On Thu, 8 Dec 2011 00:30:44 -0500, Wes McKinney <[email protected]> wrote: > On Thu, Dec 8, 2011 at 12:15 AM, Andreas Kloeckner > <[email protected]> wrote: > > Hi Patric, > > > > On Thu, 31 Mar 2011 09:28:40 +0200, Patric Holmvall <[email protected]> > > wrote: > >> Of course! That's my opinion too. As I said though, I'm probably not > >> experienced enough to make it into a patch. I guess you have too much on > >> your plate to fix it at the moment? > >> By the way, we found a bug in the current PRNG. The mean value and variance > >> decays to very low values after a while. All the more reason to get it > >> updated. You'll probably get more info about it from my colleague soon. > >> > >> Maybe we could make a request or something on the mailing list? There ought > >> to be a demand for a better PRNG. > > > > I just found this email at the bottom of my inbox--just wanted to make > > sure you know that PyOpenCL 2011.2 (just released) comes with RANLUXCL. > > > > Andreas > > > > _______________________________________________ > > PyOpenCL mailing list > > [email protected] > > http://lists.tiker.net/listinfo/pyopencl > > > > Any of you guys looked at Random123 yet? > > http://www.deshawresearch.com/resources_random123.html > > I saw a talk about it at SC11 this year and it looked very promising > for a CUDA/OpenCL PRNG
Haven't, but thanks for the pointer! Andreas
pgpOddzq1bGgh.pgp
Description: PGP signature
_______________________________________________ PyOpenCL mailing list [email protected] http://lists.tiker.net/listinfo/pyopencl
