You can use the profiler with the options file to see how many of your reads & writes are coalesced and how many are uncoalesced. It is very informative!
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Conditional Statement Divergent | 1 | 863 | June 2, 2011 | |
| Some Performance Consideration Questions warp divergence, coalescing and shared mem then and now | 1 | 2038 | March 8, 2012 | |
| reduction optimization #1 Not agree with performances explanation | 8 | 6758 | August 1, 2008 | |
| Optimization to Reduce Bank Conflicts Decreases Performance | 3 | 4037 | May 31, 2010 | |
| Some advice needed pls Doubts we have, we're starting with CUDA programming | 16 | 4799 | June 22, 2011 | |
| What should I optimize first? Divergence? Serialized Warps? | 4 | 7268 | April 7, 2009 | |
| CUDA basic discussions(Looping, branching) Why GPU faster than CPU when both use C? | 6 | 18762 | August 22, 2007 | |
| Warp branching | 11 | 10353 | October 26, 2010 | |
| Warp divergence branching inside loop | 11 | 7535 | August 20, 2008 | |
| Impact of control flow on thread performance | 11 | 14019 | January 17, 2008 |