 3 3 3 3 3 3
[4]: 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5
[5]: 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5
[6]: 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5
[7]: 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5
[8]: 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5
[9]: 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5
[10]: 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5
Hi @jeanchristopheruel,
Thank you for reporting this issue. Could you please provide a small reproducible example along with details of your testing environment, including information about the GPUs, CUDA Toolkit version, and any other relevant configurations?
Hey @mkhadatare I'm seeing the exact same issue on one system, but not on another. I am using nppiDistanceTransformPBA_8u32f_C1R_Ctx.
System with issue: RTX 4070 CUDA/NPP version: 11.8
System without issue: Jetson Orin Nano 4GB CUDA/NPP version: 11.4
The nature of the issue is exactly as demonstrated in @jeanchristopheruel's post. Locations which should have distances of 2 are reported to have distances of 1. This appears to right itself at distance 3. Then at positions which should have distance 4, we see distance 3.5. Then the rate of the distance increments correctly, though the values themselves are incorrect. (3.5, 4.5, 5.5 where it should be 4, 5, 6)
Hi @soooch Thank you for reporting this. We are currently working on the issue under bug tracker ID 4832970 and will keep you updated on the progress.
Hi we originally did not have this issue on Jetson Orin Nano 4GB. However, since upgrading to Jetpack 6 with Cuda NPP version 12.3.1 we seem to be observing this same problem now on Jetson. Here's a sample program
#include <stdio.h>
#include <stdlib.h>
#include "cuda_runtime.h"
#include "nppdefs.h"
#include "nppcore.h"
#include "nppi_filtering_functions.h"
#include "npps_initialization.h"
#define min(x, y) (((x) < (y)) ? (x) : (y))
#define nImageWidth 64
#define nImageHeight 64
int test_npp()
{
NppStreamContext nppStreamCtx;
nppStreamCtx.hStream = 0;
cudaGetDevice(&nppStreamCtx.nCudaDeviceId);
cudaDeviceGetAttribute(&nppStreamCtx.nCudaDevAttrComputeCapabilityMajor,
cudaDevAttrComputeCapabilityMajor,
nppStreamCtx.nCudaDeviceId);
cudaDeviceGetAttribute(&nppStreamCtx.nCudaDevAttrComputeCapabilityMinor,
cudaDevAttrComputeCapabilityMinor,
nppStreamCtx.nCudaDeviceId);
cudaStreamGetFlags(nppStreamCtx.hStream, &nppStreamCtx.nStreamFlags);
struct cudaDeviceProp oDeviceProperties;
cudaGetDeviceProperties(&oDeviceProperties, nppStreamCtx.nCudaDeviceId);
nppStreamCtx.nMultiProcessorCount = oDeviceProperties.multiProcessorCount;
nppStreamCtx.nMaxThreadsPerMultiProcessor = oDeviceProperties.maxThreadsPerMultiProcessor;
nppStreamCtx.nMaxThreadsPerBlock = oDeviceProperties.maxThreadsPerBlock;
nppStreamCtx.nSharedMemPerBlock = oDeviceProperties.sharedMemPerBlock;
NppiSize oImageSizeROI = {nImageWidth, nImageHeight};
Npp8u hostInputBuffer[nImageHeight * nImageWidth] = {0};
for (int i = 0; i < nImageWidth; i++) {
// Fill first row with 1s
hostInputBuffer[i] = 1;
}
printf("Input array (%dx%d):\n", nImageWidth, nImageHeight);
if (nImageWidth != 8) {
printf("Truncating printed width to %d\n", 8);
}
for (int y = 0; y < nImageHeight; y++) {
for (int x = 0; x < min(8, nImageWidth); x++) {
printf("%d ", hostInputBuffer[y * nImageWidth + x]);
}
printf("\n");
}
Npp8u *pInputImage_Device = 0;
Npp32f *pOutputImage_Device = 0;
Npp8u *pScratchBuffer = 0;
Npp32f *hostOutputBuffer = 0;
size_t nScratchBufferSize;
nppiDistanceTransformPBAGetBufferSize(oImageSizeROI, &nScratchBufferSize);
printf("Scratch buffer size: %zu bytes\n", nScratchBufferSize);
cudaMalloc((void **)&pScratchBuffer, nScratchBufferSize);
cudaMalloc((void **)&pInputImage_Device, oImageSizeROI.width * sizeof(Npp8u) * oImageSizeROI.height);
cudaMalloc((void **)&pOutputImage_Device, oImageSizeROI.width * sizeof(Npp32f) * oImageSizeROI.height);
hostOutputBuffer = (Npp32f *)malloc(oImageSizeROI.width * sizeof(Npp32f) * oImageSizeROI.height);
if (pScratchBuffer == 0 || pInputImage_Device == 0 || pOutputImage_Device == 0 || hostOutputBuffer == 0)
{
printf("Memory allocation failed\n");
if (pScratchBuffer)
cudaFree(pScratchBuffer);
if (pInputImage_Device)
cudaFree(pInputImage_Device);
if (pOutputImage_Device)
cudaFree(pOutputImage_Device);
if (hostOutputBuffer)
free(hostOutputBuffer);
return -1;
}
cudaMemcpy(pInputImage_Device, hostInputBuffer,
oImageSizeROI.width * oImageSizeROI.height * sizeof(Npp8u),
cudaMemcpyHostToDevice);
Npp8u nMinSiteValue = 1;
Npp8u nMaxSiteValue = 1;
printf("Running distance transform...\n");
NppStatus status = nppiDistanceTransformPBA_8u32f_C1R_Ctx(
pInputImage_Device, oImageSizeROI.width * sizeof(Npp8u),
nMinSiteValue, nMaxSiteValue,
0, 0,
0, 0,
0, 0,
pOutputImage_Device, oImageSizeROI.width * sizeof(Npp32f),
oImageSizeROI, pScratchBuffer, nppStreamCtx);
if (status != NPP_SUCCESS)
{
printf("Distance transform failed with error: %d\n", status);
}
else
{
printf("Distance transform completed successfully!\n");
cudaMemcpy(hostOutputBuffer, pOutputImage_Device,
oImageSizeROI.width * sizeof(Npp32f) * oImageSizeROI.height,
cudaMemcpyDeviceToHost);
if (nImageHeight != 8) {
printf("Truncating printed width to %d\n", 8);
}
for (int y = 0; y < nImageHeight; y++) {
for (int x = 0; x < min(8, nImageWidth); x++) {
printf("%.1f ", hostOutputBuffer[y * nImageWidth + x]);
}
printf("\n");
}
}
cudaDeviceSynchronize();
cudaFree(pScratchBuffer);
cudaFree(pInputImage_Device);
cudaFree(pOutputImage_Device);
free(hostOutputBuffer);
printf("Done!\n");
return 0;
}
int main() {
void *x = 0;
cudaMallocHost((void **)&x, 8);
if (x != 0) {
cudaFreeHost(x);
x = 0;
}
NppLibraryVersion * version = nppGetLibVersion();
printf("%d %d %d\n", version->major, version->minor, version->build);
test_npp();
}
as well as its output on jetson
12 3 1
Input array (64x64):
Truncating printed width to 8
1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
Scratch buffer size: 98840 bytes
Running distance transform...
Distance transform completed successfully!
Truncating printed width to 8
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0
3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5
4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5
5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5
6.5 6.5 6.5 6.5 6.5 6.5 6.5 6.5
7.5 7.5 7.5 7.5 7.5 7.5 7.5 7.5
8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5
9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5
10.5 10.5 10.5 10.5 10.5 10.5 10.5 10.5
11.5 11.5 11.5 11.5 11.5 11.5 11.5 11.5
12.5 12.5 12.5 12.5 12.5 12.5 12.5 12.5
13.5 13.5 13.5 13.5 13.5 13.5 13.5 13.5
14.5 14.5 14.5 14.5 14.5 14.5 14.5 14.5
15.5 15.5 15.5 15.5 15.5 15.5 15.5 15.5
16.5 16.5 16.5 16.5 16.5 16.5 16.5 16.5
17.5 17.5 17.5 17.5 17.5 17.5 17.5 17.5
18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5
19.5 19.5 19.5 19.5 19.5 19.5 19.5 19.5
20.5 20.5 20.5 20.5 20.5 20.5 20.5 20.5
21.5 21.5 21.5 21.5 21.5 21.5 21.5 21.5
22.5 22.5 22.5 22.5 22.5 22.5 22.5 22.5
23.5 23.5 23.5 23.5 23.5 23.5 23.5 23.5
24.5 24.5 24.5 24.5 24.5 24.5 24.5 24.5
25.5 25.5 25.5 25.5 25.5 25.5 25.5 25.5
26.5 26.5 26.5 26.5 26.5 26.5 26.5 26.5
27.5 27.5 27.5 27.5 27.5 27.5 27.5 27.5
28.5 28.5 28.5 28.5 28.5 28.5 28.5 28.5
29.5 29.5 29.5 29.5 29.5 29.5 29.5 29.5
30.5 30.5 30.5 30.5 30.5 30.5 30.5 30.5
31.5 31.5 31.5 31.5 31.5 31.5 31.5 31.5
32.5 32.5 32.5 32.5 32.5 32.5 32.5 32.5
33.5 33.5 33.5 33.5 33.5 33.5 33.5 33.5
34.5 34.5 34.5 34.5 34.5 34.5 34.5 34.5
35.5 35.5 35.5 35.5 35.5 35.5 35.5 35.5
36.5 36.5 36.5 36.5 36.5 36.5 36.5 36.5
37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5
38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5
39.5 39.5 39.5 39.5 39.5 39.5 39.5 39.5
40.5 40.5 40.5 40.5 40.5 40.5 40.5 40.5
41.5 41.5 41.5 41.5 41.5 41.5 41.5 41.5
42.5 42.5 42.5 42.5 42.5 42.5 42.5 42.5
43.5 43.5 43.5 43.5 43.5 43.5 43.5 43.5
44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5
45.5 45.5 45.5 45.5 45.5 45.5 45.5 45.5
46.5 46.5 46.5 46.5 46.5 46.5 46.5 46.5
47.5 47.5 47.5 47.5 47.5 47.5 47.5 47.5
48.5 48.5 48.5 48.5 48.5 48.5 48.5 48.5
49.5 49.5 49.5 49.5 49.5 49.5 49.5 49.5
50.5 50.5 50.5 50.5 50.5 50.5 50.5 50.5
51.5 51.5 51.5 51.5 51.5 51.5 51.5 51.5
52.5 52.5 52.5 52.5 52.5 52.5 52.5 52.5
53.5 53.5 53.5 53.5 53.5 53.5 53.5 53.5
54.5 54.5 54.5 54.5 54.5 54.5 54.5 54.5
55.5 55.5 55.5 55.5 55.5 55.5 55.5 55.5
56.5 56.5 56.5 56.5 56.5 56.5 56.5 56.5
57.5 57.5 57.5 57.5 57.5 57.5 57.5 57.5
58.5 58.5 58.5 58.5 58.5 58.5 58.5 58.5
59.5 59.5 59.5 59.5 59.5 59.5 59.5 59.5
60.5 60.5 60.5 60.5 60.5 60.5 60.5 60.5
61.5 61.5 61.5 61.5 61.5 61.5 61.5 61.5
62.5 62.5 62.5 62.5 62.5 62.5 62.5 62.5
Done!
@mkhadatare any updates on 4832970? Thanks so much for your help. Also @soooch and I work together which is why we're both observing the issue on Jetson
Sorry to spam this issue but we found that pDstVoronoiIndices is accurate. So we'll be using that instead of the pDstTransform output. Hope this helps someone
Hi, the issue has been resolved in the latest CUDA Toolkit 12.9. Please refer to the release notes for more details.