I am using Allwinner processor without any GPU acceleration. The problem is that remap() funtion takes about 20ms to remap 640x480 3 bytes image. It is too much for me. I compiled with multiprocessing flag (MP) and tried to write me own piece of code. Any ideas how to speed up this function or what can I make on (ORANGE-PI H3) to get some GPU acceleration of this function?
Asked
Active
Viewed 1,291 times
1
-
What kind of GPU does your board have? Does it support OpenCL? – zindarod Oct 07 '17 at 19:08
-
Mali 400, it does not support any acceleration suitabled for OpenCV – RobertLT Oct 07 '17 at 20:28
-
Then you cannot give OpenCV GPU support if your on board chip does not support OpenCL/CUDA. – zindarod Oct 07 '17 at 20:56
-
Please share your own code - there are some smart folks on StackOverflow... – Mark Setchell Oct 07 '17 at 20:56
-
Also, try tagging your question with `SIMD` and `NEON`. – Mark Setchell Oct 07 '17 at 21:01