Probably because the integer opcode worked just fine and there was no point?
Hack of the day: Integer comparison can be used to compare positive IEEE floats. I need this bc there's no atomicMax() for floats in CUDA.pic.twitter.com/RMsQrfALEj
-
-
-
It gets a bit more complicated if you want to support negative values as well. Otherwise they could just provide the API without hw change.
End of conversation
New conversation -
-
-
I tried to prove this with ESBMC first, but apparently ESBMC is not using IEEE floats to represent float values. Which C model checker does?
-
Well. Any C checker will tell you your programs behavior is undefined.
- Show replies
New conversation -
-
-
Hacker's Delight has a whole chapter on such FP tricks.
#IEEE754Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
1. I don't see where you're computing the max :P! 2. Why are you converting to float anyway (something something strict aliasing something)?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.