Hi. I found there is an instruction “bra.uni” in PTX besides “bra”, to mark a branching point is non-divergent. So is there some equivalent way to specify this in CUDA’s C/C++ programming language just like the “__builtin_expect()” in gcc for branching prediction? Thanks.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Bra.uni optimization Compiler optimization | 3 | 9164 | October 27, 2010 | |
| Branch Predication in CUDA Any example in SDK | 1 | 1333 | December 25, 2009 | |
| Branch or not | 7 | 2860 | February 28, 2018 | |
| CUDA compiler needs too much help in order to use select instead of branch | 6 | 566 | October 12, 2021 | |
| Is there any idiom to indicate that branches are uniform? | 0 | 481 | June 16, 2016 | |
| Branching Performance Hit | 15 | 2683 | June 30, 2009 | |
| about the divergent branching | 1 | 1201 | February 3, 2009 | |
| how to use the .uni suffix....?!? | 4 | 2774 | August 13, 2009 | |
| branchless exchange based on condition ? | 1 | 996 | February 9, 2009 | |
| PTX: "call.uni" vs "call" | 0 | 447 | December 17, 2020 |