[Beowulf] bring back 2012?
pbisbal at pppl.gov
Fri Aug 19 07:30:19 PDT 2016
I'm not going to dispute your claims, but I think you're probably a
statistical anomaly in this regard. I'm guessing your code isn't that
old, so it was built using current 'best practices' for parallelization
(loop independence, data parallelism, overlapping communication and
computation, etc) for the get-go, or was at least kept 'modern' which
made it easy for the compilers to optimize for the Phis.
Can you comment on this?
On 08/18/2016 09:28 PM, Stu Midgley wrote:
> we get EXCELLENT performance from the phi's... just look at our
> marketing with Intel.
> The 72XX series are also superb.
> On Fri, Aug 19, 2016 at 7:41 AM, Christopher Samuel
> <samuel at unimelb.edu.au <mailto:samuel at unimelb.edu.au>> wrote:
> On 18/08/16 00:10, Prentice Bisbal wrote:
> > I wouldn't be so glad. You're still going to have to rewrite
> your code
> > as mentioned above to get any meaningful performance.
> To be fair to Stu though he did post back in 2013 that he got good
> performance improvements on his codes on Xeon Phi for little initial
> effort, so that's why they're so attractive to him.
> To us though they've proved not that useful, we can only run them
> in accelerator mode (due to external constraints) and so we've had
> virtually 0 uptake. :-(
> All the best,
> Christopher Samuel Senior Systems Administrator
> VLSCI - Victorian Life Sciences Computation Initiative
> Email: samuel at unimelb.edu.au <mailto:samuel at unimelb.edu.au>
> Phone: +61 (0)3 903 55545 <tel:%2B61%20%280%293%20903%2055545>
> http://www.vlsci.org.au/ http://twitter.com/vlsci
> Beowulf mailing list, Beowulf at beowulf.org
> <mailto:Beowulf at beowulf.org> sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit
> Dr Stuart Midgley
> sdm900 at sdm900.com <mailto:sdm900 at sdm900.com>
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Beowulf