Possible 4-way Woodcrest config?

2»

Comments

  • Reply 21 of 38
    Quote:

    Originally posted by backtomac

    You might want to check out this link below. Dual core conroe was beating quad core opteron systems. Even without the on die memory controller looks like core will outperform AMD.



    http://www.digitalvideoediting.com/a...jsp?id=53163-1




    4 cpu intel systems will hit the FSB wall.
     0Likes 0Dislikes 0Informatives
  • Reply 22 of 38
    hmurchisonhmurchison Posts: 12,464member
    Quote:

    Originally posted by Joe_the_dragon

    4 cpu intel systems will hit the FSB wall.



    Joe it's about time you stop making stuff up. There is no FSB Wall.
     0Likes 0Dislikes 0Informatives
  • Reply 23 of 38
    Quote:

    Originally posted by hmurchison

    Joe it's about time you stop making stuff up. There is no FSB Wall.



    I would like to see Intel try to make 4 quad-core cups beat an amd system with the same thing.



    Intel up coming quad-core are doing the same thing as there first duel cores.
     0Likes 0Dislikes 0Informatives
  • Reply 24 of 38
    backtomacbacktomac Posts: 4,579member
    Quote:

    Originally posted by Joe_the_dragon

    I would like to see Intel try to make 4 quad-core cups beat an amd system with the same thing.









    Did you see the link I posted? Current dual core 2.9ghz conroe systems are already beating quad core opteron systems in many tests. Do you think the Intel quad cores are going to be slower? If so give a legit reason.
     0Likes 0Dislikes 0Informatives
  • Reply 25 of 38
    zandroszandros Posts: 537member
    Quote:
    Originally Posted by hmurchison


    Joe it's about time you stop making stuff up. There is no FSB Wall.



    There is a FSB wall, but not the kind he is speaking of. The setting where you can't increase the FSB more on a motherboard is generally referred to as a FSB wall.



    @ mwswami: Are those lineups for MWSF? Otherwise they aren't realistic at all.
     0Likes 0Dislikes 0Informatives
  • Reply 26 of 38
    Quote:
    Originally Posted by backtomac


    Did you see the link I posted? Current dual core 2.9ghz conroe systems are already beating quad core opteron systems in many tests. Do you think the Intel quad cores are going to be slower? If so give a legit reason.



    amd beat intel in CineBench Rendering and Night Flight

    also they a testing a new intel chip to the old opteron the new ones will add ddr 2 and more ht links and k8l is comeing soon after that.

    and the intel system had a better raid system in it.
     0Likes 0Dislikes 0Informatives
  • Reply 27 of 38
    mwswamimwswami Posts: 166member
    Quote:
    Originally Posted by Zandros


    There is a FSB wall, but not the kind he is speaking of. The setting where you can't increase the FSB more on a motherboard is generally referred to as a FSB wall.



    @ mwswami: Are those lineups for MWSF? Otherwise they aren't realistic at all.



    If you read carefully, it says that the first two machines are coming at WWDC and the last, presuming Intel comes out with Clovertown by the end of this year, will come out at Mac World SF '07.
     0Likes 0Dislikes 0Informatives
  • Reply 28 of 38
    backtomacbacktomac Posts: 4,579member
    Quote:
    Originally Posted by Joe_the_dragon


    amd beat intel in CineBench Rendering and Night Flight

    also they a testing a new intel chip to the old opteron the new ones will add ddr 2 and more ht links and k8l is comeing soon after that.

    and the intel system had a better raid system in it.



    I hope I understand you correctly. The quad core opterons win two tests, and are getting the ddr memory and you think that will be enough to hold off quad core intel cpus. Are you not impressed with the conroe performance? I reiterate, a 2.9 ghz conroe outperformed a 2.4 ghz quad core opteron system in many tests. A quad core Kentsfield and Cloverton will creme AMD unless they significantly improve their architecture. ODMC aint gonna be enough.
     0Likes 0Dislikes 0Informatives
  • Reply 29 of 38
    Quote:
    Originally Posted by backtomac


    I hope I understand you correctly. The quad core opterons win two tests, and are getting the ddr memory and you think that will be enough to hold off quad core intel cpus. Are you not impressed with the conroe performance? I reiterate, a 2.9 ghz conroe outperformed a 2.4 ghz quad core opteron system in many tests. A quad core Kentsfield and Cloverton will creme AMD unless they significantly improve their architecture. ODMC aint gonna be enough.



    The systems in that test used different hard driver systems and It does not say if all of the systems used the same video card.



    Also was the software useing all of the cores in the quad-core system.
     0Likes 0Dislikes 0Informatives
  • Reply 30 of 38
    hmurchisonhmurchison Posts: 12,464member
    I think what I'm seeing here is the promotion of AMD chips as being fast without a strong knowledge of exactly why the AMD is superior.



    With AMD vs Netburst Intel you not only had an advantage with with ODMC improving latency to the memory but you had a robust architecture with smaller pipelines and higher IPC (Instructions Per Clock). And AMD whooped up on Intel.



    Now with Core 2 architecture AMD no loner has the advantage of IPC because they still have a 3-issue core whilst the Core 2 architecture has moved to a 4-issue core and still clocks faster. Whilst the Intel Woodrest doesn't have an ODMC it supports either a 1066 or 1333Mhz FSB. I would expect that in a 4-way system on tasks sensitive to memory latency (ie Databases or Mail Servers) you would see AMD systems pull close or exceed Core 2 systems.



    Now that we have both CPU removing cache coherency from the FSB in lieu of onchip communication your FSB isn't likely to be as congested.



    AMD will stay close but I don't see them regaining an advantage in 2-way Servers for a while.
     0Likes 0Dislikes 0Informatives
  • Reply 31 of 38
    Quote:
    Originally Posted by hmurchison




    Now that we have both CPU removing cache coherency from the FSB in lieu of onchip communication your FSB isn't likely to be as congested.



    AMD will stay close but I don't see them regaining an advantage in 2-way Servers for a while.



    what about intel quad-cores that are just 2 duel cores in the same die linked by a FSB?
     0Likes 0Dislikes 0Informatives
  • Reply 32 of 38
    mwswamimwswami Posts: 166member
    Quote:
    Originally Posted by hmurchison


    I think what I'm seeing here is the promotion of AMD chips as being fast without a strong knowledge of exactly why the AMD is superior.



    With AMD vs Netburst Intel you not only had an advantage with with ODMC improving latency to the memory but you had a robust architecture with smaller pipelines and higher IPC (Instructions Per Clock). And AMD whooped up on Intel.



    Now with Core 2 architecture AMD no loner has the advantage of IPC because they still have a 3-issue core whilst the Core 2 architecture has moved to a 4-issue core and still clocks faster. Whilst the Intel Woodrest doesn't have an ODMC it supports either a 1066 or 1333Mhz FSB. I would expect that in a 4-way system on tasks sensitive to memory latency (ie Databases or Mail Servers) you would see AMD systems pull close or exceed Core 2 systems.



    Now that we have both CPU removing cache coherency from the FSB in lieu of onchip communication your FSB isn't likely to be as congested.



    AMD will stay close but I don't see them regaining an advantage in 2-way Servers for a while.



    Very well put. Intel Core 2 is a clear winner in 2 socket 2 core/socket systems. Things start to go AMD's way when you increase the number of sockets or cores/socket. In 2 socket with 4 cores/socket configurations, AMD's ODMC might neutralize all of Intel's advantage and might even take a marginal lead. Time will tell. AMD architecture is currently untouchable in large socket servers. My favorite example in that category is the Sun Fire X4600 with 8 dual core Opterons (going to 4 cores/socket in Q1'07). All the processors are on a separate cards and are connected to each other via Hypertransport connectors.
     0Likes 0Dislikes 0Informatives
  • Reply 33 of 38
    benzenebenzene Posts: 338member
    Quote:
    Originally Posted by Joe_the_dragon


    Clovertown-MP will not be able to beet amd with a FSB.

    ...

    Intel up coming quad-core are doing the same thing as there first duel cores.

    ...

    also they a testing a new intel chip to the old opteron the new ones will add ddr 2 and more ht links and k8l is comeing soon after that.

    ...

    Also was the software useing all of the cores in the quad-core system.

    ...

    what about intel quad-cores that are just 2 duel cores in the same die linked by a FSB?



    You make my eyes bleed.

    Why is it that almost unilaterally, CPU fanbois (either intel or AMD) can't spell worth a damn?



    As an aside, I'm glad that there is competition in the CPU market. Consumers get to reap the benefits of millions upon millions of dollars in R&D from both companies. AMD was on top for a while, and now it's intel. Big whoop, more power to me either way.
     0Likes 0Dislikes 0Informatives
  • Reply 34 of 38
    hmurchisonhmurchison Posts: 12,464member
    Quote:
    Originally Posted by Joe_the_dragon


    what about intel quad-cores that are just 2 duel cores in the same die linked by a FSB?



    a 1333Mhz FSB offers a lot of memory bandwidth. Plus Intel should be using Dual Independent Bus as well. I think AMD will continue to shine in areas where memory latency is important but Intel's more efficient core is going to pay dividens across the wide range of computing applications.
     0Likes 0Dislikes 0Informatives
  • Reply 35 of 38
    placeboplacebo Posts: 5,767member
    Quote:
    Originally Posted by benzene


    You make my eyes bleed.

    Why is it that almost unilaterally, CPU fanbois (either intel or AMD) can't spell worth a damn?



    Yeah, I don't get that either. Computer geeks are usually above-average in smarts, why are they dumb as shit in spelling?
     0Likes 0Dislikes 0Informatives
  • Reply 36 of 38
    tomhayestomhayes Posts: 128member
    I don't see them anywhere.



    I have to buy a new Mac becuase I sold my Dual 2.7 GHz a month ago to have the money on hand to buy a new one.
     0Likes 0Dislikes 0Informatives
  • Reply 37 of 38
    Quote:
    Originally Posted by hmurchison


    a 1333Mhz FSB offers a lot of memory bandwidth. Plus Intel should be using Dual Independent Bus as well. I think AMD will continue to shine in areas where memory latency is important but Intel's more efficient core is going to pay dividens across the wide range of computing applications.



    Dual Independent Bus will not fit in a single cpu socket. Intel will need a new socket with more pins for that
     0Likes 0Dislikes 0Informatives
  • Reply 38 of 38
    slugheadslughead Posts: 1,169member
    Quote:
    Originally Posted by Outsider


    Possible 4-way? in my dreams. I'll settle for a 3-way.



    Not a fan of DVDA?
     0Likes 0Dislikes 0Informatives
Sign In or Register to comment.