On Tue, 10 Apr 2012 14:06:19 +0200, Christian <mining.fa...@googlemail.com> wrote: > Hi, > > I'm more or less new in multiprocessing issues. I'm wondering if it > might be worth considering pycuda to accelerate the comparison's > of a huge amount string tuple's ( levenshtein ,jaro-winkler etc..)?
the question you really want answered is whether GPUs can be of help in such a task. If you decide that this is the case, then PyCUDA can provide an easy vehicle for this work. Unfortunately, I don't know much about your problem--but whether you do or don't use PyCUDA likely doesn't factor into the answer. (That said, if what you have is a large number of Python strings (as opposed to a less granular format--say one string with indices, or a 2D numpy character array), then collecting them and copying them to the device might already too big of a penalty to pay.) HTH, Andreas
pgpYs6yvMNXl2.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda