Intel: We Will "Blow Away" the PowerPC 970



  • Reply 21 of 86
    thegeldingthegelding Posts: 3,230member
    linux and ibm isn't making the 970 for is making the 970 for ibm and apple is using it ibm will want the 970 to not only succeed, but shine...g
  • Reply 22 of 86
    occamoccam Posts: 54member
    Perhaps the 970 compares to the Itanium2 because it runs on 64 bits? That's probably one reason why Intel avoids comparing the P4 to it. It does seem flattering to the 970 though, that Intel is taking aim at it already before it's ever been released in any incarnation.

    As for IBM's motives to "help" Apple? I do not think they're in it just to help Apple. Rather, I think Apple and IBM have found a common ground. Apple is willing to pay a handsome amount for chips to increase IBM's 970 ROI (return on investment), perhaps even making the 970 a healthy profit center. Plus, IBM has plans for the 970 chip in its own blades, etc., so IBM itself has a vested interest in keeping the 970 current. Finally, if the 970 is successful, it has a chance to further loosen the grip of the Wintel monopoly perhaps at the benefit of Linux (as open source Wintel alternative) and Apple (desktop Wintel alternative).

    The 970 could well be the wedge which opens the door into the Wintel monopoly, making the pie bigger for the rest of the industry. Apple and IBM are happy to work together toward that goal.
  • Reply 23 of 86
    kidredkidred Posts: 2,402member
    [quote]Originally posted by spooky:

    <strong>If I was a betting man (which I'm not), I would be willing to bet that the 970 would quickly pale into insignificance next to Intel's coming offerings. I don't understand why people think that IBM would be so keen to work on a 970 suitable for apple's needs. sure they'll probably power the next range of powermacs but what evidence is there that they would bust a gut to keep the 970 on top? why would IBM succeed where MOTO so willingly fail?</strong><hr></blockquote>

    Because IBM is bigger. Because IBM makes the power4. Because IBM currently makes chips for Apple. Beacuse IBM currently makes PPC for a few clients. Beacuse there will be more then 1 customer for the 970. Well, it seems the 970 will destroy the P4 because of the comparision to the iTanium. As noted many times above, the Itanium is in a different class yet Intel is comparing their large fry to a medium fry, 'yea, we have more fries'. Let's compare items from the same biggie menu, shall we.

    One thing tho, how can intel say the Itanium would beat the 970 when the 970 would be faster clock speed? A 1ghz Itanium would beat a 1.8ghz 970? Sounds like roll reversal to me.
  • Reply 24 of 86
    Another reason is more companies are trying to move away from Microsoft OS servers to Linux OS servers. Intel does not care what OS you use as long as it using Intel processors and chip sets.
  • Reply 25 of 86
    eugeneeugene Posts: 8,254member
    <a href=""; target="_blank"></a>;

    AMD just lost... I think they needed a headstart on Intel to push x86-64 into the mainstream. Intel's going to win by a TKO...

    [ 01-31-2003: Message edited by: Eugene ]</p>
  • Reply 26 of 86
    Another reason IBM wants 970 business is that they need the volume. They've got $M's in equiment, and they can't afford to just let it sit. They need the volume to keep it and their people busy.

    That's not meant to say that they will lose any money on the 970 program. I'm sure they will do well. I'm just saying that IBM isn't TOO big to care about the 970 program.

    Oh, and Eugene, your link don't work.
  • Reply 27 of 86
    rickagrickag Posts: 1,626member
    [quote]Could IBM Be the Next Computer Chip King?

    By Tiernan Ray

    NewsFactor Network

    January 31, 2003

    " Performance is a moving target, however, and Intel contends it will blow away any power advantage the 970 currently holds with its next Itanium, dubbed "Deerfield," to be previewed in February at its developer forum. Moreover, Intel spokesperson Bill Kircos says the business relationships behind Itanium are unstoppable."<hr></blockquote>

    This is just too easy.

    NO, Intel will just BLOW

    It is really comical that Intel is responding to a chip that has yet to ship with blowing hot air about another chip that isn't shipping yet.

    Let's see now, the 970 has how many transistors and the Deerfield has how many transistors again??

    <a href=""; target="_blank">Intel Deerfield</a>

    [quote]"First consumer IA-64 processor for mid-range servers and workstations. Includes around 1 MB of L3 cache and a slower bus than McKinley.""<hr></blockquote>

    <a href=""; target="_blank">OPPS forgot the apps!!</a>

    [quote] Jamel Tayeb, a technical marketeer at Intel's European software enabling group, said that while the architecture, the software and apps developed for the Itanium II will run on Deerfield, a chip which will be produced using 130 nanometer technology and with large caches, there will be some differences.

    Those are probably differences in the socket, because Tayeb also pledged that Intel would retain the Itanium II, McKinley socket design for the foreseeable future.

    He showed a chart that demonstrated that Deerfield would bridge the IA-32 and the IA-64 architetures.

    But it's clear that Intel will not go out of its way to improve IA-32 support in its Itanium family, despite competition from AMD with its Opteron dual 64/32 bit designs. Tayeb said that support for IA-32 would fundamentally be the same in future generations of the Itanium family.

    Jayeb demonstrated a "Tiger" chassis with Bandera inside, a four way CPU which seems to run floating point architecture like the blazes.

    He said that although there were 100+ applications for the Itanium already, other applications are en-route. Independent software vendors, he said, had longer roadmaps for their software products than the microprocessor company. But, he added, most of the big names had got with the programme.<hr></blockquote>

    am i misreading this or what. 100+ applications for Deerfield that targets the desktop market. - you go girl.

    [ 01-31-2003: Message edited by: rickag ]</p>
  • Reply 28 of 86
    powerpcpowerpc Posts: 109member
    [quote]Originally posted by Relic:

    <strong>He he, reminds me of something Bill Murray said, "We came we saw we kicked it's ass".</strong><hr></blockquote>'s a dog.......
  • Reply 29 of 86
    rickagrickag Posts: 1,626member
    Just a note. Intel is saying they will blow away the 970 with the "Deerfield", which is a supposed consumer chip. The comparison is not to the Itanium. We'll see.
  • Reply 30 of 86
    'Blow Away'?

    Kenobi: 'I don't think so...'

    Occam. Excellent post.

    Kid Red. I agree.

    When you're oponent is rubbishing you. It's a sign that you've got them rattled.

    Intel are rattled. They should be. They aint dealing with Moto or AMD who are bleeding red ink. They're dealing with IBM. Intel know they're going to have a tough time with the 970.

    I'm not so sure about the assertions that the 970 will be a 'medium performer'.

    IBM say that at the same clock, the 970 outguns the G4. 2:1 At 1.8 Gig = 3.6. Extra bandwidth. Better memory. Superior architecture. 2fpu. 5 execute instructions per clock?

    Sounds good to me. While the G4 uses 'Altivec' to 'hang in there' against the latest Pantium 4 the 970 looks like being ahead without altivec. With? It looks like being a real world benchmark 'blood bath'. Whereas the G4 needs altivec to look respectable (and needs 2 to gang up on one Pentium 4...) the 970 on its own looks ahead of the game.

    I'm a G4 in 'Towers' critic but if a 3.6 gig G4 existed I'd feel confident of it giving the Pentium 4 a real black eye! Similarly, I await the 970s performance 'with great interest.'

    If the 970 can hang with the latest Pantiums in Spec...I think current Mac owners can look forward to a 2-3 fold increase in performance in the next 8 months. Presumptiously yours...

    Lemon Bon Bon

    And note, it's a difference in design philosophy. IBM are going for wide. Well, wide and fast!!! And add in multiprocessing, an area which IBM has great experience in.

    X-serve anyone? With 8 chips 'in there'? Blow people's minds with performance ala Uber? I wouldn't want to take an 8-way 970 in a fight.

    This could give Apple Enterprise huge options!!!

    Intel? Fast and spinning tyre's going nowhere?

    Apple have taken their performance punches on 'mhz' over the last few years.

    Tables are turning. Hammer is hyped and delayed. Itanic...a slow moving ship... The 970 looks good on mhz and performance. How will Intel attack it?

    Heh. By debunking their own mhz myth?

    I've got to see the next year. I'm dying to see how it pans out. <img src="graemlins/cancer.gif" border="0" alt="[cancer]" />

    [ 01-31-2003: Message edited by: Lemon Bon Bon ]</p>
  • Reply 31 of 86
    amorphamorph Posts: 7,112member
    [quote]Originally posted by Lemon Bon Bon:


    When you're oponent is rubbishing you. It's a sign that you've got them rattled.</strong><hr></blockquote>

    Exactly. This is 100% pure FUD. It actually reads: Intel: We are afraid we'll get blown away by the 970, because we just realized how badly we screwed up with the Itanic.

    [quote]<strong>I'm not so sure about the assertions that the 970 will be a 'medium performer'.</strong><hr></blockquote>

    That's the punchline: It is a medium performer, by IBM's own description. Don't forget the POWER series.
  • Reply 32 of 86
    Did anyone else notice that the subject line for this thread is abbreviated to "Intel: We Will "Blow... " on the main forum page?

    [ 01-31-2003: Message edited by: ericj551 ]</p>
  • Reply 33 of 86
    stoostoo Posts: 1,490member
    However, Intel don't know about the 980.
  • Reply 34 of 86
    jbljbl Posts: 555member
    [quote]Originally posted by Stoo:

    <strong>However, Intel don't know about the 980. </strong><hr></blockquote>

    Would it be possible to create a version of the 970 with a x86 instruction unit (maybe replacing the altivec unit or something)?
  • Reply 35 of 86
    bartobarto Posts: 2,246member
    Yes, not by replacing the AltiVec unit though. There is a Decode, Crack & Group unit. This concievably could be extended into a x86 compatibility unit.

    But why would you want this?

  • Reply 36 of 86
    stoostoo Posts: 1,490member
    IIRC, a dual ISA x86/PowerPC chip was tried by IBM in the early days of the PowerPC but didn't work well enough.

    It would be quite cool if Apple/IBM aided Connectix im making Virtual PC faster than IA64 at running x86 32 bit code.
  • Reply 37 of 86
    telomartelomar Posts: 1,804member
    [quote]Originally posted by Utmost:

    <strong>I would agree with you spooky. By the time of the 970's release, it will only be a very modest competitor in comparison to Intel and AMD's offerings.</strong><hr></blockquote>

    AMD isn't in the same league anymore. It's the addiction of Intel haters now and not a lot more. Unfortunately those aren't the people who place the server orders and AMD now faces the problem that the people interested in their processors are mainly interested in cost. AMD doesn't want to be the budget server platform especially since they have bled cash being the budget processor maker through 2002 and they haven't shown the reliability to be anything more.

    The 970 is very nice for n-way servers and IBM is very well regarded by business. Compound that with the strong support IBM is throwing behind Linux and a lot of businesses will consider them.

    What you'll see is IBM chase the total cost of ownership sales line and Intel chase the initial payment argument.

    I have no doubt Intel will remain number 1 but I also have no doubt IBM is going to be very competitve with their upcoming products.
  • Reply 38 of 86
    The 970 will be a great server chip. Looks like it might be ready as soon as mid-summer?

    Isn't the X-serve due for an update mid-summer? Then just as DDR did, find it's way into the PowerMacs in the fall?
  • Reply 39 of 86
    The Deerfield is an IA-64 processor and it runs IA-32 (e.g. x86) code only via a form of hardware assisted emulation. Slowly. Intel is going to have a bit of a struggle to get this thing into the mainstream, and its going to confuse the market substantially. The server crowd can understand the architecture differences, but the consumer is going to be awfully confused.

    Apple, on the other hand, is going to have a chip that is dramatically faster on existing code and it can run new 64-bit apps that are specifically "optimized for the 970". From a marketing viewpoint Apple definitely has an edge this time around, hopefully they can exploit it.

    The Deerfield may definitely be fast and more power efficient than the Itanic, but I remain to be convinced that it'll be a match for the 970 in terms of price / performance / power. To coin a new phrase, the 970 is designed for "fast, wide and many" and its going to do that better than Intel's... and IBM will probably make it to 0.09 micron first.

    Things are heatin' up.
  • Reply 40 of 86
    The part I find scary is

    [quote]Chekib Akrout, vice president of PowerPC and networking technology development at Big Blue, told NewsFactor that the company's real target is servers running Xeon and, in the future, Itanium 2 chips.<hr></blockquote>

    [quote]According to Akrout, developers struggling with the total power budget for racks of multiprocessor Web and database servers will be attracted to the 970's power savings: It is expected to consume just tens of watts, compared with more than 100 watts for the next version of Itanium 2, code-named Madison.<hr></blockquote>

    What does this say to me? These chips

    <ol type="1">[*]Won't be cheap. They'll be priced like Xeon's (I *hope* not Itaniums) and workstattions based on them will be priced accordingly.[*]It will be another processor that Apple uses (if it does) in which Apple is not really part of the target market.[/list=a]

    In other words, if Apple can manage the memory controller design (no sure thing), they may have the next $6,000 XServe.

    I have dreams of a nice 970, but I strongly suggest people start to manage their expectations. Most of what IBM has stated about the chip indicate that this will not be an ideal chip for Apple to use. The Power4 is a nice chip too, but it doesn't mean that Apple can use it.

    The only bright spot is they did add Altivec, which means that they hope Apple will use it *somewhere*.
Sign In or Register to comment.