1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

so Jobs gets screwed by IBM over game consoles, thus Apple-Intel ?

Discussion in 'Intel' started by Guest, Jun 10, 2005.

  1. Guest

    Guest Guest

    It seems that Steve Jobs got screwed over by IBM, because IBM didn't care
    that much about making the CPUs that Jobs wanted for Macs, because IBM
    decided that game consoles were more important. 'IBM Inside' every
    next-generation game console: Sony Playstation3 (IBM 'Cell' CPU) Microsoft
    Xbox 360 (IBM 'Waternoose' CPU)
    Nintendo Revolution (IBM 'Broadway' CPU) which will ensure tens of millions
    of IBM CPUs sold every year, compared to maybe 5 million Mac CPUs every
    year. even if it was more than 5 million Mac CPUs, it would still almost
    certainly be less than 10 million. so it seems that is one of the
    major reasons why Apple has hooked up with Intel for CPUs



    related article
    http://arstechnica.com/columns/mac/mac-20050607.ars
     
    Guest, Jun 10, 2005
    #1
    1. Advertisements

  2. This is in tune with what most people have been saying. Your point is?
     
    Bjorn Olsson d.ä, Jun 10, 2005
    #2
    1. Advertisements

  3. Guest

    Judd Guest

    I doubt IBM gets as much per CPU in game consoles as they do for regular
    PCs. You can sell them per thousand for up to 1000 USD in PCs. It's
    probably $100 tops for game consoles and likely much less than that. Yeah,
    it's a great deal for them, but the likely revenue stream isn't as great as
    the previous poster made it seem. If Intel sold 10 million Mac CPU units,
    it may very well offset any loss of the X-Box when they practically gave
    away the P-III.
     
    Judd, Jun 12, 2005
    #3
  4. Guest

    keith Guest

    Oh, my! :-0
     
    keith, Jun 12, 2005
    #4
  5. Guest

    steve.travis Guest

    Since you obviously have $$ figures to back up your arguement, please
    feel free to share them...
    FWIW, neither IBM nor FreeScale really give a rat's ass about losing
    Apple's business. The level of R&D $$$ that both would have to put in
    to deliver what Apple was looking for, just wasn't worth it for them.
     
    steve.travis, Jun 12, 2005
    #5
  6. Guest

    imouttahere Guest

    It's well known that Intel gave a low bid for the xbox to kneecap AMD,
    and walked away from the xbox2 since IBM lowballed them in turn.

    IBM has done some baaaad things to the PPC core to get it fabable at
    3.2, and Sony is willing to accept at 12.5% defect rate (1 SPE out of 8
    per die being nonfunctional).
    What's more, TMK either company would only undertake this R&D if Apple
    paid for it. Freescale makes router chips and embedded stuff now,
    nothing suitable for portables. IBM doesn't feel confident on taking on
    Intel any more I guess, while Intel is already making exactly what
    Apple needs so Apple gets a free ride, or better, with them.
     
    imouttahere, Jun 12, 2005
    #6
  7. Guest

    websnarf Guest

    I assure you, that's not how it happened. AMD was able to match the
    price, but, Microsoft, gave Intel a much larger margin, because MSFT
    knew that Intel had the manufacturing capacity to guarantee delivery of
    CPUs for Xbox, while AMD (unless they dedicated their fac capacity
    almost totally to Xbox) did not. In fact, MSFT never had any real
    intention of going with AMD -- they just leaked the rumours that they
    were to scare Intel into giving them a lower price. Same thing with
    GigaPixel versus nVidia. Notice that *both* Intel and nVidia have
    walked away from Xbox 360? MSFT didn't exactly make a lot of friends
    on that transaction.
    Yeah, more pipelining, adding more dispatch slots, rename registers and
    an industry leading out-of-order window is really baaaaad ... NOT! The
    G5's biggest problem is not performance (it was only about 10-20%
    slower at the high end versus x86s at introduction, which is actually
    quite an accomplishment versus the previous generation Motorola crap)
    but rather its lack of clock scalability and ridiculous power
    consumption.

    See the clock scaling business is clearly something AMD learned with
    the K6, and Intel learned with the original Pentium. Just looking at
    the history of these two companies, and you can see that both know how
    scale a processors clock throughout its lifetime to match up with
    Moore's Law. The K6 went from 233Mhz to 550Mhz, the Pentium 33Mhz to
    200Mhz, Athlon from 600Mhz to 2.0Ghz, Pentium II/III from 300Mhz to
    1.266Ghz, Pentium IV from 1.5Ghz to 3.8Ghz, and the Athlon-64 got
    started at around 2.0Ghz. Look at the anemic PPC 970 processor;
    introduced at 2.0 Ghz, now a measley 2.5 Ghz but only by using *liquid
    cooling*.

    And lets talk about power consumption. Because of the threat from
    Transmeta (with their amazing low power cores) both AMD and Intel have
    reacted by making lower and lower power cores. Intel will eventually
    switch to them in their mainstream processors (via the Pentium-M path)
    and AMD has put their low-power technology (gates clocks, and other
    transistor techniques) into their entire product-line. In the
    meantime, IBM, not feeling any competitive pressure from anyone, just
    decided to crank the power consumption through the roof.
    Interesting. I am not aware of any other CPU manufacturer willing to
    share their yeild rates publically, so I don't really know what to
    compare that to.

    Possibly true. Not exactly good news for Apple -- if their suppliers
    didn't care about them, how could they expect high quality parts?
    That's right IBM, cannot hope to accomplish what even the lowly humble
    AMD has done. IBM simply doesn't have the competence, drive, or will
    to compete with either Intel or AMD. Confidence would not have helped
    them. And obviously Freescale is more than happy to stick to the
    confines of the embedded market.
     
    websnarf, Jun 12, 2005
    #7
  8. Guest

    imouttahere Guest

    Right. I don't think Intel found that the xbox was worth its while, tho
    seeing it stay x86 probably was.
    eh? TMK the Cell has none of those things, the PPC core has been cut
    down to the bare minimum, 604e-level of basicality.
    The G5 has nothing to do with Motorola crap, other than Apple getting
    Motorola to license VMX to IBM behind the scenes (this is lore, but
    AFAIK is true).
    ? The G5 seems identical to Opteron/Hammer to me. Hell, AMD and IBM
    closely collaborate now on process and fabbing.
    2.7 with liquid cooling. And the G5 came out at 1.6-2.0 and the Opteron
    at 1.4-1.8.

    A pair of 2.6Ghz Opteron 252s will set you back over $1700 at newegg,
    so at least for this generation Apple is better off on G5 than AMD.
    ? Part of the simplification of the Cell was to keep heat down. Plus
    IBM is planning to ship triple-core 3.2Ghz for MSFT later this year,
    Intel's dual-core 3.2Ghz offering costs more than an entire xbox2 will.
    This is also common with GPUs, the manufacturer sells the dodgy parts
    (that have point failures) at mid-range pricepoints with the bad
    modules disabled (8 pipes instead of 16 or what have you). Sony is
    doing the same thing to save some money.
    Indeed. Apple has had this problem for 5 or more years. More lore is
    that Moto was none to happy to lose its Mac OEM license, but not sure
    how true that is since I don't think they were really making any money
    on that.
    I think making IBM the bad guy out of this is a mistake, and I don't
    know you bad-mouth them so much since AMD is also relying on IBM's
    technology for their products (I don't think it's any accident that
    Opteron and G5 clockspeeds have been so closely matched these past 2
    years).

    My somewhat uninformed perspective is that IBM lacked the resources to
    match Intel's roadmap of P-M parts which are coming out next year.
    Remember too that Apple relies on IBM to design the memory controllers
    too, so really Apple is on the hook for 2x the R&D with the G5.

    That the Mac mini had to come out with 2 year-old technology instead of
    a nice & tight low-wattage Sonoma architecture was plain sad.

    I just think after 4 revs of OS X, it was time for the platform to
    break the PPC ball and chain that has dogged it for the past 5 years.
    Moving to x86 is not a big deal technically, and it's going to be
    interesting to see Apple partnered with a very strong hardware company
    for a change.
     
    imouttahere, Jun 12, 2005
    #8
  9. Guest

    websnarf Guest

    IIRC, VMX was not a Motorola design, but actually an IBM design.
    Remember that Mot and IBM had originally formed a group that worked
    together to specify the PPC; this meant that specs like VMX was shared.
    Motorola just happened to be the first to implement the specification.
    Opteron does not use liquid cooling. AMD basically bought fab capacity
    from IBM, since they can't fund the purchase of a second fab by
    themselves.
    The low end clock rates are just for product placement differentiation.
    I am just counting from the top of the clock rate range since that
    represents what the manufacturer can really do. Either way, my point
    is that the EOL clock rate ends up roughly 2x-3x of the original clock
    rate from the two major x86 vendors. Intel tends to have longer clock
    lifes, because they generally redo the same architecture twice in its
    lifetime (for eg, the P5, and P55c cores, the P-II, and the Deschutes
    core, Willamette and Prescott, etc.) Both IBM and Mot have typically
    had great troubles ramping the clock rate of their CPUs by comparison.
    Opterons are server CPUs, and thus are more comparable to Power
    processors (both have lower clock rates, but are better for SMP), not
    PowerPC processors. You should compare the Athlon-64 line to PowerPCs.
    Heat/MIPS possibly. The Cell has more active parts -- and assuming
    that each CELL processor has a floating point multiplier, there is no
    way in hell that the max power draw is *low* in a 3.0Ghz processor.
    Cost is just a question of marketing and positioning. (And only very
    rarely correllated with yield, like the Pentium 4 EE.) I know this is
    a common thing with Mac advocates -- i.e., trying to justify false
    comparisons but claiming that you need to normalize them by either
    price or clock rate.
    GPUs are different. There is a complete driver software layer that
    insulates bugs. I.e., people don't write "binary" to graphics cards --
    you write to Direct X or Open GL (or Quartz or GDI).
    Mot must have known they were going to lose Apple as a customer. They
    simply weren't making an effort to improve their processors in any
    reasonable way. Moore's law works on *all* processors where the vendor
    is making an effort. And serious CPU vendors *know* this. Mot allowed
    their processors to languish for *years* -- they must have known that
    they would eventually lose the Apple contract the next time IBM felt
    like making a new CPU.
    AMD is using IBM's *fabrication process* (which *is* state of the art),
    not their CPU design people.

    The clock rate similarity is just a question of ALU design limitations
    wrt the fabrication process. Its a common CPU design idiom -- work out
    the fastest you can make your generic ALUs in a given process, then
    design the rest of your pipeline with stages that run at the same
    speed. The reason Intel's clock rate is so different is because their
    ALUs are kind of these staggered half width things (which they call
    Netburst) which are a lot faster than regular oridinary ALUs.
    I think IBM lacked the interest in doing it. Like me, they assumed
    that Apple simply didn't have the cahonies to move to x86. And
    Mot/Freescale are a joke, so I think IBM just assumed that they had the
    contract no matter what they did.
    Well actually it *is* a big deal technically. Steve Jobs gave a great
    demo of a quick recompile, but Photoshop is clearly a very assembly
    language and endian sensitive application. And I suspect a lot of the
    media apps on the Mac are in a similar situation. The joke
    applications will just be a recompile, but not the serious stuff.
     
    websnarf, Jun 12, 2005
    #9
  10. Guest

    Travelinman Guest

    Yep. Joke applications like Mathematica, right?
     
    Travelinman, Jun 12, 2005
    #10
  11. Guest

    imouttahere Guest

    VMX's origin is rather unclear.
    IBM's PPC designs were G3, with no VMX. IBM made not a single G4 for
    Apple AFAIK.
    Apparently IBM was not a big fan of the additional diespace that the
    VMX stuff took up, prefering to go for lean & mean instead of making
    the CPU fatter.
    Doing some research I see this is more or less accurate.
    This is largely proportional to pipeline length. The P4's direction of
    longer pipeline has proved to be a mistake and Intel has already
    cancelled all designs and has moved to a P-M future.
    Athlons can't do MP so a dualie G5 would murder an athlon box for
    CPU-intensive stuff.
    yeah yeah
    that made no sense, I made no such claims.

    You claimed:
    while I pointed out that IBM is giving a 3.2Ghz triple-core part to
    Microsoft that enables Microsoft to build its entire xbox2 at a
    pricepoint less than the current 3.2Ghz dual-core Intel part.
    Another nonresponse. Whatever.
    True I guess.
    I think IBM wanted money to move forward. Intel will *give* Apple money
    to move (backward).
    Not really. Endianness was solved, more or less, 10+ years ago when
    NEXTSTEP became OPENSTEP.
    Hardware wise, Apple is already dealing with bass-ackward endianness
    with the PCI and AGP subsystems.
    It's a vector unit sensitive application, and Apple has been working on
    abstracting the vector units from the application with the
    Accelerate.framework, which combines vector routines and image
    processing routines in one architecture-neutral package.
    Nah. Real apps use OpenGL, and swizzling in OpenGL is pretty easy to
    set up.
     
    imouttahere, Jun 12, 2005
    #11
  12. Guest

    Nasht0n Guest

    Oh yeah, Mathematica was shown to run flawlessly after the recompile.
    It took, what, 30 seconds to demonstrate that it...launched on a Wipple
    machine. Kinda like those -probably- rigged demos of how PS for PPC was
    faster than PS for PCs.
    It certainly doesn't mean that it can't be done, but to suggest that
    it's easy, is only a notion a fanboi would entertain.

    Now you've done it! No more Kool-Aid for the others!

    Nicolas
     
    Nasht0n, Jun 12, 2005
    #12
  13. Guest

    websnarf Guest

    Mathematica is written mostly in the Mathematica language (so its like
    'porting' a Java program). The core stuff, does not require low level
    details like endianness or AltiVec or anything like that (I know, I've
    written a small symbolic math program myself -- you just don't obsess
    over low level details like this). And given that there already exists
    Mac and PC versions of Mathematica, I am actually surprised that it
    took them anything more than a recompile.

    Apple's headline application has always been Photoshop. You'll notice
    they made no mention of how that port is going.
     
    websnarf, Jun 12, 2005
    #13
  14. Guest

    websnarf Guest

    Not to me its not. If you read comp.arch at all, you'd know this. IBM
    did all instruction set design, since PPC is really derivative from the
    Power/RT processor designs, which predates PowerPC. Mot was just along
    for the ride.
    Yes, writing stuff down on paper is different from fabricating it. Mot
    just picked up the ball faster than IBM of fabrication.
    No, they just hadn't gotten around to it. Note that IBM actually added
    VMX to Power processors as well -- i.e., they always wanted to do it,
    they just hadn't put together a design until the Power4/PPC 970.
    Uhh ... this is a side effect of AMD buying fab capacity from IBM.
    They have to be familliar with the IBM process for this to happen,
    meaning they need to learn IBM's techniques for fabrication, which
    obviously they could incorporate into their own fabrication process.
    So they make a big deal about the sharing because it plays better than
    "AMD buys fab capacity from IBM".

    (The real point of the story is to convince investors that AMD will be
    able to compete with Intel's latest fabrication capacity and
    technology. I.e., its *possible* for AMD to win DELL's business.)
    It was not a mistake in the main requirement of being able to scale the
    clock rates (which they clearly succeeded at.) It was just a *failure*
    in the sense the AMD's design was superior.

    Actually both the K8 and PPC 970 are deeply pipelined cores as well.
    But the K8 is very clever in that certain stages which appear in
    Intel's pipelines, don't appear in the K8 at all. No drive stages and
    no rename/issue stages. (Nevertheless, the K8 definately has rename
    registers.) So the K8 may look less pipelined, just because it has
    fewer stages; but that's a bit artificial -- they just don't need the
    additional stages due to their ingenious design.

    Clock rate scaling is *not* just a question of pipelining. It also has
    to do with controlling clock skew. You can learn more here:


    http://stanford-online.stanford.edu/courses/ee380/050330-ee380-100.asx
    You brought in the issue of price.
    I don't know any details about the IBM 3.2 Ghz triple core part. If
    this part isn't somehow crippled in other ways, then why isn't Apple
    shipping with one? Doesn't it at all seem fishy to you?
    Sorry you don't understand these things. CPUs cannot ship with any
    "non-functional parts" unless its half of an L2 cache or something like
    that (but there are superior techniques used in modern CPUs). The
    reason is that all software gets compiled to machine language -- so
    everything just has to work. GPUs can ship with major
    non-functionalities, so long as they can be covered up in the drivers;
    I know this from first hand experience.
    Intel is not giving money to Apple, I assure you. They don't care
    *that* much about gaining Apple's business.
    This has nothing to do with anything. On x86 its very common to use
    the a trick for converting from double to integer by reading out the
    bits of the double (casting the pointer to the same memory location
    from (doube *) to (int *)) then adding a magic constant to the contents
    and reading out the results. Tricks like this require certain endian
    assumptions, that have nothing to do with the operating system. So
    long as memory is still accessible directly by software, there will
    always be endianness consistency issues.
    The devices on the other side of PCI and AGP are abstracted by drivers.
    General software does not talk to them directly -- its always through
    an OS level API which is itself shielded by drivers.
    And if Apple is still working on it, do you think Adobe, with its
    *shipping* requirements is paying any attention to it?
    Media includes audio dude. OpenGL doesn't do anything for that.
     
    websnarf, Jun 12, 2005
    #14
  15. Guest

    Travelinman Guest

    I have no way of checking whether that's true or not, but it sure
    negates your comment (above) where you claim that anything other than a
    'joke application' would be hard to port.
    Other than Adobe saying that they were committed to porting all their
    apps. IIRC, they claimed that they'd be one of the first.
     
    Travelinman, Jun 12, 2005
    #15
  16. Guest

    imouttahere Guest

    deja has 14 articles mentioning VMX and IBM.

    This one:

    http://groups-beta.google.com/group/comp.arch/msg/8c55bd0149aa57a7?dmode=source&hl=en

    is how I remember it.

    Though this one:

    http://groups-beta.google.com/group/comp.arch/msg/0cf5708fb80ac5da?dmode=source&hl=en

    supports your claim tht IBM was the prime mover initially.
    Well, they apparently took over the ride ca 1998. IBM was NOT using VMX
    in its chips, which was, TMK one of the major reasons they got to 1Ghz
    way before moto.
    I doubt this. VMX is/was only single-precision for one thing. Not a
    good match for Power.
    Well this is too technically wonky for me so instead of pipelining I'll
    just stick to sacrificing IPC for frequency...
    Just the point that IBM was making a triple-core 3.2Ghz part for
    Microsoft that will be much cheaper than Intel's dual-core 3.2Ghz part.

    Seems like IBM was willing to compete with Intel well in this
    particular arena.
    Microsoft owns that IP since they paid for it. TMK, they're taking it
    to TMSC or whoever for fabbing.

    IBM would be perfectly willing to do such an exercise for Apple, should
    they too also agree to pay some hundred(s) of megabucks to get the ball
    rolling.
    That makes more sense, but I don't think that is so common that drivers
    work around a variable number of missing features.

    In this case Sony knocking out 1 core is pretty much indentical to
    NVIDIA knocking out half their pipes, and also Intel shipping chips
    with half their cache physically deactivated.
    They give everyone money (that's why all Wintel OEM commercials end in
    the Intel tones). They're a regular ATM.
    Right. These silly things were obviated by NeXT taking the time to
    create an endian-neutral API to abstract this away. NSFloat, NSNumber,
    unichar, etc.
    true enough but much of Apple's existing driver codebase is already
    dealing with swizzling thanks to PCI.
    It shipped with 10.3, IIRC. Adobe doesn't use it since they have/had 5+
    years invested in altivec.
    10.4 features CoreImage additions, which Adobe won't use either most
    likely.
    True enough, I expect audio apps will be in for an especially tough
    time.
     
    imouttahere, Jun 12, 2005
    #16
  17. Guest

    imouttahere Guest

    Adobe should have a not-horrible time, just a lot of copy/paste from
    the existing Windows codebase for the intel binaries.

    Lot of drudgery, but the hard stuff has already been done.
     
    imouttahere, Jun 12, 2005
    #17
  18. Guest

    websnarf Guest

    Well that reasoning only applies if you have limited design teams that
    can only work on one thing at a time. MMX/SSE, etc never had any clock
    rate impact on any x86 processor. OTOH, I was left with the distinct
    impression that the AltiVec "byte permutation" instruction was
    difficult to implement in hardware, and may have caused clock rate
    pressure. If so, that would be a huge price to pay for 1 instruction.
    Uh ... excuse me, but VMX *IS* in the current Power CPUs.
    Right. You gotta stay in the reality distortion field; must not
    disturb it with annoying things like technical details.
    Ahahahaah! No, I mean *real* benchmarks by independent outsiders. You
    know, like Anand's, Tom's, Tech-fest, HARD-OCP, Sharky, FiringSquad,
    3D-Mark, SPEC-CPU, SPEC-GL. Oh I forgot, nobody ever benchmarks an
    Apple do they? ... Oh wait! Here's one:

    http://www.barefeats.com/macvpc.html
    You clearly don't know the Xbox history. Intel was not a player in the
    Xbox 360 by their own choice. Otherwise MSFT would gladly go with
    Intel again, to have a simple "backwards compatibility story", like
    SONY did with PS2.
    Ah! TSMC is a second tier fabrication facility. I.e., they don't
    support things like copper interconnect, SOI or shallow trench
    isolation. So this design has to be a completely stripped down,
    probably comparable to a G3 or something, but designed for clock rate
    and a generic fab process. If this thing has an IPC even as high as a
    P4 I would be greatly surprised.
    No, licensing the design is cheap, and Apple can clearly get the same
    deal that MSFT did at the drop of a hat. Apple rejected it, and with
    good reason. Apple needs the clock rate to scale, but MSFT doesn't.
    No, just the most common defects. You take care of all your low
    hanging fruit and all of a sudden your yield looks a heck of a lot
    better.
    Like I said, Intel (and AMD) doesn't do that anymore. You make the L2
    caches slightly redundant with built-in testing and with spare cache
    lines. The survive fab defects by remapping the defective lines during
    initial chip testing. The CELL having an additional processing unit,
    with the requirement of disabling routing to exactly one of them (with
    a special pin in the packaging, say) makes sense though.
    Its call comarketing. They don't need to pay for the whole ad, they
    just need to pay enough of it to convince the OEM to pay the balance
    themselves. Its actually a more effective way for Intel themselves to
    advertise, because people tend to buy complete systems more than they
    buy individual CPUs -- but Intel doesn't want to play favorites with
    system vendors. Intel is doing this as a means of competing with AMD,
    not Apple.

    This only seems out of place to you because Mot and IBM never truly
    competed with each other for Apple's business, and therefore they never
    had a reason to pull the same trick.
    I see -- and they included IEEE-754 representational tricks in those
    APIs? For an example of when IEEE-754 representation tricks are
    useful:

    http://www.pobox.com/~qed/sqroot.html

    Look. Any performance sensitive, standard binary format (like PKZIP,
    for example) is going to require low level endian swizzling on
    Mac-based software. Furthermore, a lot of software will just assume
    endianness when they can precisely because they associate the
    endianness with the operating system. The fact that Next happens to
    have made endian-neutral APIs doesn't mean anything to anyone who isn't
    writing code which is cross platform to begin with.
    And similar time invested in SSE. That's not the issue. They need to
    mate an OS X front end, with an x86 back end. That's just going to be
    a hell of a lot of work.
    *Can't* use, is more like it. To do the level of image manipulation it
    does, its all assembly language.
     
    websnarf, Jun 13, 2005
    #18
  19. Guest

    keith Guest

    On Sun, 12 Jun 2005 15:48:40 -0700, imouttahere wrote:

    The Power4 certainly does *not* have VMX, while the 970 does. ...just
    wanted to make this clear. ...carry on! ;-)
     
    keith, Jun 13, 2005
    #19
  20. Guest

    keith Guest

    1) Uh, "Power" is a marketing term, not a current architecture (the
    architecture is PowerPC, or more precisley for this class; PowerPC/AS).

    2) VMX is certainly *not* in the Power4 or Power5 processors. It was
    added to the 970, for the obvious reason.

    <snip>
     
    keith, Jun 13, 2005
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.