I remember the pricing to be a lot lower, but now seems to be $2000 (eee) (Unless non-profit and that's $800) http://spec.org/order.html
The ACCEL benchmarks do validation on a lot of the kernels, but not 100% - so it may be suitable if you don't mind spending real money. Feel free to email me offline if you need a hand setting it up. If you don't mind something small and quick-n-dirty I may be willing to try to help figure out something.. On Wed, Aug 31, 2016 at 9:07 PM, Michael Di Domenico <[email protected]> wrote: > On Wed, Aug 31, 2016 at 8:53 AM, John Hearns <[email protected]> wrote: >> I guess you have looked at gpu-burn >> https://github.com/Microway/gpu-burn > > i have not, i'll add it to my bag. but a cursory look at the code, it > doesn't jump out at me how this actually checks the correct answers. > > there are several check calls that examine the CUDA return codes, but > I'm worried what I actually have is akin to silent data corruption, > which this might not catch > _______________________________________________ > Beowulf mailing list, [email protected] sponsored by Penguin Computing > To change your subscription (digest mode or unsubscribe) visit > http://www.beowulf.org/mailman/listinfo/beowulf _______________________________________________ Beowulf mailing list, [email protected] sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
