Yeap. I tried that. My problem was the same on this post http://forums.nvidia.com/index.php?showtopic=54599
I used the timing functions from CUT instead of using cudaEventRecord etc.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| How to get peak rate with simple opeartion Question about performance optimization | 17 | 13686 | June 2, 2008 | |
| multiple texture fetch bandwidth | 1 | 5199 | April 21, 2008 | |
| texture memory vs global memory | 10 | 13869 | August 20, 2007 | |
| Reading data | 12 | 2737 | July 18, 2011 | |
| Benefits of Texture Memory couldnt use them... | 6 | 3241 | February 13, 2008 | |
| CUDA vs. OpenGL textures read-only vs. read-write | 5 | 10981 | March 30, 2007 | |
| How to do this (and be fast) | 7 | 7911 | February 9, 2008 | |
| About texture cache and spatial locality | 15 | 11320 | July 24, 2009 | |
| Texture? Just a short lesson... | 5 | 2748 | March 9, 2008 | |
| Matrix multiplication using texture | 6 | 4873 | April 17, 2008 |