The problem is that I don't want an array of random numbers in the end
(as your code does), but want to pass generators to a cuda function
that then simulates a stochastic process. So I need access to the
curandState.

On Fri, Dec 9, 2011 at 5:38 AM, Tomasz Rybak <[email protected]> wrote:
> Excuse me stupid question, but why are you reimplementing
> XORWOWRandomNumberGenerator instead of using it?
>
> The code should look like this:
>
> import numpy
>
> # This initializes random number generator
> r = pycuda.curandom.XORWOWRandomNumberGenerator()
> # This will create GPUArray of 32-bit floats with
> # sizes 100x200x300 and fill it with normalized random
> # numbers
> array = r.gen_normal((100, 200, 300), dtype = numpy.float32)
> # Or, if you already have array...
> array = pycuda.gpuarray.GPUArray((200, 400), dtype = numpy.float32)
> r.fill_normal(array)
>
> It will use as many cores as you have, check for errors, etc.
>
> Best regards
>
>
>
> --
> Tomasz Rybak <[email protected]> GPG/PGP key ID: 2AD5 9860
> Fingerprint A481 824E 7DD3 9C0E C40A  488E C654 FB33 2AD5 9860
> http://member.acm.org/~tomaszrybak

_______________________________________________
PyCUDA mailing list
[email protected]
http://lists.tiker.net/listinfo/pycuda

Reply via email to