We’ve posted an initial version of a CUDA frequently asked questions list here:
[url=“http://forums.nvidia.com/index.php?showtopic=36286”]http://forums.nvidia.com/index.php?showtopic=36286[/url]
Please check this before posting on the forums, and let us know about any necessary additions or corrections. Thanks!
I would like to see a section about large integers. What happens when I use __int64 in CUDA? Will it need to fetch the int from memory in two stages?
Current GPUs are basically 32-bit machines, so operating on 64-bit or larger values requires multiple instructions.
It is possible to load 128-bits from memory in a single instruction.
Thanks for your comment.
When I want to do a bitwise operation (e.g. AND two 64 bit-integers) on a 64 bit integer the compiler will handle that for me?