Hi Daniel, Thanks for contacting me regarding PyCUDA. I would like to request that in the future you use the mailing list for such discussions, though.
On Sun, 07 Nov 2010 21:57:49 -0000, "Daniel White" <[email protected]> wrote: > I'd like to create an application which allows the user to define an > arbitrary function, > through a GUI and the result is built and then ray-traced to the screen. > It has to be > as fast as possible, so I was looking at using CUDA with NVidia's C/C++ > libraries directly. > > However, the CUDA forums pointed me to PyCUDA which really seems very > interesting > as it supports metaprogramming (and so therefore would seem to allow for > user generated > code to run on the fly, right?). Correct. > One more thing I think I heard was that PyCUDA uses nvcc to compile > anyway, so perhaps > it's not as 'interpreted' as I think it is? No, running an interpreter on the GPU would not be the right approach IMO. Instead, one of the main purposes of PyCUDA is to make (C) code generation simple. > Do you think NVidia would allow me to distribute nvcc with my program > (which may be > partially close-sourced for now, and could be commercial). This may make > PyCUDA less > necessary (although the feature-set of PyCUDA, such as the runtime > optimization does > look enticing in any case). I'm prepared to pay them (and you if I end up > using PyCUDA) > something, though I'm not rich yet :) If you're concerned about compiler availability, why don't you check out OpenCL? There, the compiler is part of the API and is guaranteed to be available at run time, without further distribution worries and licensing charges. Hope this helps, Andreas > Regards, Daniel > http://www.skytopia.com > >
pgpHiExpVcYP8.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list [email protected] http://lists.tiker.net/listinfo/pycuda
