On Fri, 29 Apr 2011 11:17:26 +0200, Jonas Einarsson <[email protected]> wrote: > Thanks for your answers, > > Christoph: I am slightly confused on the version of your supplied > binaries. The documentation (and source!) for 2011.1 regarding > prepared calls says that block size is deprecated in prepare and > should be supplied in prepared_call. Using the compiled version > requires me to use the old syntax, or I'll get a type error: > > kernel_advanceSystem.prepare("PP") gives > TypeError: function_prepare() takes at least 3 arguments (2 given) > > but > kernel_advanceSystem.prepare("PP", block=blockDim) > works. > > Either I am mistaken on the versions, or there might have been a > mix-up with the sources. > > Easy way to reproduce is to open examples/hello_gpu.py and add a line: > multiply_them.prepare("PPP")
There's no such version as 2011.1 yet--you're looking at two different development snapshots, and I take it Christoph's binary reflects a slightly older version from more than two weeks ago. Andreas
pgpIjZUtUtav8.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list [email protected] http://lists.tiker.net/listinfo/pycuda
