Slashdot Mirror


K7 to exist in socket variety by 2000

EB News is reporting that the K7 should be out by the end of June, featuring a 512Kb or 1Mb L2 cache. In 2000 it should exist in a socket variety, hinting that a lower cost version may become sooner than later.

39 comments

  1. Single Vendor Considered Harmful by Anonymous Coward · · Score: 0

    That's handy, though Cyrix/IBM chips working in Slot A would make me happier. Clue, guys - if I wanted to prop up a potential monopoly, I'd buy Intel.

  2. All good and well, but... by Anonymous Coward · · Score: 0

    ..the pricing of Intel (I know, I know, this is AMD we're dealing with..) 512k and 1024k-cache'd chips was nothing short of startling. Prices more than trebled for more cache.

    Are AMD going to be any different?

  3. ...not if they wanna keep 200 MHz bus speed by Anonymous Coward · · Score: 0

    Fastest they could do is probably 100 MHz FSB. Although, they might be able to pull another rabbit out of their hat and push the Super 7 bus to 133 MHz (Super Duper 7?), but that would be unlikely I think. Even if they could, it would only be 2/3 speed of 200 MHz bus. So you would have processor that could wait really fast. How much life is left in Socket 7? Anyone have any more info on its limitations?

    Could they be using a(nother) proprietary socket (not a slot) for the K7? I doubt this would be true.

  4. Need OSS for Hardware!!! by Anonymous Coward · · Score: 0

    Damn, all these new proprietary buses and slots are making me mad. I wish that one day we can maybe make it so that we have one interface that we plug stuff into (bus, chip, etc.) and make it all work. That way, a company or a community (a la OSS) could build hardware components that can interconnect, no matter what.

    Yes, yes, I know the potential of the upcoming Device Bay spec... Hope that it becomes a reality. And extends to all things, not just storage.

    Just my $0.02 :)

  5. wtf is this?? by Anonymous Coward · · Score: 0

    The only reason Intel ever went to a slot design was so that they'd have somewhere to shove the cache with it's little 1/2 clock speed bus (and some would say, to have their own proprietary socket). Now that they're moving to on-die cache, there's no reason to use a slot, and it's cheaper not to. Slots or pins, it's all just contacts, and I don't really think either is "more advanced" than the other.

  6. No Subject Given by Anonymous Coward · · Score: 0

    Looks like the Pentium III is gonna kick K7's ass.. It beats it both with the release date and, even more important, it outperforms it...

  7. Half Right by Anonymous Coward · · Score: 0

    The Pentium III, a gross misuse of the pentium name, will beat the K7 out but preformance is wy too early to tell. If we go on specs alone the K7 will win. K& features: 128 L1 Cache (2x more than a Pentium III) L2 cache from 512k to 8 MB (512k for P3) 200 MHz bus (133 MHz for P3) a FPU = to intels in size, but can accept 3, not P3's 2, instructions at once. K& Preview

  8. Single Vendor Considered Harmful by Anonymous Coward · · Score: 0

    Commodity Alpha hardware! That *does* have potential....

  9. SMP? by Anonymous Coward · · Score: 0

    What's the deal with SMP? Is there going to be any?
    Currently being a happy K6-2 owner, but if they don't put in SMP in K7 they can forget me!
    Regarding whether they invent a new socket I really don't care, when I buy a new CPU it's usually also time to buy a new MB (and yes, I _am_ a "poor" student!).

    If the indications that Transmeta will be shipping a good SMP chip, perhaps that's where I'll be.

    Death to single-processors (and non-distributed systems :)

    Jonas / not-ac'ing, just lazy

  10. 128k FS L1?!?!? by Anonymous Coward · · Score: 0

    FYI, the lead architect for the K7 was formerly one of the 20164 designers...so the similarity isn't without coincidence.

  11. Lay off that crack pipe.. by Anonymous Coward · · Score: 0

    Whoa...not sure where you're thinking that the PIII will be faster than the K7...have you looked at the K7 design? The FP performance in particular should pretty much rape the PIII.

  12. Alpha/K7-machine by Anonymous Coward · · Score: 0
    Imagine a SMB-machine that contains a fast alpha-processor for Linux-binaries AND a K7 for Intel-compatibility (both use the same EV6-bus). The kernel should check which commands should go to which processor.

    Guess this would be a pretty fast machine and it would be a KILLERFEATURE compared to dumb old WinNT.

    Dont think it would be too-hard to realize...

  13. K7 is way off yet... by Anonymous Coward · · Score: 0

    The K6-3 isn't even out yet and AMD are telling OEM's that they shall receive their deom chips at the end of February. That means a full release isn't likely until April. The K7 is surely a good few months away behind that wchich will put it at a release date similar to Intel's next gen chips. I wish AMD well (after all, they have significantly affected Intels pricing policies) and their competition is a good thing. But, I can't help feeling that they are running a bit late and Intel could well have stolen the march already by then. Also, the new motherboard architecture needed for the K7 will be hideously expensive as it will be very complicated and surely no motherboard manufacturer will make them in sufficient numbers to begin with. It's a BIG gamble on AMD's part I reckon...

  14. Sockets are technically superior to slots by Anonymous Coward · · Score: 0

    Intel only went to the slot to stave off competition. The socket is smaller and closer to the mainboard and has tighter CPU fit. This all adds up to a more reliable and faster connection. It's why the big boxes (HPs, Alphas, etc.)still use sockets rather than slots. Throw this slot trash in the garbage and get back to serious computing.

  15. Transmeta needs to.. by Anonymous Coward · · Score: 0

    keep cost of CPUs under $250. Then I'd certainly stampede over and buy a EV6 SMP motherboard and buy 4 of these things, with the potential to add m more CPUs (256 CPUs on one motherboard with 8GB makes me drool... sorta)

  16. Draining You by Anonymous Coward · · Score: 0

    All CPU maker are currently able to produce 1 Ghz but does not, due to the amount of money they can make thru slow advances. Draining consumers of they money, by keeping up with those slow advancements (upgrading, to keep up).

  17. Draining You by Anonymous Coward · · Score: 0

    All CPU maker are currently able to produce 1 Ghz ,but does not, due to the amount of money they can make thru slow advances. Draining consumers of their money, by keeping up with those slow advancements (upgrading, to keep up).

  18. What colour are the clouds in your world? by Wakko+Warner · · Score: 1
    I'm guessing they're an attractive teal shade, myself.

    - A.P.
    --


    "One World, One Web, One Program" - Microsoft Promotional Ad

    --
    "Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
  19. Wrong.. by HoserHead · · Score: 1
    the celeron's cache speed is at 1/2 the CPU, not at 1/1 (as you say).
    Nope, the Celeron has 128 kB of cache on-die, running at the processor's speed.
    1/2 cache/cpu speed ratio compared to 1/1 isn't as big a difference as you might think.
    It's more than enough to make the AMD K6-3 with 256 kB of on-die cache the clear winner when compared with Intel's 1/2 speed cache chips.
  20. wtf is this?? by gavinhall · · Score: 1

    Posted by modefan:


    AMD going to release a K7 at 1/3-1/2 the speed of the CPU? They better have 600+ mhz machines out ASAP, otherwise they'll still be playing catchup. I have used AMD chips for a while, but when it comes to playing FPS, then AMD don't cut da mustard.

    The Celeron processor sounds like it could still kick K7 in the jock strap. With a full speed cache (albeit 128k), it can really rock when it comes down to games and such.

    But all of that is within the i386 thinking, not the x86 thinking. If AMD finally engineers an awesome chip so that the cache's speed isn't as important as the size, then you'll see me getting in line for a K7.

    Lastly, what's up with the socket re-revolution? Intel said "no more socket". Now we've got celery coming out on socket. AMD stated that socket would last until 2000, now they plan to extend that support.

    Drop the socket for all its worth and go slot. If you made enough slot instead of socket, then it would be cheaper =)

    Legacy has go to die sometime.

  21. wrongo buddE by gavinhall · · Score: 1

    Posted by modefan:


    lookup Intel's developer information for the Celeron and you'll be schooled. check your facts before you post.

    celeron 128k cache is running on a 1/1 ratio with the cpu.. meaning that the kicks more ass than ike turner on smack (from billy at ve3d).. the celeron is like a mini xeon..

    schooled on /.
    how pathetic.

  22. wtf is this?? by gavinhall · · Score: 1

    Posted by Mephie:

    BZZT! Ok, since bus speed is 200mHz, that means cache is running that fast, or you've got a bottle neck there. So,that means 200mHZ is the 1/2 or 1/3rd, and bus speeds should be 400-600.
    The Celery may the K7 on cache, but the K7's FPU's are going to be ungodly.
    That's why I'm building a quad.
    --Mephie

  23. No Subject Given by gavinhall · · Score: 1

    Posted by Mephie:

    And is the III in the name it's SMP limit like the P-II ?
    --Mephie

  24. No by six · · Score: 1

    K7 is by no means Super7 compliant. It uses the alpha EV6 bus protocol and needs a new chipset.

    I think we'll see a "Socket A" soon !

  25. Good news. by Bwana · · Score: 1

    Very Good news. I'm looking forward to getting this processor.
    "Electric Relaxation" - ATCQ
    Bwana McCall
    bwana@nye.net

    --

    "Electric Relaxation" - ATCQ
    - Bwana
  26. wtf is this?? by Dastardly · · Score: 1

    The Celeron processor sounds like it could still kick K7 in the jock strap. With a full speed cache (albeit 128k), it can really rock when it comes down to games and such.

    Wrong!

    The K7 has a 128K full speed L1 cache. Put that on top of a moderately fast L2 and you have a screamer. Imagine a PII with a Celeron core, but better.

  27. Toms Hardware by vinn · · Score: 1

    Yesterday Tom of Toms Hardware had a little editorial and he ranted a bit about AMD's problem getting the K6-3 out the door. Pretty interesting.

    Even more interesting was something a saw a while ago about AMD development. It seems the people working on the K7 are not the same people who worked on the K6. The K6 was developed by the NexGen guys AMD bought out. However, the NexGen guys have already moved on and are working on what will presumably be called the K8. Of course that will be competing in the Merced world, god only knows what that will be like.

    --
    ----- obSig
  28. not likely by Akira1 · · Score: 1

    The K7 will not launch at 30% of the cost of the equivalent Intel chip like AMD usually does, but it will be somewhere around 800 dollars, or so I've heard. That is still far better then a Xeon is now, and what a Xeon will be in 6 months. Shure it isn't launching at under 400 bucks as AMD usually does, but it is going to be one hell of a chip. Needless to say I want to have a dual/quad K7 system just for the hell of it.

    --
    Food: It's whats for dinner
  29. Cache speed difference by Akira1 · · Score: 1

    Intel also had to fab the cache memory for the Xeon chips, and their facilities are not setup to manufacture cache efficiently. Previous chips/mobo's farm the cache memory production to small Asian companies that merely make cache.

    --
    Food: It's whats for dinner
  30. wtf is this?? by Akira1 · · Score: 1

    The K7's FPU is supposed to absolutely stomp intel's. It is supposedly able to execute 3 instructions per clock cycle, not to mention the 3DNow support. From what I've read so far, this will be AMD's best FPU ever, and the best X86 FPU we have seen. Now it stands to reason to take into consideration the fact that 3DNow support might fade with katmai on the horizon, but the fact that developers are using it now, with katmai 2 months away says that we might have both of these technologies simultaneously on the market. And of course the K7's integer performance will just completely smoke. then there is the 200mhz FSB. Think about that for compiling, 1.6GB/sec pipe to your 1GB+ of RAM, oh did they mention SMP support. This is major drool factor here folks.

    --
    Food: It's whats for dinner
  31. Finally a release date!!!!! by i_hate_windows · · Score: 1

    I've been reading about sharptooth for 6 months and can't wait for it to come out. Intel better watch out lots more L1 cache and AMD says they will release K7's with up to 8 megs of L2!!
    The K7 will also be a 64 bit processor and have a 200 MHz front side bus! Also plan for clock speeds up to 1 GHz!
    This is the great stuff you get when Dirk is making your processors

  32. It is possible by kellman · · Score: 1

    That is what the article was saying... Socket7 by 2000 for the K7

    --
    I don't want to sell anything, buy anything, or process anything. I don't want to sell anything bought or processed...
  33. Is it possable... by walflour · · Score: 1

    for them to stuff the K7 into super socket7 MBs?

    --

    --
    When she told me I was average, she was just being mean.
  34. amd? intel? nope. by Chomsky · · Score: 1
    Look for the dark horse to come into the picture and make both pentium III and K7 irrelevant.


    I'm referring of course, to the transmeta apocalypse. =)

    --
  35. Need OSS for Hardware!!! by jonbrewer · · Score: 1

    Look what happened to CHRP and the PowerPC.
    M$ killed NT 5.0 for PowerPC and it died.

  36. Maybe this is a dumb question... by Omnifarious · · Score: 1

    Did Intel give up on Slot 1 finally?

    Everybody seems to be saying that the K7 is going to Alpha's bus and board architecture. I'm not up enough on this kind of thing to be able to say much about that, but it seems like a good idea since the Alpha runs in the range that x86 style processors are now starting to reach.

    Why would anybody want Socket 7 anyway? It mostly seemed to me that Socket 7 was what the K6 did because Intel locked them out of the Slot 1 market. If they've found a nice new bus architecture that works better and supports PCI, I say go for it.

    Is anybody interesting in explaining the differences between EV6 (Alpha's bus?) and Slot 1? Is the question even relevant anymore?

  37. Speaking of which... by Kludge · · Score: 1

    Does anyone know of anyone who makes EV6 motherboards other than DEC? I'd love a cheap K7, but if the motherboard is going to drain me, forget it!

  38. woohoo! by flirzan · · Score: 1

    I've been waiting for the K7 since the K6-2 came out, now lets all hope that AMD will stick to their normal practice of setting the pricepoint below Intel's, and there will be much rejoicing.

    flirzan

    --
    Twinkies sure taste good for something that is 68% air.
  39. wtf is this?? by Ravi+Swamy · · Score: 1

    Intel's "no more socket" talk was all marketing BS. If you look at a Celeron 266, 300, 300A, or 333 you'll see that the PCB is virtually empty. The point of the PCB was for the P-II cache which is now gone or on chip so get rid of it and put the CPU back in a socket. The packaging is much cheaper. If they made more slot instead of socket they'd just make less money.