[Rpm-ecosystem] Requires based on destination hardware

Igor Gnatenko i.gnatenko.brain at gmail.com
Sat Dec 12 18:19:41 UTC 2015


It is already implemented, but I'm thinking how to make better dependencies
on opencl realizations. For example, we have program which uses OpenCL, but
how we can provide to user one of implementations of OpenCL? First idea
which comes to my mind to make Provides: opencl to all implementations and
Requires: opencl in this program, but it is not looks very good..

On Sat, Dec 12, 2015 at 5:52 PM Zbigniew Jędrzejewski-Szmek <
zbyszek at in.waw.pl> wrote:

> On Sat, Dec 12, 2015 at 01:54:12PM +0000, Igor Gnatenko wrote:
> > Hi all,
> >
> > With OpenCL currently we have some implementations:
> > * Beignet (Intel GPU)
> > * POCL (Any CPU)
> > * mesa-LibOpenCL (AMD GPU, probably something supported fro NVIDIA GPU)
> >
> > I am thinking how to make it user-friendly. Any ideas?
>
> Detect if the hardware is unsupported and fail nicely, printing an
> clear message?
>
> Please don't try to encode the hardware support in packaging —
> it is not flexible enough, and even if it was, installations should be
> tied to particular hardware.
>
> Zbyszek
> _______________________________________________
> Rpm-ecosystem mailing list
> Rpm-ecosystem at lists.rpm.org
> http://lists.rpm.org/mailman/listinfo/rpm-ecosystem
>
-- 

-Igor Gnatenko
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rpm.org/pipermail/rpm-ecosystem/attachments/20151212/dd7a6ad5/attachment.html>


More information about the Rpm-ecosystem mailing list