#ServerSmart – Chat 2 Recap: High Performance Computing

ServerSmart - May 29 ImageStay ahead of the curve with #ServerSmart. The chat series continued on the blog’s chat page on Thursday, May 29, 2014 where the Intel® Xeon® and Xeon® Phi Team of Gary Paek, Mike Lafferty, Mike Donat, and Whitney Gross hosted the conversation about High Performance Computing.

Here are the discussion highlights:

(1)    Technical computing defined:

  • Mayura Garg: Can we start by clarifying what do we mean by ‘Technical Computing’?
  • Mike Lafferty: Technical Computing – Includes workstation and HPC clusters to solve engineering, simulation, design problems
  • Gary Paek: Technical Computing, by the way, is the superset that includes High Performance Computing.
    We have Technical Computing overview and training available, under NDA, for you: https://www-ssl.intel.com/p/s/en_US/techprovider/products/technical-computing

(2)    Gary Paek, Worldwide Channel Marketing Manager for Intel Xeon Phi Coprocessors, shares high performance computing resources:

(3)    Processor Performance:

  • Robert Whitmire: What’s the fastest mob/processor combination that’s all Intel?
  • Michael Donat: Hi Robert,  the S2600GZ and the E5 V2 CPUs are your best combinations that are all Intel for the general server arena
  • Whitney Gross: @Robert Whitmire – take a look at ark.intel.com and search on E5 v2 Xeon CPUs.  You’ll find CPUs like the E5-2667 v2, E5-2697 v2.
  • Peter Cole: Yep…Still need the best proc for the most performance, often, I pick the 2680v2.  just checking…
  • Whitney Gross: @Peter Cole – the advanced SKUs of the E5 2600 v2 family would typically be a university target.  E5-2697 v2 maximizes core count at 12/CPU.  E5-2697 v2 is top bin, there are other lower bins that affect frequency and/or core count.  ark.intel.com can help research.

(4)    Supercomputing speed:

  • Leonard Lamberg: Do you see speed increases in large supercomputers coming from faster individual CPUs or clusters of CPUs working together?
  • Michael Donat: So Supercomputing speed is a mix of CPU speed, clustering CPU speeds, and the speed of the internconnects. With high bandwidth internconnects, and streamlined parallel applications you can get fantastic Supercomputing speeds.
  • Leonard Lamberg: We work on single-purpose systems that go into long-lifecycle product lines.  Which Xeon processors are long (embedded) availability?
  • Virginia J Monte: @Leonard – E5-2609v2, E5-2648Lv2, E5-2680v2, E5-2658v2, E5-2630v2, E5-2640v2, E5-2628Lv2, E5-2618Lv2,

The Intel® Big Data and Intel® Internet of Things Team will host the next chat topic, Big Data and the Internet of Things on June 5, 1 p.m. EDT/ 10 a.m. PDT.

RSVP and participate in every #ServerSmart for a chance to win a one of two Dell* Venue 8 Pro Windows Tablets. Chats will run every Thursday in the month of June (June 5 – June 26).

Stay sharp and chat with you soon!

Mayura

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>