AMD brings Quad-cores to the roadmap

Discussion in 'AMD x86 CPUs and chipsets' started by stmok, May 4, 2006.

  1. stmok

    stmok Member

    Joined:
    Jul 24, 2001
    Messages:
    8,878
    Location:
    Sydney
    Website is in chinese, but the roadmap is in English.
    http://www.hkepc.com/bbs/news.php?tid=593736

    It'll be about 12 to 18months behind Intel, but know that its coming. ;)

    Codenames:
    "DeerHound" => High-end multi-processor server variant
    "Greyhound" => Desktop variant
    "Zamora" => High-end multi-processor server variant, 2nd Gen. (uses FB-DIMM)
    "Cadiz" => Workstation and Server variant.
     
  2. MrRevhead

    MrRevhead Member

    Joined:
    Dec 12, 2002
    Messages:
    1,610
    Location:
    Sydney, Australia
    Nice find :thumbup:

    I like how all inter-related news gets released close together. Socket F, HT3 and now this which shows their projected plans for future cores utilises both the technologies.

    Another interesting read that is related to some of the features that are meant to be released with these new cores is this link which analyses the differences in more recent release hypertransport links comparible to AMD's current HT 1.1.
    Direct Souce Link
     
  3. SLATYE

    SLATYE SLATYE, not SLAYTE

    Joined:
    Nov 11, 2002
    Messages:
    26,823
    Location:
    Canberra
    Well, I guess this could be a very good use for that 'anti-HT' technology that they were talking about earlier (making two cores act like one). By the time their quad-cores are released there'll be plenty of dual-core software but not much quad-core software. Making a quad-core act like a dual-core = free performance. I wonder just how flexible they'll be - could a quad-core make three cores act like one core (to run a game) and have the remaining core run the Windows background tasks?
     
  4. qwijibo

    qwijibo Member

    Joined:
    Apr 22, 2004
    Messages:
    408
    Location:
    Old Sydney Town
    Well, in a utopian world, when going to the expense/effort of multithreading programs to support dual cores, developers would make the code work with as many cpus as you have, so it might be a bit of a non-issue. Generally, going from 2 threads to N threads is much easier than from 1 thread to 2.

    Unfortunately, we don't live in that world.

    In theory, if they can get the anti-HT stuff to work, then there is no reason that they can't have odd numbers of cores doing it - the same as if you were splitting up code manually, there is no reason you can't just 3, 5, etc threads.
     
  5. nodeblaster

    nodeblaster Member

    Joined:
    Aug 9, 2004
    Messages:
    44
    Location:
    Melbourne
    It also seems that AMD will utilize a shared cache with these Quad Cores.
    Similar to Intels Core Cpu's, which is what I beleive gives them decent boost in some applications.
     
  6. Phido

    Phido Member

    Joined:
    Jun 20, 2003
    Messages:
    4,895
    Location:
    Dark City
    Zamora seems interesting.

    FB-Dimm?
    Shared L3 cache? Additional cache over the L1 and L2 Just for servers?
    But why have deerhound?

    I guess these are more of IBM, Sun, etc. Where as regular multi cpu users will get Cadiz
     
  7. R3xx

    R3xx (Banned or Deleted)

    Joined:
    Mar 12, 2003
    Messages:
    2,453
    Location:
    Melbourne
    Either way... this means Intel get a new asshole torn again! :D
     
  8. SLATYE

    SLATYE SLATYE, not SLAYTE

    Joined:
    Nov 11, 2002
    Messages:
    26,823
    Location:
    Canberra
    Not really. If you read the original post:

    It looks like Intel will be doing it first - and by quite a long way. Depending on how you read the benchmarks, Conroe may well be significantly faster than the A64 clock-for-clock anyway.
     
  9. -=N0N@ME420=-

    -=N0N@ME420=- Member

    Joined:
    May 9, 2004
    Messages:
    7,562
    ??

    AMD has always been behind Intel, as chainbolt blasted me in another thread Intel were the first with basically everything ever invented to do with the cpu.

    AMD has Hypertransport and that's about it, but I don't really care, Intel can have DDR2, DDR3, DDR8 first, they can have quad-core, 8-core, or even 100 core because at the end of the day, it takes these products a year or even 2 years or more to mature and AMD usually comes in with a huge boast in performance, look how far 64 Bit has gotten, dual core has been out for a year now, Intel and AMD released around the same time has the market for dual core software improved considerable? probably, but only because dual core is "latest" not exactly because it's the greatest ie intels entire dual core lineup.
     
  10. SLATYE

    SLATYE SLATYE, not SLAYTE

    Joined:
    Nov 11, 2002
    Messages:
    26,823
    Location:
    Canberra
    AMD's been ahead of Intel in terms of performance (especially with dual-cores) for a while now.

    I assumed that R3xx was referring to the poor performance of Intel dual-cores relative to AMD ones; he seemed to be saying that we would be facing a very similar situation this time round (with quad-cores). I disagreed, since Intel's going to have their quad-cores out well before AMD do.
     
  11. R3xx

    R3xx (Banned or Deleted)

    Joined:
    Mar 12, 2003
    Messages:
    2,453
    Location:
    Melbourne
    Correct... I should have been clearer with my post. Yes Intel generally come out with the latest tech first but the later AMD incarnation usually performs better.
     
  12. slobber

    slobber Member

    Joined:
    May 1, 2005
    Messages:
    5,207
    Location:
    Mandurah!
    Yep, intel invented the wheel, bread and butter, electricity, HUGE PR campaigns to brainwash ppl into believing all this crap, and lastly supplying all the intel shills with all the propaganda and h/ware needed :sick:
     
  13. stalin

    stalin (Taking a Break)

    Joined:
    Jun 26, 2001
    Messages:
    4,581
    Location:
    On the move
    IIRC AMD had better performing 386's :) so Intel pushed the release of the 486 faster :)

    AMD made it to 1GHz first, intel made it shortly after, but it was a token effort and 1GHz PIII's were impossible to find at the time and overheated.

    AMD are not the back water company people seem to think they are.

    Intel tends to release most tech first, AMD utlise that similar technolgy (which they develop in parallel generally) and proceed to eclipse the performance of Intel. This game of leapfrog had occured for many many CPU generations. The original Athlon was the first were AMD really took a lead.
     
  14. FlameHead

    FlameHead Member

    Joined:
    Jul 14, 2001
    Messages:
    2,072
    Location:
    Sydney
    He didn't say it was easy to do 2 to N but it is definetly easier than going 1 to 2.

    I know in my own programming, as soon as I start doing multithreading I end up with entire handfuls of threads. Limiting it to 2 or saying that 2 is easier to manage than N is not in the least bit accurate.
     
  15. morbo

    morbo Member

    Joined:
    Jun 26, 2002
    Messages:
    53
    Good luck with that.

    edit:
    Making an application multithreaded has little to do with making it SMP aware.
     
    Last edited: May 8, 2006
  16. Kurosaki

    Kurosaki Member

    Joined:
    Dec 8, 2005
    Messages:
    4,770
    Location:
    Australia
    Can you please provide a list of technological 1st from AMD that doesn't include HT?
     
  17. heath_303

    heath_303 Member

    Joined:
    Sep 8, 2002
    Messages:
    1,305
    Location:
    Brisbane
    first to 1ghz for one, i think that is one of the most significant milestones in computing.

    i think that intel has definately surpassed AMD with new tech over the years but AMD has suprised intel over the last 6 years.
     
  18. Hipcrostino

    Hipcrostino Member

    Joined:
    Apr 24, 2006
    Messages:
    29
    Ummm isn't everyone forgetting that AMD produced the first 64-bit chips capable or running 32-bit applications??? IT was one of the biggest milestones the company ever had in the recent past.

    Yes intel had 64-bit chip before AMD, but Intel actually expected everyone to buy 64-bit chips that could only run 64-bit apps. Meaning to upgrade, you would need to buy all your software again. Thast what they never bought it out, because the market wouldn't accept it and they knew it. AMD figured out a way, and thats why AMD64 has been so successfull.
     
  19. Kurosaki

    Kurosaki Member

    Joined:
    Dec 8, 2005
    Messages:
    4,770
    Location:
    Australia
    Wrong. The Itanium ran all 32bit code in emulation mode conducted on the CPU. It was just slow.

    IBM, Sun, DEC, Alpha all had 64bit gear before Intel. Mass server farms aren't exactly things anyone can buy into. Itanium has a small market niche as does the other CPUs I've just mentioned. For example NASA has built entire server farms around Itanium CPUs for their space research. Besides building on a foudation laided almost 30yrs ago is hardly earth shattering stuff.

    Commercial success != Great feat of engineering.
     
  20. slobber

    slobber Member

    Joined:
    May 1, 2005
    Messages:
    5,207
    Location:
    Mandurah!
    Hehehehe yes the P4.............NOT, its only reason for commercial success was the fact that demand far outstripped supply

    intel could have been selling rebadged P3 celeron's and would have still made money :sick:
     

Share This Page