Print

Print


hi

I used the tables in this article to relate performance of the new Intel  processors with processors all the way back to X5650.  From that I can estimate a hepspec value. 

https://www.nextplatform.com/2017/09/01/huge-premium-intel-charging-skylake-xeons/

The estimated results I'm using are

Intel silver 4116, QMspec = 514 
Intel gold 5118, QMspec = 557 
Intel gold 6126, QMspec=628 

For AMd I had to look at a range of benchmarks comparing AMD to Intel and then guess a value for HepSpec, these are more guess work then estimates.

AMD 7301, QMspec=616 
AMD 7351, QMspec=668 

Disclaimer: Note theses are guesstimates so I've called them QMspec rather than HepSPEC ;)

* Dr Daniel Traynor, Grid cluster system manager
* Tel +44(0)20 7882 6560, Particle Physics,QMUL

________________________________________
From: The UK High Energy Physics GRID List <[log in to unmask]> on behalf of Paul Kyberd <[log in to unmask]>
Sent: 25 January 2018 11:15
To: [log in to unmask]
Subject: Re: CPU comparisons/benchmarks

Ivan,

Here is the specfp2006 on the 6126. I can try tracking down the others

The CERN benchmarks for meltdown after the fixes are only showing a 2-4%
slow down, but maybe it gets worse when a set of reliable fixes are available.

Also not sure how dangerous the bug is for a set of machines sitting behind
a firewall and accessible through a server machine. What is Raul's opinion

Paul

´╗┐On 25/01/2018, 10:55, "The UK High Energy Physics GRID List on behalf of Dr Ivan D. Reid" <[log in to unmask] on behalf of [log in to unmask]> wrote:

    Hi;
        We're evaluating some offers for new machines, but the CPUs being
    offered are new enough that I'm having trouble finding any comparisons or
    benchmarks, let alone HEPSPECs.  Does anyone know of any comparisons between
    these compute CPUs?

    Xeon 6126, Xeon 6130, Xeon E5-2683v4, EPYC 7501

        Also, is there any consensus on the desirability or otherwise of
    shunning Intel to avoid the Meltdown slowdown?

        Thanks,
                ivan

    --
                       Ivan Reid (ivan.reid@[brunel.ac.uk|cern.ch])
    Engineering, Design & Physical Sciences                CMS Collaboration,
    Brunel University London. Room TOWD405                 CERN, Room 40-1-B12