1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Socket AM2 Forward Compatible With AM3 CPUs

Discussion in 'News' started by malware, Jul 6, 2006.

  1. malware New Member

    Joined:
    Nov 7, 2004
    Messages:
    5,476 (1.54/day)
    Thanks Received:
    956
    Location:
    Bulgaria
    AMD has revealed that Socket AM2 will accept both the current AM2 and the next-generation desktop AM3 CPUs. The most recent AMD roadmap is also very clear to state AM3 "supports either DDR2 SDRAM or DDR3 SDRAM, but not on the same motherboard". This means that AM3 CPUs will have DDR2 and DDR3 integrated memory controller allowing them to run either on a Socket AM2 board with DDR2 memory or on a future Socket AM3 motherboard with DDR3 memory and HyperTransport 3.0. HT-3 allows 5.2 Giga-transfers per second, while current AM2 motherboards top out at 2.0 Giga-transfers per second. There is no word about the pin configuration on the future AM3 CPUs but it has to be 940 in order to fit the current AM2 Socket. AMD has not announced when Socket AM3 will be released, although since it would be the first HT-3 "ready" socket design, it seems likely that K8L and AM3 for the desktop will launch simultaneously.

    Source: DailyTech
  2. v-zero Guest

    Pretty sweet news for anybody who thinks (wants to believe?) AMD are still in the race for the next three-quaters...
  3. wazzledoozle

    wazzledoozle New Member

    Joined:
    Aug 30, 2004
    Messages:
    5,414 (1.49/day)
    Thanks Received:
    161
    Location:
    Seattle
    Well really 939 could have been used, since the memory controller is on the cpu all they would have needed is a new cpu and a mobo with DDR2 slots.
  4. v-zero Guest

    I imagine a 939 compatible pin layout for DDR2 was beyond the realms of feasibility, otherwise it would have been done...
  5. wazzledoozle

    wazzledoozle New Member

    Joined:
    Aug 30, 2004
    Messages:
    5,414 (1.49/day)
    Thanks Received:
    161
    Location:
    Seattle
    No, because AMD needs to make money, and new socket= new motherboards AND cpu's. I am sure every motherboard manufacturer pays a small fee to AMD to make boards compatible with their cpu's.
  6. v-zero Guest

    We both know that doesn't stack up at all. No CPU manufacturer moves to another socket without needing to. It may be that the new socket has been introduced to allow them to iron out the kinks when the "real" processors (K8L) aimed at Conroe are released... However, from a revenue standpoint the release of AM2 makes no sense.
    Chipset manufacturers don't have to pay a fortune to develop for AMD, hell I bet nVidia and ATi pay nothing at all. AMD are helping, not hindering mobo manufacturers here...
    Let's face it, if they could have kept 939 they would have - but the socket and pin layout was obviously not suited... AMD would sell more 939 CPU's rather than change CPU sockets if they could. They have taken huge cuts on the 939 line, and that hurts revenue.
    So AM2 is not an immediate revenue move, it is probably there as a forerunner for something bigger, and also just to keep up with the technology tides...
  7. NamesDontMatter

    Joined:
    Oct 1, 2005
    Messages:
    788 (0.24/day)
    Thanks Received:
    30
    As the gamers say AMD will be PWNED for 3 quarters.
  8. v-zero Guest

    Bingo NDM, bingo!
  9. Alec§taar New Member

    Joined:
    May 15, 2006
    Messages:
    4,677 (1.56/day)
    Thanks Received:
    94
    Location:
    Someone who's going to find NewTekie1 and teach hi
    Gotta admit, I am totally "with" v-zero's original post's reasoning

    See subject-line/title: Absolute agreement!

    :)

    (Good "business-sense/intuition" imo, v-zero!)

    APK

    P.S.=> I just said this today in another posting, & it appears to be true here as well: The "big 2" in CPU's, just like in memory & graphics, tend to "change-the-rules" on hardware when needed for that "competitive edge"... & it makes you wonder just how FAR AHEAD in performance we truly COULD be, if they just went "all-out" & produced what they probably could if they opened up patent safes + R&D findings of theirs... apk
  10. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    19,726 (6.21/day)
    Thanks Received:
    5,869
    I believe that 939 was perfectly capable of running DDR2 memory and AM2 processors. I believe they made the socket change for marketting. The processors add nothing but DDR2, not really a big deal, however if you all of a sudden throw in a socket change too it makes is seem like more of a big deal.

    I also believe that they did it to avoid confusion. Since the AM2 processors don't support DDR1 they didn't want to make the processors seem compatible with DDR1 boards. We all know that if they did make them 939 people would be putting them in their DDR1 motherboards and then wonder why it doesn't work.
    Crunching for Team TPU 50 Million points folded for TPU
  11. Alec§taar New Member

    Joined:
    May 15, 2006
    Messages:
    4,677 (1.56/day)
    Thanks Received:
    94
    Location:
    Someone who's going to find NewTekie1 and teach hi
    DDR4 is here already too! Will wonders NEVER cease...

    See subject-line/title, & this URL (here @ these forums) as well:

    http://forums.techpowerup.com/showthread.php?t=14053

    :)

    * I wonder if this will be a "HUGE" improvement & what gains are available using it... in other words, is there mobos that can exploit its benefits today?

    APK
  12. Jimmy 2004

    Jimmy 2004 New Member

    Joined:
    Jan 15, 2005
    Messages:
    5,491 (1.58/day)
    Thanks Received:
    267
    Location:
    England
    I reckon AMD could have used 939 sockets to control DDR2. It may have other advantages, but the main aim of businesses is to make money and if the best way for them to do that is to introduce a new socket, then they'll do that.
  13. KeepHerSlidin New Member

    Joined:
    Jul 9, 2006
    Messages:
    2 (0.00/day)
    Thanks Received:
    0
    I am not sure what the k8l is But the big thing to combat conroe is the amd 4X4 due out in fall .

    as for gaming I am not so sure intel is the best as the top scorer in 3dmark06 is someone using a amd board including sli of the newer nvidia cards a physx card also i think.

    amd processors are usually better performing at lower speeds and when you consider sli and quad sli plus ddr2 800 mhz with am2 motherboards your good. the xfx 1900 doesnt come close to the nvidia 7900gtx/gtx extreme edition or the 7950gx2 and with sli they would kill a crossfire setup of the top cards.

    Are you guys trying to say nvidia cards work better with intel processors?? one other thing Ive seen people with nice cooling heatsinks in benchmarks increase the 1 ghz/2000mt/s hyper transport speed to over 2.5 ghz and using a dual core 2.2 or 2.4 ghz to go to dual core 3.1 ghz. I am willing to bet a lot of people with great heatsinks such as the ultra 120 or tuniq tower 120 and good cooling fans could keep the fx62 running at dual core 4.2 ghz in the 50s celsius(chipset) and the cpu around 35-40 celsius.

    i think whether intel or amd is better for games depends on your overall setup and cards used. obviously any system with the newest cards will be pretty damn great.

    In the end its all really preference and how much performance you need or want from your computer.if you are hardcore the fx62 or intel dual core 3.8 ghz and the 7950 gx2 or x1900 are probably what you want.

    One thing I would like to see is the fx 62 at dual core 3.8 ghz and the stock intel dual core 3.8 ghz with the same video card setup and same ram and see what system comes out on top for the latest games.

    any opinions are welcome its just between a 2 ghz and 3.4 ghz single core intel processor i dont see much better performance application loading wise and video encoding wise.
  14. v-zero Guest

    We're talking about Conroe, which is faster than any AMD chip fior gaming, amongst other things... It has a higher IPC (faster at lower speeds) than the A64 lineup (K8), and is looking to overclock better too...
    There is nothign between ATi and nVidia at the moment, so that argument is wasteful.
    Last edited by a moderator: Jul 10, 2006
  15. rpg711

    rpg711 New Member

    Joined:
    Jun 17, 2006
    Messages:
    781 (0.26/day)
    Thanks Received:
    0
    Location:
    MA
    wat about the am2 fx 62? that should atleast beat the conroe by alittle
  16. KeepHerSlidin New Member

    Joined:
    Jul 9, 2006
    Messages:
    2 (0.00/day)
    Thanks Received:
    0
    in all of my stuff i wasnt considering conroe i am talking about existing systems i know conroe and fx 62 will be ruling until k8l which i guess is code named the 4x4. when i was saying dual core i meant the amd series mainly the x2 and fx series.

    also i said processor not card at lower speeds. the reason its called the 3800 and its only at 2.0 ghz is because it compares to a intel 3.8 ghz in speed.of course with the fsb of ram being higher for intel now that gives them a little of an edge.

    and i do mean in the 50s under load with overclocking about 1.2 ghz higher in amd or intel.the heatsinks usually getting these results is tuniq tower 120 and ultra 120 by diff companies.

    and i said the system with the newest cards not processors will be the greatest for gaming. i meant graphics cards not processors.

    im not fighting for invalid reasons im talking current gen only conroe isnt out yet.
    i want to see dual core intel 3.8ghz unoverclocked performance against amd dual core fx62 am2 overclocked to actual 3.8 ghz performance.

    last but not least when i said a amd 2 ghz and single core 3.8 ghz intel i meant the lga775 processors and i meant both as single core which means i wasnt comparing dual core to single core.

    as for spelling errors i did not spell wrong maybe punctuation wise.
  17. rpg711

    rpg711 New Member

    Joined:
    Jun 17, 2006
    Messages:
    781 (0.26/day)
    Thanks Received:
    0
    Location:
    MA
    ok... slow down there... i have no clue wat ur trying ot get at
  18. rpg711

    rpg711 New Member

    Joined:
    Jun 17, 2006
    Messages:
    781 (0.26/day)
    Thanks Received:
    0
    Location:
    MA
    lol ndm... its a new forum member so excuse the member's stupidity
  19. mikelopez New Member

    Joined:
    May 12, 2006
    Messages:
    170 (0.06/day)
    Thanks Received:
    0
    Location:
    Tampa, FL
    Hey, NDM, why are you being such a jerk? I know you just said sorry and after you went on your first power trip you said you weren't trying to be mean but that is exactly what you did.

    It is hilarious that you try to correct his "English language" but you commit worse mistakes. In fact, as you were correcting him on his spelling (which didn't need correcting) you misspelled properly. Then when you tried to explain yourself you misspelled grammar. Well I have some advice for you "Please drop your post into MS word next time!"
  20. rpg711

    rpg711 New Member

    Joined:
    Jun 17, 2006
    Messages:
    781 (0.26/day)
    Thanks Received:
    0
    Location:
    MA
    dont try to make enemies with a person that is names dont matter... lol anyways why are we even talking about being mean to a newbie... we all should have better things to do. anyways u probably made him mad enough to quit the forums
  21. Steevo

    Steevo

    Joined:
    Nov 4, 2005
    Messages:
    8,121 (2.54/day)
    Thanks Received:
    1,127
    Bagh.


    AMD released LGA style CPU's at it relieves them of a few issues that pinned CPU have had.

    1) EMI, or crosstalk, between pins on CPU's cause data corruption at high speeds.
    2) Who here has never either dropped a CPU, or bent a pin or two, who hasn't recieve a damaged one from shipping? No pins=no bent pins. Simple huh?
    3) More Ghz+++ needs more I-O capabilitys, really, so at what point do pins become too small and too dense to be reasonable? 939 was packed pretty tight, and to think a few pins were trying to move 80 watts at 1.5 volts? Damn warm if you ask me.
    4) Easier assembly for OEM's and for us.
    5) Stronger board, easier to mfg.
    10 Million points folded for TPU
  22. rpg711

    rpg711 New Member

    Joined:
    Jun 17, 2006
    Messages:
    781 (0.26/day)
    Thanks Received:
    0
    Location:
    MA
    how is the board stronger? i have a 775 board and it doesnt seem much differnt
  23. NamesDontMatter

    Joined:
    Oct 1, 2005
    Messages:
    788 (0.24/day)
    Thanks Received:
    30
    I deleted my post because I was far to tough on the new member. Whether you like it or not we all say stupid things and we all make mistakes. I apologies for my approach to the situation. By the way mike Lopez you forgot a comma. :roll:
  24. Steevo

    Steevo

    Joined:
    Nov 4, 2005
    Messages:
    8,121 (2.54/day)
    Thanks Received:
    1,127

    Each layer of the board has holes and pins through it that extend all the way through and are soldered in place. At least in the case of high quality boards.

    A standard board has to have hollow pins and half of one face removed to allow for pin insertion and clamping.



    Sit at the masters feet and learn little cricket.


    http://www.research.ibm.com/journal/rd/466/corbin.html
    10 Million points folded for TPU
  25. v-zero Guest

    No. Conroe = lower IPC and higher/similar Clock Rate, so it will kill the fx-62!

    Is this the 1207-pin layout? Otherwise I don't know what AMD chip you're talking about... AM2 is not LGA...

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page