The package compyte  contain a new gpu nd array that is in
development. Currently there is indexing with int, slice and tuple
with a mix of int and slice that are supported. Those create gpu nd
array that have in some case not contiguous memory. No code in pycuda
accept that, but you can make a copy of the gpu ndarray to use it with
pycuda code. There is no direct conversion from a gpu nd array to
The package have a bad installation system. Look into the Makefile and
change it to your platform.
Currently we don't support numpy "advanced indexing" or list of int
in the __getitem__ function.
On Sun, May 29, 2011 at 9:54 AM, Andreas Kloeckner
On Sun, 29 May 2011 16:08:28 +1000, <Brett.Bryan(a)csiro.au>
> Just wondering if there are plans/timelines for implementing the Numpy
> multidimensional and fancy indexing/slicing in PyCUDA? We use this
> feature a lot to do repetitive data processing. There is not much of a
> performance improvement from indexing on CPU then passing to GPU,
> processing, then passing back. It seems to me that passing a single
> multidimensional array to GPU memory then indexing/slicing and
> processing entirely on the GPU would provide substantially greater
> performance improvement?
This is being worked on--see the recent announcement regarding the
'gpundarray' list. No precise time line, but potentially soon--faster if
you help. :)
PyCUDA mailing list