2016-09-06 02:43:14 -0600 | received badge | ● Student (source) |
2016-09-05 23:38:07 -0600 | asked a question | GPU cv::scaleAdd is slower than CPU cv::cuda::scaleAdd Hi, I used the following code to compare execution time between CPU and GPU versions of scaleAdd. I am in Ubuntu 16.04, cuda 8.0, 740M card, and core i7 5th gen in my laptop. However, the results I get is that the CPU version is always faster. Could you execute the code and provide me your results with machine configuration? I compile with |
2016-08-06 18:52:57 -0600 | received badge | ● Editor (source) |
2016-08-06 03:55:25 -0600 | asked a question | cudacodec::VideoReader returns 4 channels Hi, the regular video reader returns 3 channels (BGR) but the gpu version returns 4 channels for the same video. Is there any documentation that could help me understand what format is it returning them? edit, I am on ubuntu 16.04, cuda 7.5, geforce 740M edit2: Further testing revealed that for a 1440x1080 video, it reads a frame of 1440x1088. It adds 8 rows to the bottom. |