[PyCuda] async memcpy

[Nicholas Tung]

On Wed, Mar 11, 2009 at 18:55, Andreas Klöckner <lists at informa.tiker.net>wrote:

> On Mittwoch 11 März 2009, Nicholas Tung wrote:
> > On Wed, Mar 11, 2009 at 16:56, Andreas Klöckner
> <lists at informa.tiker.net>wrote:
> > > On Mittwoch 11 März 2009, you wrote:
> > > > Hi Andreas,
> > > >
> > > >     For asynchronous memory copies, what do I use to load data to a
> > > > page-locked array?
> > > >
> > > > Does page_locked[:] = original[:] work?
> > >
> > > That should do it. Even if it comes out pagelocked_something(), it's
> just
> > > a plain old numpy array, just in special memory.
> > >
> > > Andreas
> >
> > Thanks. However, when I try to do any GPU operations on separate threads,
> I
> > get [e.g.]
> > "LogicError: cuStreamCreate failed: invalid context"
> >
> > Is this avoidable?
>
> Each thread needs a separate context created. See
> examples/multiple_threads.py
> in (very) recent git.
>
> Andreas
>

Thanks so much, but can I not pass memory between threads?

"LogicError: cuMemcpyHtoD failed: invalid value"

Thanks,
Nicholas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tiker.net/pipermail/pycuda/attachments/20090311/f9ea48bc/attachment-0001.htm>