Hi Alex, On Wed, 8 Feb 2012 19:10:57 -0500, Alex Nitz <alex.n...@ligo.org> wrote: > I'm new to using pycuda and I've run into the following issue. It seems > that the definition of __rmul__ in GPUArray asserts that the dtype is > float32. Is was wondering if there a reason for this? I would like to do > this operation with other types such as float64. > > (line 328 of gpuarray.py) > def _rdiv_scalar(self, other, out, stream=None): > """Divides an array by a scalar:: > > y = n / self > """ > > if not self.flags.forc: > raise RuntimeError("only contiguous arrays may " > "be used as arguments to this operation") > > assert self.dtype == np.float32 > > func = elementwise.get_rdivide_elwise_kernel(self.dtype) > func.prepared_async_call(self._grid, self._block, stream, > self.gpudata, other, > out.gpudata, self.mem_size)
Fixed in git. Thanks for the report. Andreas
pgpwXY1WxJW6o.pgp
Description: PGP signature
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda