Yes, exactly this.
From my understanding, there isn't anything pyopencl would do to
different codepaths for differend gfx cards, is that correct?
And aside, my Nvidia driver is 331, not 334 as previously stated (I
mixed it up with the 304 alternative); and I am using Python 2.7.
How could I proceed with this? From looking at pyopencl/__init__.py
build(), it all looks very normal and not-forking-per-card.
It would also be better if someone could reproduce this bug -- it
completely baffles me, and even though I tried it with a second Ti780
(failed also) and my older card and different computers with Titan Black
and GTX480, respectively, it is still a mysterious one-off.
On 14.04.2014 02:04, Andreas Kloeckner wrote:
Kai Ruhl <kain(a)land-of-kain.de> writes:
> thanks for the quick response -- I also tested it on a Titan Black
> which gives correct results. It is *just* the Ti780. I use the x-swat
> nvidia drivers 334 for Ubuntu 14.04.
> In the meantime, I used the AMD SDK lib to compile pyopencl with
> OCL1.2, still the same error.
As in, AMD OpenCL gives you 88.0 as well?