- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
the ippiReduceBits_32f8u_C1R gives me different results on different machines:
Ipp32f input[] = { 0, 0.5, 0.75, 1 };
Ipp8u output[4];
Ipp8u buffer[4];
ippiReduceBits_32f8u_C1R(input, 16, output, 4, { 4, 1 }, 0, ippDitherNone, 255, buffer);
On one machine result is 0, 127, 190, 255
On second machine result is 0, 127, 191, 255
Is this a known bug (IPP Version 9.0)?
Best regards
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Steffen,
Depending on different system, there could be some small epsilon difference. Here is an article discusses the possible difference https://software.intel.com/en-us/articles/getting-reproducible-results-with-intel-mkl. Let us know if you have further question. Thanks!
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
but in this case this is no small epsilon I think.
The used formula here is: 0.75*255 which is 191,25. So, how can 190 be a valid result?
Regards Steffem

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page