On Mon, 21 Mar 2011 20:20:39 +0100, Magnus Paulsson <paulsso...@gmail.com> wrote: > The visual profiler shows overlaping mem-copies and execution for the > Working.py. You are probably staring at your computer so if you are in > doubt, try it :D > > (and this was one of my original questions ... how do you profile the > code if the profiler is obviously broken?)
Wrt the profiler--I guess then the best answer I can provide is "I don't know". What I'd do is run the thing in a loop (the first run through a set of kernels on CUDA often (and inexplicably) results in weird timings) and see which is faster. I'm guessing I wouldn't rely too much on the profiler. HTH, Andreas
pgpjTgnzEGDVM.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda
