Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
AMD

AMD Officially Rolls Out 1Ghz Athlon 179

spudwiser writes: "AMD has a press release on their Web page concerning shipment of the 900, 950, and 1000MHz Athlon processors. Also included are times for the live satellite interview with the CEO and VP of AMD." Check out some of the benchmarking info about the new chips as well. I wonder how Andy Grove [?] is feeling today.
This discussion has been archived. No new comments can be posted.

AMD Officially Rolls Out 1Ghz Athlon

Comments Filter:
  • I'm not quite sure why they spent all this effort being the first to 1GHz. So, they beat Intel by two days. Who cares? Do they think the consumers do? I don't. The average computer buyer doesn't, either.
  • by Anonymous Coward
    Are you kidding? The L1 cache is DAMN important... Sure you have to feed it, but remember it gets fed much more than the item you are reading! It has spatial locality too... This means that sequential loops will run mostly in the L1 cache. The L1 gets fed basically once if you can keep all the inner data inside the L1. You don't need to load an entire application - just the loop that is currently happening. The difference between a loop in the L1 and a loop in the L2 is STAGGERING! Try benchmarking for x for y compared to for y for x and you will see the difference. A larger L1 cache should make an immediate difference in perf0ormance
  • by Anonymous Coward
    My 400MHz G4 is still faster than the 1GHz Athlon...and 3times faster than the PIII 800. Suck on that Wintel / Linux fanatics!
  • by Anonymous Coward

    So? Next year it'll be $199 for qty 1, when $1000 gets you the 2 GHz chip. Just buy two or three steps behind the top-of-the-line, and everything is quite affordable.

  • by Anonymous Coward
    The 1GHz Athalon uses a 1/3 clock divider for the cache. If Intel uses a full speed on die then I think it will be faster than the Athalon overall. Even though on Sharkys, that K7 beat out every thing, the P3 800 was not too far behind. The P3 will prevail at 1GHz until AMP gets out an on die implementation. Enough of this. Go here for some Scooby action [scoobydoo.com].

    Trolling for Scooby doo!
  • I came across some old PC magazines at a library a year or two ago. It was cute to see 1988/89 Gateway ads with 486 SX/25s and DX/33s for $4000. Ah, the memories...
  • Most importantly, when will these come out in Dual/Quad configurations? Anyone who only uses one phone at a time is borderline Amish if you ask me.
  • Every schoolchild knows that Bull Aldrin got there second

    That should be "Buzz", as in Lightyear.
  • Nice.. Now I just need to find a decent motherboard manufacturer that produces dual/quad processor configurations for these beasts.

    --
  • A pet peeve: People refer to the new Athlon chipsets as "SMP". They're not. That's Symetrical MultiProcessing. They're Point to Point Multiprocessing, which I guess would be PTPMP or something.

    Can we call it PPMP and pronounce it "pimp"?

    Seriously, now: Would a Linux kernel know the difference? Do they work equally well with Linux? I mean, there's only one option (wrt multiprocessing), for SMP. Does anyone with a multi-Athlon box care to comment?

    --Ben "needs a job that involves work" August

  • Maybe you all forgot, he stepped down as CEO last year. Isn't he on some island drinking from coconuts right now?
  • I think larger L1 cache will compensate it. BTW - it's strange how people only keep in mind L2 cache when talking about caches. Considering that L1 cache does 90% (or so) of the speeding job.

    Not sure where you are getting your numbers on this one,
    [...]

    The rule-of-thumb is that every level of cache catches (hits) about 90% of the accesses thrown at it. This means that for 100 accesses, you spend: 90*[time to access L1] + 9 * [time to access L2] + 1 * [time to access Main RAM] . (And so the average is 0.01 times that. )

    So, no L2 speed does not really matter all that much. Sure you can generate a benchmark that fits in L2 cache, but not in L1. Then it will depend highly on the L2 speed. But in general, the on-package L2's are good enough.

    Roger.

  • The Athlon 800 uses a 2/5 ratio cache, the Athlon 1000 uses a 1/3 [so the /800 has 320MHz L2 cache and the /1000 has 333MHz]. That's on the Anandtech site.

    Comparing P3 and Athlon benchmarks is a bit fiddly, because the best P3 results are obtained using Rambus memory, an i840 motherboard, and Intel's v4.5 compiler which can vectorise FP operations and do prefetch.

    For Athlon systems, you're tied to PC133 memory on a KX133 motherboard, and it's not possible to tell ICL v4.5 to generate prefetch operations but not SSE, so you don't get the prefetch benefit.

    So you get a SpecFP95 figure of 29.4 for the K7/1000 from AMD; if you go to specbench and search for the P3/800, you get

    24.5 - BX motherboard, PC100
    28.9 - i820 motherboard, Rambus, Intel
    32.4 - i840 motherboard, Rambus, Dell

    If you compile on the P3 with options such that the code also runs on the Athlon, you get a score of about 19.5 and the P3 appears incredibly slow - but this is what you expect when running code which is essentially unoptimised.

    Running hyper-optimised cache-blocked code, like Prime95, an Athlon/500 is 25% faster than a P3/500E and 45% faster than a normal P3/500.

    Of course, the K7 system is enormously cheaper than a P3-and-Rambus system here in the UK.
  • It drives down the costs of the middle (read 700mhz) processors and the true test of who's in better shape will be to see the availability of the systems. I get the feeling AMD's yeilds are much better than Intel's at this hairy edge of speed.
  • Hillary as the first to scale Mt. Everest. Nobody remembers who got there second..."

    Tensing got there second.
  • No... AGP is a completely separate bus from PCI. That's how it's able to manage much higher bandwidth than PCI, otherwise it'd be limited to PCI's maximum speed.

    And you're right, the bandwidth numbers i punched out are flawed, in that you'd only ever see that if you were basically displaying random noise on your screen. On second though, photo editors would also bump into bandwidth limitations when scrolling through large documents.

    If and or when that were the case, then a PCI card would indeed pull down the rest of your system as it tried to fill it's buffers.

    I've always been really disappointed with AGP, in that Intel originally promised so much more (shared memory, etc...) but ultimately only delivered a faster pipeline.
  • Unlike every OS out there aside from Win 9x, Win NT, and the Mac, Windows 2000 will be able to run Microsoft Office. There will also be BackOffice Server, soon enough. I believe Oracle is bringing Oracle8 to 2000, as IBM is doing with DB/2. Lotus, however is not bringing it's office productivity suite to 2000, or if it is, it isn't getting it certified. Say bye to another MS Office alternative.

    Joe consumer doesn't care about memory footprints or anything. They'll just go buy a new machine to run 2000 on.

    Though yes, I think Intel's doomed when Merced finally arrives... First it had no backward compatitbility, meaning that it would have been completely dead in the water. Now they've bolted on an x86 core, which means that there'll be little incentive for a LOT of apps out there to get ported to it (we don't really need 64 bit browsers, word processors, or email clients), except for non ported tasks, it'll run slower than any other chip because of it's lethargic clockspeed...

    In the mean time, AMD is going ahead and adding their own 64 bit extensions to the x86 architecture. I wonder what the new monicker will be ? Wintel = AMDSoft? Winamd? Winalt?

    I don't know...
  • But Linux is pretty much 2nd tier across every arch. except x86... Yeah, it runs on PowerPC's... Yeah, it runs on SPARCs... But where are applications like Oracle, Sybase, DB/2, WordPerfect for those archictectures? Linux needs those to win it's "war"...
  • So far as your PCI-to-AGP thing goes....

    IF you were running your screen at 1024 by 768 at 32-bit color depth, with a 75 Htz refresh rate, you
    re already moving more data per second than PCI can handle (according to my calcs, that's >160 MB/Sec - PCI does 132 MB/s)... But video is really the most demanding operation in a desktop computer.

    Switching your video card to an AGP one was a great move, because you've moved the graphics into another bus. Aside from that typical bus usages would be:

    CD quality sound: 176 KB/Sec
    10 base T ethernet: 1.25 MB/Sec
    100 base T ethernet: 12.5 MB/Sec
    UW2SCSI: 80 MB/Sec
    USB: 1.25 MB/Sec

    So basically, with video on a separate bus, you can completely saturate 2 100 Megabit connections, while churning away at disk array, type constantly, and listen to several streams of CD Audio...
  • I wonder how fast someone will try and overclock one of these...
  • I seem to think someone suggested that Tensing was actually first up. Not that it matters much, neither of them would have got to the top w/out the other.
  • The general public has a fear of overclocking.. pretty simple.
  • Do you honestly think that anyone was confused in any way about what the unit of measure was? I understand "correct is correct," but c'mon! I don't even know why I'm responding, except I'm on vacation and bored. :-)

    Now, if the unit of measure was meters, then the case could be confusing, but I think everyone can figure out that mhz, MhZ, or even mhZ all really mean MHz.

  • I remember getting a brand-new 486DX33 in 1992 and being the guy with the fastest machine on the block, so 10 years ago, 100 Mhz was unheard of!

    If we've gone from 33 to 1000 (or 1050 in some cases) in just 8 years, imagine what will happen in another decade.

    Mr. Jetson, you have a call...

  • His claims are far from outrageous. That you look to Sharky for technical insight doesn't rack up any points for your team, either.

    If you like benchmarks, you should look here [anandtech.com], here [gforces.net], here [gamersdepot.com], or maybe even here [gamepc.com].

    The Coppermine has a few technical advantages over the Athlon, and even outperforms it on most platforms. The Athlon still suffers from a selection of mostly sub-par motherboards, and Intel's 820 is a dud. There are no great chipsets available to the motherboard manufacturers right now (though Via's KX133 is off to a good start) -- a "wait and see" attitude is probably the best thing for us about now. Poot.
  • It's not '1000 mhz', it's 1000 MHz. 1GHz is more correct still. 1000 mhz isn't in any accepted units at all, and 1000 mHz is 1000 x 0.0001 Hz, which is a very silly way to express 1Hz.

    Surely slashdot readers understand that case is significant!
  • 1) About PCI: Yes, I know there are 64bit/66Mhz PCI motherboards but cards working that way are hard to find. Plus, while I don't know the gory details about PCI internals, I experienced good speed improvements after replacing my PCI video card with an AGP one... maybe I'm wrong again but this seems a PCI limitation to me because the only element I changed was the video card and I don't think there's a noticeable difference between a Matrox Millenium and a GeForce in pure 2d drawing speed!!
    2) About memory: yes, Rambus is not bad in itself: it's how Intel drove the whole thing that makes me laugh (faulty chipsets etc. etc.)... and there's the latency issue: I think DDR-RAM is more intresting but cuold be a matter of taste :)) Ah, with this one I didn't want to compare apples and oranges (that's what you said :), I just wanted to point out that while companies are struggling about megahertz they are doing nothing to improve PC architecture, nothing besides this Rambus thing that will help only when prices will drop and when Intel will do a good, reliable chipset like BX.
    3) About s-bus: you're right, I'm reading Ultra2 specsheet and the s-bus is indeed very close in design but inferior in performance to PCI: I was thinking about Sun's UPA (that seems more like a "multiple AGP with arbitration controller"), and Intel's NGIO: I think this will be the future, but frankly it's too expensive nowadays...
  • by / ( 33804 )
    Every schoolchild recognizes Neil Armstrong as the first man to walk on the moon.... Nobody remembers who got there second...

    Every schoolchild knows that Bull Aldrin [buzzaldrin.com] got there second, if not because they actually care about learning such things then because they watch tv and have seen the NASA commemorative coin commercials he did a few years back. Celebrities are notorious for extending their face-time with endorsements -- how many people do you suppose only recognize Shatner as the Priceline guy?...
  • "Ceci ne pas un sig."

    That sentence has no verb. That's like saying "This not a sig". Should be

    "Ceci n'est pas un sig."

    Merci.
  • We're not talking about Hertz; we're talking about hertz. (The unit is spelled in lower case, to distinguish it from the man it's named after.)

    And I know you're joking, but it's a bit silly to get the case wrong when replying to a post pointing out how slashdotters should understand case is important.
  • Why won't AMD and Intel stop with this stupid mHz wars and actually worry about quality? We say what happens to Intel when they rush a product (Pentium I's)... Lets hope AMD isn't prone to the same mistakes. But seriously, now that the race-to-the-gigahertz is over, PLEASE, AMD-- Start concentrating on developing that powerhouse chip of yours: Get a chipset out that uses DDR-SDRAM. Develop a chipset that takes advantage of Athlon's ability to have the 64 processor SMP that you hypes so much when the chip was first announced. Make a chipset that's worth a damn and don't rely on VIA to cover up your crappy chipset.

    Desco
  • Apparently its important to AMD, here's a quote from the press release:

    "Every schoolchild recognizes Neil Armstrong as the first man to walk on the moon, Roger Bannister as the first to run the four-minute mile, and Edmund Hillary as the first to scale Mt. Everest. Nobody remembers who got there second..."
  • I wonder how much of an impact this will have on performance, compared to Intel's Kamati offerings with a 1/2 speed cache? Also, does anybody know what the multiplier is on Intel's newest P3's? I can't remember their stupid name.

    In fact the latest Pentium III (which are replacing the old ones) are codenamed Coppermine, and the cache is not 1/2 speed of the CPU but full speed (the size is only 256 KB but this still makes it faster than the old Katmai core)
  • From the Article :
    Pricing and Availability
    AMD is currently shipping its 1GHz AMD Athlon processors priced at $1,299 in 1,000 unit quantities. AMD is also announcing the availability of 950MHz and 900MHz AMD Athlon processors. The 950MHz AMD Athlon processor is priced at $999 in 1,000 unit quantities. The 900MHz AMD Athlon processor is priced at $899 in 1,000 unit quantities.


    I don't know about everyone else, but that's a little pricey for a processor. You can build a decent full system for that kind of money. The benchmarks are pretty decent though. [sharkyextreme.com]

    The price to stay ahead of the Jones' isn't moving much, even with heated processor competition.

    //Phizzy

  • Just as the achievement of Chuck Yeager signaled the beginning of a new era in aviation, the
    1GHz processor ushers in a new era of information technology. AMD plans to lead in the gigahertz era.


    I don't know... certainly the x86 tech that permitted AMD to enter the "gigahertz era" won't take it much farther. I'd compare this to a 10Mhz 6502 - extremely interesting for its time, but less a presaging of things to come than a logical endpoint to an old tech.

  • 1) The Athlon 800 has the same cache divider as the Athlon 1Ghz.
    No it hasn't...

    The 800Mhz Athlon has a 2/5 cache divider, which yields a cache speed of 340 Mhz.

    The 1Ghz Athlon has a 1/3 cache divider, which yields a cache speed of 333Mhz.

    The effective cache speed of the 1Ghz chip is thus slower than the cache of the 800Mhz chip.

    Your claim is thus blatantly wrong.


    Okay... I'll do the stupid things first, then you shy people follow.

  • Well, AMD rightfully thinks it is important because people can go to COmpaq or Gateway *today*, custom build a system at 9xx+ mhz, and have it arrive within a month. A ghz system is probably price comparable with a Intel 8xx mhz system due to use of RDRAM, and the shipping times are probably about the same .

    As an aside, Dell couldn't manage to get 2 P3 700 systems to a client at once. A 700 and all the 600s arrived, but the second 700 took two more weeks.

    matt
  • by nutty ( 70104 )
    Damn

    My 900Mhz cordless phone isn't the fastest anymore!

    I'll have to upgrade to the 2.4GHz phone.

    Megahertz is everything, right?

    /nutt
  • Its called the Network Effect.
    Lots of people using something increases the value of that good. Thats the only reason that x86 has survived for sooo long, and cause of its market penetration Intel/AMD are not about to switch to a new instr. set! Imagine people thinking about a new CPU might decide to look beyond xx86 and may go with something other than Intel/AMD!!!
    The same freaking reason with OS's!!! But i think Win2k HAS to be the answer!!! I mean with its memory footprint i think people may just try other alternatives (Be, FreeBSD, Linux)
  • It was my understanding that Athlon Chips were at 100Mhz bus so a 1Ghz would be 10.0x 100=1000Mhz. I believe (though I don't know for sure) that most Athlon motherboards only have 10.5x as the fastest mulpilier so unless you increase the bus speed, which I believe is not supported by the motherboard, or the overclocking cards the fastest you could get is 10.5x100=1050Mhz. Not great for over clocking.......
  • i just heard this on the radio, looks like they beat Intel. Go AMD! :) -motardo
  • Well this got rated as funny, and possibly rightly so. Reducing speed to "n times faster!" is pointless.

    But my G4/450 has gotten 4+ megakeys/sec on RC5.

    Where is my mind?
    mfspr r3, pc / lvxl v0, 0, r3 / li r0, 16 / stvxl v0, r3, r0
  • Wow... I never guessed these processors were so $$$$-y. The G4, as I remember (I can't find pricing info any more... odd...) is about half as expensive or less, at half the clock rate.

    Where is my mind?
    mfspr r3, pc / lvxl v0, 0, r3 / li r0, 16 / stvxl v0, r3, r0
  • Read this message aloud in a James Kirk / William Shatner voice.

    Where is my mind?
    mfspr r3, pc / lvxl v0, 0, r3 / li r0, 16 / stvxl v0, r3, r0
  • My motherboard (an Apple G4) has a software-controlled L2, so once I wrote a mini-program to determine just what the cache does. It read 512k of memory over and over again (the cache is 1M, but it's no different). It turned out that setting the cache to 150MHz doubled or tripled speed, but increasing it all the way to 300MHz (overclocked) only made a 10% or so performance gain.

    So, the main advantage of cache RAM is latency, not transfer rate.

    Where is my mind?
    mfspr r3, pc / lvxl v0, 0, r3 / li r0, 16 / stvxl v0, r3, r0
  • that should be increasing it to 150MHz *from being turned off completely*, and a 10% gain over the 150MHz speed. Sorry...

    Where is my mind?
    mfspr r3, pc / lvxl v0, 0, r3 / li r0, 16 / stvxl v0, r3, r0
  • I stand by the statement. Look at Anandtech's latest review of the 1 GHz chip [anandtech.com].

    I picked the SYSMARK2000 to demonstrate my claim but look at them across the board.

    PIII (800) BX PC-100 Ram - 30.8
    Athlon 850 AMD-750 - 30.8

    Sure, if you enable superbypass (not supported by my K7M motherboard in the revision I bought) or use a Via KX-133 chipset motherboard you can eke out a few more points. The PIII can also get a small jump with RDRAM but it costs a fortune so I'm not using it for comparison.

    You busted me. I work for Intel. Read my previous posts.
  • I'm just waiting for the new 1000uHz processors!!! Oh BOY!
  • by Anonymous Coward
    I was going to buy one of those 900MHz cordless phones, but I think now I'll wait for the 1Ghz ones, I think they dial faster or something...
  • So did the k6. It's just different from Intel's. To the best of my knowledge, noone ever built a chipset for dual k6's. The athlon bus is licensed from alpha, so maybe this time around . . .
  • No, the Athlon is NOT an implementation of the x86 hardware. Furthermore, neither is the x86 (at least anymore).

    Ever since the K5 and Pentium Pro, the chips have shared nothing in common with the 8086 except the instruction set.

    I agree with you otherwise, though. Code morphing and other hardware-assisted JIT techniques really do look like a major advantage.

    -Billy
  • Sounds like Sun needs some competition in their sparc suppliers (and/or design teams). Especially if they want to move folks to mostly interpreted code in servers.

    Sun has some. TI makes the UltraSPARC. Fujutsi (or Samsung?) is working on the HAL SPARC V9 (and has been for some time). In the past Cypress, Ross, Fujutsi, and Samsung, and Mekio have all made SPARC CPUs. Sun publishes the SPARC ABI and ISA, and promotes compatation. They even publish the bus specs.

    The problem is componies don't see as much profit in that design space as in x86 compatables. There is a lot of money to be made there, but if Sun doesn't pick you to make CPUs for them, you end up in SPARC clones, or 3rd party CPU modules, and there just isn't the same kind of money there.

    And Intel & AMD's IO seems to be coming along, including memory bandwidth (> Sun's IBM's). Especially when you divide the number of processors into the memory subsystem bandwidth. And Sun is now says they are following Intel's lead on next generation IO?

    Sun switching from SBus to PCI isn't the same as Sun following Intel on I/O. Sun still does their own memory systems (well low end boxes use PC memory because that is the only way to get prices even remotely close to PCs).

    Morover PCs almost allways have one PCI bus, plus one or more bridged PCI busses. The bandwidth from the CPU to any pair of PCI devices is limited to half the PCI bandwidth (half for one, half for the other, and in reality some switching cost). Sun's high end machines (even their midrange) have multiple independent PCI buses, so you can push a Ultra160 SCSI at 160Mbytes/sec on one PCI bus, and not interfere with bandwidth on another PCI bus. The Sun SPARC4500 can have 4 diffrent sets of PCI busses (the 4500 is also one of the last SBus machines, what it really takes is up to four backplane bords which can have memory and CPUs and/or PCI and SBus slots; each backplane bord communicates over a very high speed bus the "FHB")

    PCI (esp. the 66Mhz 64bit kind found on many Suns) has enough bandwidth to handle most all perhrials, and is cheep. So it is a fine bus for most things. Why is Sun wrong to use it?

    Why are the Sun boxes still popular for non-scientific work (perhaps even scientific)? There are lots of faster Unix boxes (and it sounds like even Linux on PCs runs rings around equivalent systems - given the same number of processors and memory and disks). It is hard to find benchmarks for comparison (can't blame them).

    At the low end I susspect:

    • Reliability. The things are way more reliable then PC servers. The hardware that is. (well I donno about Suns UltraCheep boxes, I have no Ultra5/10 experiance)
    • Easy of assembly. Ever used Suns SCA drives? Don't you wish PCs had them? No messing with cables, pull the lever up toss the old drive, put the new one down there, push the lever, drive installed. No cable mess. No mounting screws.
    • Name brand
    • "We have other Suns"
    • Lots of software

    At the high end

    • I don't care that you can get 4 800Mhz Intels in one box, i have 40 400Mhz CPUs! Ha! Ha! Ha-Ha!
    • Have you seen our TPC-C numbers?
    • If you want a big Oracle system, well there is really only one place to go (and have you seen our TPC-C numbers?)
    • Did you notice we are slowly becoming mainframes (and have you seen our TPC-C numbers)?

    Or at least that is my guess, I don't buy big Sun boxes. Just the medimum ones.

  • re: chip competition. doesn't seem to have worked even though you'd think it would be in their best interest not to be trailing the pack.

    It doesn't seem to have put them in first, second, or even 3rd place in CPU speed. On the other hand maybe without it they would be dead, and dead last. After all Sun was never focused on CPU design, they only did the first SPARC in house. DEC had been good at it over twice as long as Sun had been in existance when that stratagey was picked. Sun had no fab plants then either.

    Then again maybe if Sun had decided to build a pool of talent and do it in house the SPARC could be at the top of the SPECfp list, not the Alpha (or does the 1Ghz K7 manage to beat the 700Mhz Alpha? I doubt it, the Alpha had a 3x SPECfp lead last I saw).

    I don't think any of these companies had a chip that wowed folks at introduction (does this mean there's something wrong with the basic processor architecture that constrains speed so that none could succeed?).

    The first SPARC, the 4/110 and 4/220 were three times faster then machines costing five times as much. The were a revolution. The didn't wow folks, they stright out floored them. At least the ones that didn't think it was a lie. They made no impact on the PC market since they were $10,000 machines.

    The first really popular SPARC the one in hte SparcStation1 was again so much bloddy faster then anything anyone else had that DEC had to drop their plans to design a RISC CPU in-house and start the fastest workstation design-to-market project ever done (I think it was a little under a year, or a little over, I forget which).

    When DEC finally brught their MIPS baised machines to market and edged out past the SPARC Sun brought the SPARCStation 1+ out (as an in-line upgrade -- existing SS1 orders were shipped the SS1+ at the same cost!) forcing DEC to drop it's brand new DS2000 because it was laughable next to the 1+ (or maybe this was the 2, I forget).

    It was only after DEC managed to bring the Alpha out that they managed to beat the SPARC, and keep beating it for the rest of the decade. Not bad for a little upstart workstation peddling "snake oil" and hoping to one day "piss with the big dogs".

    The SuperSPARC was impressave, but not ground breaking. The MicroSPARC was pretty wowing, if you were intrested in low cost CPUs (it was very cheap from the very start -- and not too slow). The Ultra1 and Ultra2 were not awe inspireing. The SPARC-V9-US3 is not wowing in terms of clock rate, but in terms of L1 and L2 load to use latency they are indeed wowing. But it is Mhz that makes the headlines. Even if SPECint/SPECfp would be a better set of numbers to chase.

    Too bad CPUs cost so much to make, it would be intresting to see what would happen if a compony designed an extreamly long pipelined CPU with a fantastically fast clock that wasn't really all that fast (the long pipeline would make a fast clock "easy", and at the same time make it impossable for the CPU to actually be fast without monster good load bypassing and branch prediction). Would it sell great because it's clock is 2x to 3x to 4x as fast as everyone else's, or would it flop because it would be slow as a dog on any real code?

    And wasn't it 5 or so years ago Zander told anyone selling Suns they'd forfeit their contract if they stocked things like Solbournes(sp?)?

    Probbably more then five, I think Solborne was only around from '91 to '94. The Solborne was a competing SPARC baised system, not a new SPARC CPU Sun could put in their box. Worse yet it did multi-CPU suport much better then SunOS did (Solaris wasn't out at the time).

    It was probbably a dumb move (long term), but it wasn't quashing a rivel CPU. they even used the same CPUs most of the time.

    re: TPC (?). Last I looked at http://www.tpc.org Compaq and IBM solutions had it all over the big Suns (in the top 10 performance leaders). Granted the best Sun number shipped late last year and the IBM & Compaq shipments come later this year (so Sun may be able to retake the lead before they ship).

    I assume that is true, but I don't know for sure. Also while the Compaq's are priced similar to Suns (I think) the IBMs cost a lot more. Then again they have a better rep for reliability.

    Though it does look like Compaq can just keep on adding machines (rather than telling the customer "sorry, your problem just got too big".) Seems they get linear scalability going from 32 to 64 to 96 processors.

    I doubt it keeps scaling linerally after that -- unless they had all 10 top 10 spots they would have kept showing better systems. The PR coup of holding the top 10 spots (or top 5!) would make it worth it. Even for an expensave benchmark like TPC-C.

    [lots of other Big Compaq better then Big Sun quotes]

    I have no idea. We have strayed far from my area of knolage. I will say Alpha kicks some serious ass, and I'm not supprised you can make nice systems from them. But it is really a market where I don't buy machines, I don't evaluate them, and I don't really understand the needs of. I'm a "small server" guy myself. If I can't lift it, I probbably don't run anything on it. (note, I can lift systems one at a time and still get a rack full of small servers).

    re: reliability. The big Dells we have have not had a cpu die that I know of, but I know our big UE has had multiple processors replaced, and NICs die, I think. Ok, it is a little older (but I don't think I've ever seen the even older pentium Pro 200s fail). What's your experience with the Sparcs? I suspect they run too hot for their cooling (or need more filter maintenance, or they are really serious about their humidity requirement - so the air is not too dry).

    I havn't had a CPU or NIC on the SPARCs go bad yet (I have had a DOA or two shipped to me). I have had a DEC NIC go very bad (caused the machine not to boot), one of hte DE-100s. It happened to be in my home machine though, so heat/dust/humidy wasn't the same. All my DECs are PCs. Most of them ageing. I've had drives go bad on both (many more on the DECs, but that is the fault of the HP SporeStore drives). The DECs lock up randomly sometimes (like once a month per 100 machines), the SPARCs get an ECC'ed bus fault less then once a year per 100 machines, and the OS (not solaris, same OS on the PC and SPARCs) re-try code for that didn't work last time it happened (it has happened maybe three times). Most of the SPARCs are newer, but some are quite old SPARC20's with 3rd party ROSS HyperSPARC modules.

    The PCs are all in a nice machine room. Some of the SPARCs are in a not-so-nice Telco-co-lo (well, Ok, it's pretty nice too). I would never do that with a PC. We have done "hands off" OS upgrades on the SPARCs (one of the benifits of being a small server guy is I can send all the load to machines B and C when i take A down for an upgrade). Literally we schedule someone to put the new OS CD in the SPARC, and then at out lesure we schedule the upgrde and do it from halfway across the USA with nobody at the facility.

    Christensen in his Innovator's Dilemma implies that all the "ilities" follow investment which follows volume times price point at a given interface. So the mass market stuff has got to have better reliability (else it would not be mass market).. (and the maintenance fees would be outrageous - and they're not). where appliances are the example. 10s-100s of millions of anything have to be more reliable 10,000s of something (else there are not enough phones in Baltimore to answer the calls).

    I beleve that as a genneral rule, but there are niche markets where reliability is the goal, not price. I totally expect a IBM390 to be far more reliable then the best x86 machine built. Even if part of that is only that the 390 is designed to detect the error, and let you replace the part with no intrruption of service (or minimal).

    Also the SPARC and Alpha and mainframe buyers take much more time on the phone. They won't hang up until the problem is solved. PC buyers have been conditioned to take various forms of "you'll have to upgrade", "that isn't a supported configuration", and "oh, that's Microsoft's fault". They don't insist you give a root cause failure analisis.

    I do belve he is right about the ford Mustang being less prone to brakage then a Lotus, or Ferrari.




    This has been a long reply. Hoe someone reads it :-)

  • the AMD 1Ghz just barely outperfom the Kyrotech 1Ghz chips. When you look at the cache, the kyrotech is an 800 upped to 1G, therefore the cache, which was at 320 is at 400 in the kyrotech, vs 333 in the AMD chip. What else was changed in the AMD chip to make it outperform an older version of itself with faster cache?

    As far as I can tell, nothing. The motherboard the air-cooled 1Ghz part ran in has a better chipset (the KX133). If you look at page 3 air cooled part is shown with the KX133, and 256M of PC133 memory. If you happen to magically know what the Kyrotech part is (the benchmark didn't say) it is one of the old AMD750/751 baised motherboards, like the second set of systems. It doens't support PC100 memory, and it had may not have supported the PC100 memory as well as it could (many 750s can't use "Super Bypass" mode, which as far as I can tell is a way to skip 75% of the latency in the common case, without getting the wrong answer sometimes).

    In other words this is kind of a lame CPU benchmark since the systems were not made as similar as possable. It is a fine system benchmark, since you won't get the Kyrotech in any other motherboard. You probbaly wouldn't want to put a new AMD in a non KX133 system. Still the totally diffrent hard drives and sound cards and stuff, and diffrent amounts of memory arn't a good idea to get a isolate component test! Maybe not even for the system test (what gamer would spend over $1000 on a CPU, and stick with crappy 2 chanel on the motherboard sound chip?)

  • The benchmark makes a big deal of the slow L2 cache. It doesn't mention that the L1 cache is 128K, quite a bit larger then the P-III L1 cache (which I think is 32K). So when the Athalon gets the faster L2 cache, while I have no doubt it will help a great deal, it won't help as much as, say going from no cache on the Celeron to a 128K cache.

    The P-III also has a much lower latency to it's small L1 cache, and to it's decent sized L2 cache. That can make a big diffrence for anything that chaces pointers down a linked list, or any other extreamly latency sensitave application.

    All that siad, I really like the Athalon, it is a good value for the money. My new Unix box sitting next to my desk here is an Athalon (only 650Mhz, I figure I'll pick up a DP system later in the year if I'm lucky).


  • Basically, the world has changed slightly, and the x86 people now have the best process technology.

    Intel and AMD have 0.18u; Alpha and HP and MIPS are still on 0.25u, and Sun will get to 0.25u with the Ultrasparc 3 this summer.

    Alphas are still a lot better at scalar FP per clock (with two FP units and without that dumb register stack), but if you're doing single-precision work a P3 or K7 with SIMD instructions will be as capable as a $5000 21264 machine.

    HP's PA-8600 chip is amazing; 1536k of L1 cache running at 550MHz (as fast as the L2 on the fastest Xeons around), with a brainiac design even more sophisticated than the 21264, delivers ridiculous speed for that clock rate. But at the price, I'd still rather have a decent second-hand car ...
  • The prices of the older chips will drop with the introduction of the higher rated chips. If you pay attention you'll find that the high end systems always stay about the same price, just the speed increases and that trickles down the family line. 2nd fastest will be 80-85% of fastest, then 75%, 60%, 50% and so forth. We can see some of this in the announcement 1000mhz (1ghz) @ $1299, 950mhz @ $999, 900mhz @ $899 and so forth. As the supplies of these chips come in, the demand for the available 750mhz chips will drop off (from the system manufacturers) and the onesy twosy prices will drop as well. The onesy market doesn't drive prices, it's the system people that are scarfing up the 1000 piece lots that make a dent in demand. Keep in mind they're announcing a new speed bump, not the availability of more affordable middle speed range chips. Also, look at where the price has gone already on the chips (700mhz) that were top of the line 3-4 months ago.
  • Another beauty of Linux (or xBSD) is that while computers keep getting faster, the OS isn't getting any slower. In the M$ dominated world, one has to constantly upgrade one's hardware to keep the software running at the same speed. Since old computers don't really get slower with time, they are still useful to people like us. I've got a P-100 doing just fine as a remote access / web server, and a P2-266 running really well as a desktop. If I used Windows, I would be screaming for upgrades by now.
  • Seems to me that this CPU MHz race makes no sense at all: until we're stuck with PCI bandwith
    limitations these little monsters will only do more idle cycles, while waiting for data...
    I really hope that the big players will find a new architecture, but something more interesting than the ridicolous Rambus thing:
    maybe they will come out with a solution like Sun's S-BUS, this would really change the PC market!
    In fact, if I remember well, S-Bus is not really a bus but each slot has a point-to-point connection to a dedicated controller
    that handles data without CPU usage: this would be a real change! (but correct me if I'm wrong)
  • the Athlon 800 delivered a severe can of whoop-ass to the Pentium III 800

    Are you looking at the same data I am? On the various benchmarks posted, I see the following Athlon performance, normalizing P3-800/133 to 100%:

    102.6%, 96.0%, 97.6%, 102.4%, 103.4%, 86.6%

    A couple of percent on an isolated benchmark or two hardly amounts to a "severe can of whoop-ass." If anybody thinks these benchmarks show a significant difference between Intel and AMD x86 chips, they're kidding themselves. First, benchmarks are imperfect approximations of real-world performance and expecting a few percent difference to apply to the real world is naive. Second, in my experience, it takes about a 15-20% difference in speed in interactive applications before a computer becomes noticeably faster. Any speed difference less than that is only useful if you're running some kind of loooong compute-bound task. 10% might be worth if if you're running a render farm or doing weather simulation, but then you wouldn't be building a render farm based on Quake III results.

    Those of you who are either an Intel x86 advocate or an AMD x86 advocate needs to realize that you're both in one narrow corner of the CPU architecture world. It strikes me as like arguing over exactly which Corvette options make the best car while ignoring Ferraris, Porsches, etc.

  • The 1 GHz Athlon is basically a crippled chip

    Well, if so, then it's a "crippled chip" that benchmarks faster than any processor Intel has ever released.
  • Bear in mind that the Athlon uses SDRAM, not the PIIIs Rambus memory, so a 128K Athlon system will be $500 or so cheaper based on that. Of course a lot of these premuim 1GHz systems will be used as servers with 512K or more....

    Anyway, seeing as Intel don't expect to ship production quantities of the 1GHz PIII until Q3, Athlon is the only game in town, and they in the nice position of being able to enjoy that end of the price curve!

  • Consumers have known for a while that AMD has the fastest processors out, but Intel still has the reputation of being market leader.

    Given AMD's stock rise today, shareholders appreciate AMD delivering on their promises...

    The announcement isn't that empty either - remember that AMD's policy is to announce when then can deliver in production quantity... Of course Intel will now "announce" the 1GHz PIII in the next day or so, but by their own admission they won't really be able to deliver it until Q3.
  • ... it's only 3E8Mhz. What? Is slashdot now going to announce every new chip in the world? Wait at least for 400Mhz part. That would be at least somewhat round number.

    :-)

    Gee. Seriously, great job from AMD! Greets. A bit expensive though.
    BTW: When you think we gonna have Socket (or whatever) Athlon parts for 50-80 bucks?
  • > "I wonder how much of an impact this will have on performance, compared to Intel's Kamati offerings with a 1/2 speed cache?"

    I think larger L1 cache will compensate it.
    BTW - it's strange how people only keep in mind L2 cache when talking about caches. Considering that L1 cache does 90% (or so) of the speeding job.
  • Oh, wait, maybe you're right. Seriously, I could care less about a 1GHz CPU, what I want is on-board 1.44 ADSL with a monster cache and scalable super-cache CPUs (say 500MHz). Now that would rock!

  • Yes, Intel is a Big Bad Multi-Billion Dollar Corporation. And the x86 archietecture has had a deservedly bad reputation since before most Slashdotters were born. What's humorous to me is that the teeming geek masses have gotten behind the AMD like it's a real alternative to the situation. AMD is not exactly some guy hacking hardware in his house for the good of the people. The Athlon is still an implementation of the x86 architecture.

    The advantage of AMD is that they're providing an alternative to Intel, but the victory is slight. AMD isn't running down a new road that will give them a big advantage over Intel; both companies are pacing each other, and AMD has to keep the prices down in order to get anyone to look. Companies like Transmeta have a bigger opportunity, because they can do things that Intel simply can't do, like running at 1/25 the power consumption and 1/10 the price. Not that I'm a Transmeta fanatic, thank you, but I think they're the type of company that could have a more notable effect on the future of CPUs.
  • For us to have competition, AMD needs to partake in the marketing war. End of story. Having successfully done so (so far), their products are become of better quality, as their Athlon revisions with full speed L2 cache will probably debut next quarter. This will probably dovetail nicely with SMP Athlon motherboards, of which I have heard that the first may debut in the 3rd quarter.

    matt
  • There's three kinds of lies - lies, damned lies, and benchmarks.

    But anyhow, on the benchmarks you linked to, the Athlon's cache only affected it on games and synthetic 3D benchmarks. On professional level applications, the Athlon 600 was faster than the P3-800.

    I mean, who would want these for games, anyway? On demanding games, it only got 98 frames per second! Seriously, at the $1299 price point, the main purchases of these things should be professional apps. If you're playing games, either get a slower Athlon with an overclocking card, or wait for the Thunderbird with full speed L2. Or get a 1 Ghz Athlon, so you brag to your friends, and put up with that painfully slow 98 frames per second.
  • From AMD Zone [amdzone.com]:

    AMD held a very brief conference call this morning. I actually got to pretend like I was an important person and listen in. Compaq and Gateway have dibs on all March shipments of 1 Ghz processors. Everyone elso can get them in April. Also, here's the new pricing:
    1000 Mhz:$1299
    950 Mhz:$999
    900 Mhz:$899

  • If you've been following the analysis of AMD v. Intel in places like Tom's Hardware [tomshardware.com], you know that MHz isn't enough for an ultra-high end system.

    You also want to support lots of RAM. But the motherboards most places are selling (including Gateway's 1000MHz Athlon system [gw2k.com]) are limited to either 384M or maybe 768M of memory (the VIA chipset, like on Tyan's S2380 [tyan.com], is a good example of a high-end board for the Athlon.

    On top of that, there's still no multi-processor (forget that....how about DUAL processor?) motherboard for the Athlon. You can get dual processing MBs for Pentium-III's cheaply, and >2 processor MBs for Xeons, if you want to pay the price.

    Just wanted to mention.... of course, even the non-Xeon Pentium-III has relatively few motherboards available that will support over 768M of memory, but you can go to the the Xeon and get MBs with up to 2GB (easily) 4GB (just becoming available from Tyan and others). 1GB is available for regular PIII's from several vendors.

    Disclaimer: My system is a Athlon 700MHz. It rocks.

    PS: Gee, Compaq: You'd think that when you issue a press release [compaq.com] about your new system, you'd actually be selling them [compaq.com], but you're not (at least on your Web site). Gateway is....

  • Well, considering what Intel has been doing since AMD beat them in the clock-speed race and every benchmark race there is, I'd say that Intel is definitely just shooting for the MHz and not even caring about whateher the chips melt in their sockets (the 600s) or even boot!(the 800s)

    But, don't get all worried about the 1/3cache like everyone else is, just overclock it to half and it'll beat anything Intel throws on the market for 3 months.

    Esperandi
  • Yeah, and no one cared about when the Athlon hit 600 before Intel. Remember, that's why Andy Grove put his engineers under the whip and produced a nice little processor that promptly overheated and melted in the socket. Sure, they fixed it with a patch later (and reduced its performance significantly), but I think this is a good sign that AMD should try to fluster Intel all the time.

    When you sit back for years being the biggest kid on the block used to releasing processors at your leizure and leading the speed race by a mile, you get fat. And a little later on down the road, you get your ass trampled by the little skinny kid who has been running his ass off and before you know it you're sitting there with an 800MHz chip which runs slower than the kids 750 which doesn't even really matter anyways because he just released a 900, a 950, and a 1GHz.

    Everyone is railing about the cache. It's still going to be faster than the slowpoke P3.

    And all you anti-corporation nazis take note - this is how monopolies are tumbled!

    Esperandi
  • 750 doesn't support PC100 memory? Doesn't support Super Bypass? Huh? It supports both. It didn't support super bypass when it first came out, but they do now.

    Somethnig else to consider:
    1.25x800 = 1GHz
    1x1GHz = 1GHz

    Lower multiplier to overclock. The multiplier on the processor is also probably different.

    People seem to forget that a benchmarked chip is almost always slower. A chip running 100MHz x 4 = 400MHz is slower than a 200MHz x 2.

    Esperandi
  • Sigh, someone always brigns this up and they make themselves look like a complete tool. The Pentium 166 ran about $2000 in quantities off 1000.... the week after it was released. The week after a processor is released, it is SKY HIGH in price - it halves in 2 months. It happens this way with each and every single chip. The Athlon 500 I've got opened up at around $900, it plummeted in price right afterwards.

    The bottom line is do not trust price estimates unless you're an OEM who *MUST* have the fastest thing the split second it is released. Otherwise, just wait until the thing is not only shipped but in the hands of retail vendors, then hop on pricewatch.com every day forr a week and watch the prices fall through the floor.

    Computers are getting cheaper and cheaper in every aspect, only people who don't realize this just aren't experienced with watching processor prices...

    Esperandi
    (BTW, a P3 500 Xeon with 2MB of L2 cache costs $1900 *RIGHT*NOW* after its been out for ages, if you want expensive, there's the "other guy" to look at ;)
  • Ultra-ATA/66 blows away most SCSI alternatives in both price and cost.

    Wow, and my new car beats yours in both economy and miles per gallon!
  • Because you asked:
    SGI's current processors are clocked from 225 to 300 MHz, while Sun's UltraSparc II(i) is clocked from 248 to 480MHz.

    But you recognized it:
    clock ratings are little more than marketing hype.
    As you can see from the AMD-vs-Intel battle and Cyrix's PR-ratings you can't even compare MHz-ratings of x86 processors, let alone those of different architectures.

    Furthermore, in the enterprise-class server market it isn't all about CPU speed, it is I/O throughput which is way more interesting.
  • I think larger L1 cache will compensate it. BTW - it's strange how people only keep in mind L2 cache when talking about caches. Considering that L1 cache does 90% (or so) of the speeding job.

    Not sure where you are getting your numbers on this one, but keep in mind that no matter how fast the L1 cache is ([almost?] always the same speed as the processor), the L2 cache has to feed it. So if you figure a 128K L1 cache running @1GHz, and a 256K L2 cache running @333MHz, you can see where the problem comes in. Unless the entire application (read _very_ unlikely) can fit in the L1 cache, not even counting the OS operations/data, then you will see a significant slowdown as compared to a full speed L2 cache. What everyone shouldn't forget is how memory works, even if it isn't used more than once, each piece of data has to travel through each type of memory (main->L2->L1) before it can be processed.

    Just my 2c.
    The "Top 10" Reasons to procrastinate:
  • Maybe "full of old ideas" but not "insightful"... Sure, it used to be easy to show how RISC was more elegant than CISC but that was years ago guys. There have been hundreds of improvements to the x86 line that are quite innovative. Sun doesn't advertise it's CPU speeds because they're not so great. Unisys kills Sun's UE10000 with it's new Windows 2000 offering. Slamming IDE? Again, this isn't the IDE of old. IDE has had huge innovations. Ultra-ATA/66 blows away most SCSI alternatives in both price and cost.
  • You're right in being suspicious about Intel and AMD cutting corners, because they are. Right now, the MHz war is all marketing. That little speed rating marked on their CPUs make a large difference when it comes to selling machines with tier-1 computer manufacturers, which makes a large difference on their quarterly revenue sheets. This doesn't mean, however, that Intel and AMD are resting on their laurels because they can simply up the speed on each new processor by 50MHz easily. What goes on in the background is their roadmap for the future, like the Willamette and Thunderbird processors for instance, or the Itanium and Sledgehammer in the more distant future. There are long term advancements in CPU technology to be had, but they won't make short-term ideals of their shareholders happy if that's their main focus. Just look at where AMD was last year for proof of this. AMD was developing their next mother-of-all-processors, the one and only Athlon, which was to settle the playing field. However, it was only vaporware until August. They didn't have anything to keep them too profitable at the time. The K6-2 and K6-III weren't doing very well, and AMD couldn't just do something like release a K6-2 that was 50MHz faster to keep them going, because hardly anybody (except for the strictly economy-minded people and manufacturers) wanted them. Therefore, AMD and their shareholders lost quite a lot of potential earnings, even though AMD had the promise of the Athlon coming on the horizon. Now the picture is different for AMD. The market is eating up their Athlons like crazy, and they know what may be a relatively small increase in speed and performance will do very well in the market just because of the psychology of the "MHz" number, and the fact that their processor has been so successful. They can afford the time to develop their next innovation in processor technology. And it does take time. I don't mind it at all, I'll be happy as a pig in mud for the next year or so sitting on my Athlon 650, knowing that the GHz CPUs won't benefit me very much.
  • Even if fellow /.ers don't think this is such a huge milestone, AMD certainly do. Reading through that press release reveals just how much both AMD and Intel wanted to have the bragging rights to the first 1GHz x86 chip - likening it in achievement terms to 'breaking the sound barrier' is definitely a little extreme in my book, but I think the general public and the marketroids will have a field day with a 1GHz processor. I suspect that Intel will be extremely anxious to get their own press release announcing the 1GHz Coppermine out as fast as possible now to stop AMD claiming all the glory, but AMD will get some very useful publicity over the next 24 hours. Intel and the Pentium brand are still the CPU type that over half the computer-buying populace recognise in isolation and it is a major publicity boost to AMD to hit the 1GHz milestone first.

    Cheers,

    Toby Haynes

  • You can get phones on the 2Ghz band already.. :-)
  • Anyone who's read my posts will have seen my cynicism over the mhz. race.

    There are a few areas where this will help (high end calculations, etc. But I feel it is aiming at winning over pen pushers.

    In my previous support role out Finance controller was running pretty big calculations on desktops. THey had been informed that memory would allow them to do better and bigger calculations. They got me to order 3x256 MB dimms for each machine. He would not listen when I said that win 95 would never use it and in fact he would probably end up with more resource allocation failures...

    Moral of this, well, competition is knocking down the prices of smaller processors, which is good, but I think the 1000, etc. marketing ploys are aimed at people like that controller.

    I personally have a P166 MMX for my home use whihc does everything I need it to do. I can't play unreal tournament, but, hell, that's what work pc's are for.

  • Well, it's official, you're nobody.

  • AMD dropped prices on the Athlon on Jan 28. They discontinued the 500 chip.

    On Feb 28/29 prices dropped again and they discontinued the 550 chip.

    Will they drop again at the end of this month? Who knows.

    You can find a 550 at a bargain and if you get the right week number, find out what core is inside and overclock that puppy easily to over 700 mhz. Check HERE [mist.co.uk] to see what cores are in which week number production CPUs

    You won't find AMD reducing prices of their high CPUs anytime soon I don't think.

  • The race between the americans and the soviet union to get to space. Soviet beated the americans with the first satellite and first human (altought, by a very thin time margin!). Now that race can teached us that : 1) Being first does'nt necessarely bring success. 2) Being first is good PR 3) Beating the winner afterward is even better PR And that means that Intel can still be the 2000 CPU winner. Only time will tell!
  • Gateway has a full-page ad in the Wall Street Journal today (3/6) touting a system with this processor.
  • by hawk ( 1151 ) <hawk@eyry.org> on Monday March 06, 2000 @06:42AM (#1223475) Journal
    All that time I spent drooling over 4mhz systems, hoping to someday be able to afford 64k of memory . . .

    And now, the "insignificant difference" in the *cache* speed is twice the speed of my dreams, and that cache holds more memory than a room full of computers or a box of disks . . .

    Pass the Geritol and my cane, please . . .
  • Seems to me that this CPU MHz race makes no sense at all: until we're stuck with PCI bandwith limitations these little monsters will only do more idle cycles, while waiting for data...

    I agree that the 33Mhz 32bit PCI bus can be a limitation for some things, like trying to push a gigabit ethernet at a gigabit. I don't see this as the big problem in PC class machines just yet. Maybe in two years or so.

    First off you can get 66Mhz 64bit (or maybe just 33Mhz 64bit) PCI slots in motherbords that cost under $300. That is 2x or 4x the bandwidth. So if 33Mhz 32bit PCI is a "little limiting", 66Mhz 64bit PCI should have a lot of head room.

    Secondly, there are very very few things you hook up to a machine that push past the PCI bandwidth (the gigabit ethernet being on of the few, an extreamly fast RAID controler being another, and it is only a little past the 33Mhz 32bit PCI bandwidth). So you need a case where you push multiple devices to their limit before PCI is the limiting factor (like I want to read off my massave disk array and send it out these two GE ports...). (3D video cards being a modest execption, because they allready have their own "bus")

    Thirdly, and more important, it is main memory that is the bottleneck for most things. It has been for over a decade. It isn't getting better (main memory is getting faster, but not as fast as CPUs are).

    I really hope that the big players will find a new architecture, but something more interesting than the ridicolous Rambus thing:

    From a technical point of view what is wrong with RAMbus? It is 1992 technology that was great then, and great for five years after. The compony that invented it charged too much for patent rights, never managed to hit the mass market, didn't get to drive costs down, and in the end didn't make enough money to plow it back into research to keep up with the entire rest of the world. But that isn't a technological problem, it is an econmic one. More importnatly RAMbus is extreamly intresting.

    Plus if you want bandwith, and a little latency is Ok RAMBus kicks ass. Unfortunitly bandwith isn't really as useful as low latency for most applications, and at RAMBus's current price it is cheeper to build very wide normal memorys.

    Oh, and RAMBus and PCI don't play in the same space. One is a memory system, the other is a perphrial system. It is like saying "Volvo's T5 engine is out dated, they should devlop something like the VW transmition!".

    maybe they will come out with a solution like Sun's S-BUS, this would really change the PC market!

    They did make something like the SBus. It is called PCI. On paper it uses the same FCode (in practice is normally uses Intel x86 assembly). It has vender and device IDs. It has auto config. It is fast (the SBus is 32bits at about 25Mhz, slightly slower then the PCI bus). PCI abandoned the SBus form factor. PCI abondened using strings for dev/manuf IDs. PCI dropped the IOMMU (which was re-invented for AGP as "Intel's innovatave new DIME"). But PCI is relitavly similar, and being designed afterwords it is even somewhat better in some ways.

    Sun has dropped the SBus over the last few (3?) years. They are using PCI. Not just in the low end "PC priced" machines, but even their most expensave machines use PCI for perphrials (they use their own thing for memory and CPU boards, no big supprise).

    In fact, if I remember well, S-Bus is not really a bus but each slot has a point-to-point connection to a dedicated controller that handles data without CPU usage: this would be a real change! (but correct me if I'm wrong)

    I'm afarid you are wrong. The SBus is indeed a bus. many high end SPARC systems had multiple SBus's hooked up over what may have been a point-to-point controler (like the FHB), but i think that was just a bigger bus. Existing high end SPARCs hook backplane bords up over a point to point system, some backplane bords have CPU, others PCI busses, others memory. SGI does the same thing.

    You may be thinking of Intel's NGIO (Next Gen I/O), or the Cisco/20others Future-something-or-oither. Both are extreamly fast serial-ish point to point systems that could have a fast non-blocking switch at the center. NGIO has been "in the works" since like 1995. Not sure when the other started, or when we will see anything.

  • by speek ( 53416 ) on Monday March 06, 2000 @04:45AM (#1223477)
    Yeah, companies like Dell is the most logical explanation for why AMD is doing this so soon. Dell will have to explain why they can't supply a 1GHZ machine while Gateway can. There may be a very good explanation (the AMD 1GHZ processors are crippled compared to what will be out later this year), but how many consumers will understand that?

    This puts pressure on Dell and others who haven't gone for supporting AMD products.
  • by barleyguy ( 64202 ) on Monday March 06, 2000 @08:26AM (#1223478)
    Yes, the Athlon does support Multiprocessing. It's a point to point protocol of the EV6 bus, which gives each processor in a multiprocessor setup full bandwidth, as opposed to Intel's GTL+/SMP, which shares the same bus for multiple processors, giving diminishing returns on more than a couple of processors.

    Hopefully the chipsets for this should be out soon. Tyan was talking first quarter at the end of last year, but I think it will more likely be second quarter, since there's only 3 weeks left in the first quarter.

    A pet peeve: People refer to the new Athlon chipsets as "SMP". They're not. That's Symetrical MultiProcessing. They're Point to Point Multiprocessing, which I guess would be PTPMP or something.
  • by dgb2n ( 85206 ) <dgb2n.comcast@net> on Monday March 06, 2000 @04:44AM (#1223479)
    I understand the desire of AMD to beat Intel to to 1 GHz but when Intel does release more than vaporware, the benchmarks will blow the K75 core Athlon away. The 1 GHz Athlon is basically a crippled chip, with an L2 cache running at only 1/3 of the processor speed. Until the Thunderbird core is released and they have cache running at full processor speed, expect the Athlon to significantly lag the PIII at the same speed.

    I'm not convinced that AMD would ever have released this chip except to beat Intel to the punch. The significant improvements will only come as the entire architecture improves (full speed L2 cache, AMD 760 and AMD 770 chipsets with DDRAM support). Current Athlon users will have virtually no incentive to upgrade until then. I'm sticking with my K7 500 (running smooth and stable at 750 w/ 1/2 cache). I'd be willing to bet the improvement would hardly be measurable trading off 250 MHz of wait states for lowering the cache from 1/2 to 1/3.

    r/

    Dave
  • by Hortensia Patel ( 101296 ) on Monday March 06, 2000 @05:06AM (#1223480)

    OK, but I don't know that it's fair to blame AMD for this. Some of their early offerings (K5 generation, IIRC) did emphasize good architecture - pipelining, parallel execution etc - at the expense of raw MHz. And they suffered for it, because Joe Public has never heard of pipelining or parallel execution, but knows that 400 is a bigger number than 350.

    Result: AMD got the message and refocused their efforts on explicitly trying to pump clock speed as high as possible. I recall an AMD exec openly saying as much some time back. Doesn't seem to have done them any harm, you must admit.

    My point is: if the mass market is too dumb to care about anything but clock speed, you can't blame AMD for giving it to them. You might blame Intel, whose advertising seems to be deliberately aiming to sow confusion and ignorance about the technology they sell, but that's another story.

  • by KillBot ( 116344 ) on Monday March 06, 2000 @04:14AM (#1223481) Homepage
    There's an article [techweb.com] over at TechWeb that says intel plans to have a 1GHz p3 out by the 8th. And HP announced it will be the first to ship a 1GHz machine. I hope they didn't sacrifice quality for a good press blurb. But then again, I'd love to see them steal more market share from Intel.
  • by zaius ( 147422 ) <jeff&zaius,dyndns,org> on Monday March 06, 2000 @04:19AM (#1223482)
    In order to keep the SRAM cache within its operating specs, AMD was forced to chop the cache multiplier to 1/3, or 333Mhz. This is down 7Mhz from the 850's, but oh well.

    I wonder how much of an impact this will have on performance, compared to Intel's Kamati offerings with a 1/2 speed cache? Also, does anybody know what the multiplier is on Intel's newest P3's? I can't remember their stupid name.

    You can see on this page [sharkyextreme.com] some benchmarks, showing that the AMD 1Ghz just barely outperfom the Kyrotech 1Ghz chips. When you look at the cache, the kyrotech is an 800 upped to 1G, therefore the cache, which was at 320 is at 400 in the kyrotech, vs 333 in the AMD chip. What else was changed in the AMD chip to make it outperform an older version of itself with faster cache?

    thanx

  • by coreman ( 8656 ) on Monday March 06, 2000 @04:18AM (#1223483) Homepage
    Well, if it wasn't for the rush to the magic number (and the record books) we never would have seen this so soon. Intel is going to be hard pressed to get their yields up to where they can keep up with AMD's deliveries. I think we're seeing the changing of the guard at this point where AMD takes over the market leadership role and is going to drive Intel's product development rather than the opposite as has been the case in the past. Even with the superior performance in a "mhz even" case for AMD, they haven't stood still to have to prove equivalent benchmarks, they've attacked Intel in the forum Intel chose, clock rates. Congrats to AMD! Lead on!
  • by ndfa ( 71139 ) on Monday March 06, 2000 @04:19AM (#1223484)
    Well its good to see that x86 based processors have hit the 1GHz mark. Now i am no expert on CPU's when I did computer architecture we studied the MIPS architecture and it was pretty clean (the subset we studied) and small....

    I would like to know what the clocks on many of your boxes are like. I am pretty sure Sun does not sell their Eu10K's based on the MHz rating of their CPU;s. Also, how about some info on the SGI boxes and others ? ?

    Also I find it interesting how marketing has made the MHz mark so freaking important that people spend 100's of dollars to get an extra 50 Mhz and then go and get IDE drives!!!

    In parting I have to say that i have been a fan of AMD for sometime... cant wait till i start working so i can actually afford a K7! GO AMD!!!
    and kudos to the Engineers there to be able to keep the x86 arch. going... as i recall it was called the "Golden Handcuff..big money for backward compatability with a backward technology"

  • by TummyX ( 84871 ) on Monday March 06, 2000 @05:02AM (#1223485)
    The latest benchmark utility from Ziff Davis, Content Creation, is described as a system-level, application-based benchmark. Using Adobe Photoshop 5.0, Adobe Premiere 5.1, Macromedia Director 7.0, DreamWeaver 2.0, Netscape Navigator 4.6, and Sonic Foundry's Sound Forge 4.5, CC Winstone 2000 applies stress on a system's CPU to determine real-world content creation performance.

    Heh, what better way to see how stressed a processor can get than to throw Netscape 4.x at it?
  • by kwsNI ( 133721 ) on Monday March 06, 2000 @04:28AM (#1223486) Homepage
    My question would be is, are AMD and Intel going for quality now (sorry, is AMD going for quality now?) or is this processor race all about the frequency it runs at...

    The competition between Intel and AMD has been good on the one hand in that it has increased processor speed, encouraged new innovation and dropped the price of the processors down. But I'm starting to wonder how many corners AMD and Intel are cutting trying to one up each other. I think they've both gotton so absorbed with processor frequency that they forget the real benchmark of processors: How fast they run applications. There are other, non-x86 processors out there that would blow an Intel/AMD processor out of the water, even running at half the clock speed. So what if I have a bajillion-kagillion megahertz processor when my Palm Pilot runs faster.

    I think they need to start making the processors better, not faster. If they improve the quality of the CPU, the speed will come along naturally.

    kwsNI

  • by Frac ( 27516 ) on Monday March 06, 2000 @05:42AM (#1223487)
    The Athlon beats the Coppermine, clock for clock, even when the L2 cache is running at 1/3 speed.

    As you can see here [sharkyextreme.com], the Athlon 800 delivered a severe can of whoop-ass to the Pentium III 800 (both 133 and 100 bus speeds). And the following two points can be observed:

    1) The Athlon 800 has the same cache divider as the Athlon 1Ghz.
    2) The performance of the Athlon does not "severely lag" behind the Pentium, and in fact, it's a whole lot faster!

    expect the Athlon to significantly lag the PIII at the same speed

    Dude, either you work for Intel (FUD anyone?), or you better have some concrete information to back up your outrageous claims.

The reward for working hard is more hard work.

Working...