Some sources say the texture cache is for throughput enhancing not latency reduction. The CUDA best practice guide says texture are cached, potentially exhibiting higher bandwidth if there is 2D locality in the texture feteches. If an application has non-coalesced global memory access, but the access only happened once for every element(no data-reuse). Does it mean that every texture fetch is actually a miss? Do we still have benefit(or a penalty) if we bound the global memory to a texture? Thank you.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| For what case should I use texture memory? | 8 | 2655 | May 26, 2010 | |
| When is it worth copying global to texture memory | 2 | 3361 | July 7, 2008 | |
| Texture vs Global Memory Bandwidth | 5 | 6560 | March 25, 2010 | |
| Benefits of Texture Memory couldnt use them... | 6 | 3199 | February 13, 2008 | |
| global memory texture memory and race conditions | 3 | 1064 | February 7, 2012 | |
| Texture memory when to use ? | 6 | 20698 | October 7, 2009 | |
| Higher bandwidth in texture memory compared to global memory | 1 | 479 | March 9, 2023 | |
| Performance Considerations using Texture Access Does the performance depend on the access pattern? | 1 | 1394 | August 21, 2009 | |
| no speedup with textures | 4 | 3722 | December 8, 2008 | |
| Texture and Global Memory | 2 | 3841 | July 11, 2007 |