Good afternoon colleagues, please tell me with OpenACC. I have 2 arrays that I want to fill as quickly as possible. I divide each array into a block-thread tensor in which a fast parallel calculation takes place. However, the two arrays themselves are most likely not filled at the same time. What are the ways to, for example, run these two arrays simultaneously, say, on 2 grids at once. And in general, is it possible to manage grids and their launch in OpenACC?
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Run CUDA and OpenCL kernels simultaneously | 1 | 1382 | February 7, 2017 | |
parallel computations with CUDA | 7 | 3193 | September 19, 2008 | |
CUDA processor allocation | 7 | 3437 | October 5, 2007 | |
is there any function to do sync threads in a grid? | 2 | 2425 | March 30, 2015 | |
How to use cuda core and tensor core simultaneously? | 4 | 619 | August 16, 2022 | |
cuda block synchronization | 1 | 982 | June 19, 2011 | |
Synchronize threads. | 1 | 698 | March 6, 2013 | |
how to use many graphic cards simultaneously | 3 | 5554 | August 25, 2009 | |
block vs grid vs tread the best way? how to set the pointer for each tread ? | 0 | 5011 | February 9, 2008 | |
Designing a CUDA algo question Sort of a newbie question.... | 2 | 2363 | December 9, 2011 |