New article: 8 socket commodity servers

Article: 8-Socket Commodity Servers: Flourish or Perish?
By: David Kanter (dkanter.delete@this.realworldtech.com), March 9, 2010 11:06 pm
Room: Moderated Discussions
anon (anon@anon.com) on 3/9/10 wrote:
---------------------------
>David Kanter (dkanter@realworldtech.com) on 3/9/10 wrote:
>---------------------------
>>Yesterday was an odd day in RWT history. We had three articles go up in a single day - probably a record.
>>
>>The first two were Russian translations of articles on Larrabee and Hot Chips.
>>
>>The third article is a biref look at 8 socket servers. This includes a bit of analysis
>>on why 8 socket servers were not successful previously, and some reasons why this may change.
>>
>>The article was first posted through exectweets here:
>>
>>http://exectweets.com/2010/03/08/eight-socket-servers-have-big-potential/
>>
>>
>>I subsequently put it online at:
>>http://www.realworldtech.com/page.cfm?ArticleID=RWT030810232808
>>
>>>
>>
>>Enjoy - I am hoping to return to this topic sooner rather than later.
>
>I think in a broadcast snoop cc system, then the bisection bandwidth of the snooping
>interconnect is more important than just the number of links to a node (although
>the latter can influence the former of course).

>Istanbul is rather crippled by its L3 cache size. 6MB >total, 1MB per core.

It's not just the size, but it's also the amount of snoop traffic which AMD's coherency protocol generates. Due to the exclusive nature of AMD's caches, it is quite frequent that cache misses are filled from memory (thus requiring a snoop), rather than being filled from an on-die cache. In Intel's architectures, if a line is present in any cache on the chip, it will never generate memory traffic for reads.

There's an excellent paper on this topic:
http://portal.acm.org/citation.cfm?id=1669165&dl=GUIDE&coll=GUIDE&CFID=81286494&CFTOKEN=18541064

>8 core
>Nehalem has 3x this much per core, and relatively more than other Nehalems -- not
>surprising as this is an *MP* server part. Cache just >really helps larger systems
>to scale on smaller interconnects. No prizes for guessing >why POWER and IPF CPUs have massive LLCs.
>
>I'm surprised you didn't mention it in your article.

Limited time and space...I do agree with your points.

David
< Previous Post in Thread 
TopicPosted ByDate
New article: 8 socket commodity serversDavid Kanter2010/03/09 10:27 AM
  New article: 8 socket commodity serversVincent Diepeveen2010/03/09 01:46 PM
    number of sockets is wrong metric (was: New article: 8 socket commodity servers)longtimelurker2010/03/14 05:13 AM
      number of sockets is wrong metric (was: New article: 8 socket commodity servers)EduardoS2010/03/14 05:34 AM
      number of sockets is wrong metric (was: New article: 8 socket commodity servers)Wes Felter2010/03/14 10:33 AM
        Magny-CoursMax2010/03/14 04:56 PM
          Magny-Coursanonymous2010/03/14 06:33 PM
            Magny-Courslongtimelurker2010/03/15 02:54 AM
      number of sockets is wrong metric (was: New article: 8 socket commodity servers)Vincent Diepeveen2010/03/14 11:31 AM
        number of sockets is wrong metric (was: New article: 8 socket commodity servers)longtimelurker2010/03/14 01:37 PM
          number of sockets is wrong metric (was: New article: 8 socket commodity servers)Vincent Diepeveen2010/03/15 11:36 AM
      number of sockets is wrong metric (was: New article: 8 socket commodity servers)David Kanter2010/03/14 11:56 AM
        Bad mathDavid Kanter2010/04/01 01:24 AM
      number of sockets is wrong metric (was: New article: 8 socket commodity servers)slacker2010/03/14 02:51 PM
        number of sockets is wrong metric (was: New article: 8 socket commodity servers)Michael S2010/03/15 05:05 AM
          number of sockets is wrong metric (was: New article: 8 socket commodity servers)slacker2010/03/15 01:02 PM
            Memory interfacesDavid Kanter2010/03/15 01:17 PM
              Memory interfacesslacker2010/03/15 09:08 PM
                Patents on tiny components vs. large, complex thingsmpx2010/03/15 11:41 PM
                  Patents on tiny components vs. large, complex thingsRichard Cownie2010/03/16 05:58 AM
                    Patents on tiny components vs. large, complex thingsMS2010/03/17 05:42 PM
                      Patents on tiny components vs. large, complex thingsa reader2010/03/18 08:45 PM
          Serial Port Memory TechnologyDavid Hess2010/03/21 03:32 AM
  New article: 8 socket commodity serversMichael S2010/03/09 03:13 PM
    New article: 8 socket commodity serverstheluketaylor2010/03/09 05:32 PM
    New article: 8 socket commodity serversJesper Frimann2010/03/09 10:35 PM
    New article: 8 socket commodity serversDavid Kanter2010/03/10 12:38 AM
      New article: 8 socket commodity serversTim2010/03/16 08:44 AM
  New article: 8 socket commodity serversanon2010/03/09 06:59 PM
    New article: 8 socket commodity serversDavid Kanter2010/03/09 11:06 PM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell green?