“the cuda kernel does not allocate new memory or copy memory from host.”
what does the kernel do then, where does it get data from to do this, and what does it do with the result?
what are the grid/ block dimensions of the kernel?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Kernel takes more and more time at each iteration | 0 | 835 | May 24, 2011 | |
| kernel failed after few invokation | 9 | 7897 | October 30, 2010 | |
| Iterative computations Creating of efficient iterative computations using CUDA | 16 | 2479 | January 25, 2011 | |
| Memory copy very slow memory copy, image | 10 | 12637 | April 7, 2011 | |
| CUDA very slow performance | 21 | 17042 | March 6, 2020 | |
| cuda kernel call within for loop gets slow, crashes | 5 | 5891 | April 1, 2012 | |
| Getting Different Execution Times of Running Same Kernel Twice | 2 | 74 | August 13, 2024 | |
| CUDA trouble | 3 | 1036 | March 19, 2013 | |
| Same Kernel called multiple times in a loop has different runtimes | 2 | 735 | April 8, 2017 | |
| Decreased performances if CUDA kernels are not run continuously | 1 | 488 | June 8, 2018 |