I read in an early paper by Volkov that it was like ~4 microseconds…
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| fundamental cuda kernel launch questions | 2 | 16534 | July 31, 2008 | |
| How big is the kernel invocation overhead? | 9 | 5094 | December 17, 2008 | |
| Does saturating a stream hide kernel launch latency? | 23 | 2697 | October 28, 2014 | |
| Kernel enqueue overhead Bringing kernel overhead down? | 9 | 13822 | March 12, 2010 | |
| Kernel design problem Performance difference in number of times a kernel is launched | 1 | 469 | January 9, 2012 | |
| Can too many kernel calls affect the performance ? | 4 | 1047 | June 10, 2010 | |
| kernel launch time expensive? | 2 | 1630 | July 28, 2009 | |
| Launch Overhead as a function of Kernel Size... Is it Proportional? Characterization? | 1 | 5370 | June 24, 2008 | |
| Kernel execution overhead | 2 | 1188 | July 6, 2009 | |
| Running One Thread per CUDA Core Or at least allow for highly divergent, register-intensive kernels | 11 | 4732 | December 1, 2011 |