<strong>Obviously Apple is *meant* to use this chip. That doesn't necessarily that it will be *able* to effectively use the chip. If Apple can't design the appropriate memory controllers, it'll have to give the chip a pass, or at least a big delay.</strong><hr></blockquote>
And who says Apple can't make a memory controller for it, or that they haven't already got one?
[quote]<strong>Was that top 5 client or top 5 microprocessor client? If Apple is really a top 5 client, then you are right and its unlikely the Motorola is going to just stop selling G4's. On the other hand, there are continued rumors about Motorola just selling the whole division and getting out while it can. Not something you do if the division is making money...</strong><hr></blockquote>
And not something that has happened either. Moto's semi business represents a big part of their revenue picture even in the current economic climate. And Apple's commitment to the G4 is a large part of that despite Moto's embedded business.
[quote]<strong>Garbage. IBM is so diverse that there is *nothing* that is a huge part of their total sales. It's why they're so resilient. Think OS/2. They put a *huge* effort into it (probably about 100 times what they're putting into the 970). It only had hundreds of thousands of paying customers, so it was effectively killed.</strong><hr></blockquote>
No, it effectively died. Partly because IBM relied heavily on MS to assist in development when MS was more interested in developing a competing product.
[quote]<strong>For crying out loud, IBM has pretty much stopped making desktops because the return wasn't good enough! This is a company that can rationally look at a market and say "We're not making 10% annually on our investment, kill the project and move on."</strong><hr></blockquote>
Neither of your examples would support the idea that IBM would kill the PPC970 before it had a chance to prove itself.
[quote]<strong>Of course, IBM doesn't *intend to fail*. However, unlike Apple, if they run into development trouble or slow sales, they will ditch it.</strong><hr></blockquote>
You really don't have a clue do you? If they had had development problems they never would have announced the chip at MPF. It's a done deal, IBM will manufacture this chip and it will ship this year.
[quote]<strong>Apple doesn't have that choice, but they'd be fools to depend entirely on a product that is 0.1% of IBM's sales.</strong><hr></blockquote>
You mean 0% of IBM's sales since they haven't shipped yet. But when they do ship, what makes you think only Apple will be buying them? How do you know what IBM's plans are?
[quote]<strong>Or does noone remember document-centric computing anymore? Or Newton? Or...</strong><hr></blockquote>
Worse and worse. Cyberdog and Newton, huh? Oh yeah, the 16 Newton developers who were going to get rich off it were dissapointed, hardly anyone else was. And object oriented operating systems were a passing fad that no one could explain to users much less sell them.
[quote]<strong>I think it almost a certainty that the 970 will be produced unless IBM has a incredibly bad 2003, in which case they may just kill all sorts of projects as "too much additional investment required". </strong><hr></blockquote>
What additional investment? At this point they've invested too much NOT to go forward no matter what. And if IBM is such a huge behemoth, how is it they couldn't afford this teensy-weeny, itsy-bitsy project no matter how bad their bottom line?
[quote]<strong>However, the chance that it (and its follow-ons) are in production in 5 years is about 75% in my opinion.</strong><hr></blockquote>
Given the level of ignorance you demonstrate in regard to these matters, why should we concern ourselves with your opinion?
[quote]<strong>Would I throw my company completely around a chip with a 25% (my guess) chance of disappearance? Not in your life.</strong><hr></blockquote>
Again, I should care... why?
[quote]<strong>And I doubt IBM would be willing to guarantee 10 years of multimillion dollar investments...</strong><hr></blockquote>
You really have no clue, do you? Power4? PPC architecture? RISC? Have you heard any of these terms? Do you understand that the PPC970 is just one whistle stop on a roadmap that extends much farther than your limited ability to comprehend it?
[quote]<strong>Only because if the microprocessor division was making money, they wouldn't have essentially stopped putting R+D money into it.</strong><hr></blockquote>
Oh yeah, that's right. They stopped development of the Hip7 process and forget about all those announcements about 300mm fabs and HiP8 and all that, that's just a smokescreen to hide the fact that Moto isn't investing any R&D in semiconductors anymore. Uh huh.
[quote]<strong>I doubt that anyone is going to fight for a market that's already marginal. Even the Intel processor market is looking weak for anyone but Intel.</strong><hr></blockquote>
This is an observation that is so stunning in it's sheer paucity of any sense whatsoever that it defies comment.
"Even the Intel processor market is looking weak for anyone but Intel". That's just inspired. I'm going to have to remember that one.
[ 01-13-2003: Message edited by: Tomb of the Unknown ]</p>
[quote]IBM wouldn't have spent the R&D in developing a cpu they thought would be an insignificant part of their total sales. Me personally, I suspect IBM has BIG plans for this chip in the low-end server market, going head to head with competitors.<hr></blockquote>
Try Blade Servers--its a fairly low powered chip now (compared to Intel/AMD offerings) and a move to .09u and its an ideal blade server CPU. Perfect with Linux.
<strong>I think we all can agree that the PPC970 is coming some time this year. We just don't know how soon.
... depending on how motherboard design goes. The 970 is *not* just a plug in replacement for the G4 and Apple is (from reports) having a hell of a time with the board design. There is a considerable chance that that Apple never produces a 970 based machine. (Want a more technical description of why not? Check Ars Technica's Macintosh board.)
</strong><hr></blockquote>
If Apple really can't design support chips then they would need to buy them. IBM is undoubtably designing support chips and quite possibly one or two third parties as well(that already design PPC support chips).
Apple has had years to work on board design and is likely at a point where initial costs are not much of an issue--they must put out faster machines.
Also, I read Ars every day and your characterization of the threads there is very deceptive: practically every one there--including people within the industry--assume Apple is going to use the 970. They'd be insane not to.
And don't forget the possibility that Apple will use a memory controller of IBM origins, like when they used the first G3 and MPC106 memory controller combo from Motorola. Don't deny that IBM IS working on a memory controller, especially since we may be seeing IBM and Apple machines based on this processor coming out at relatively the same time. IBM is heavily involved in RapidIO. If I were to guess, their memory controller would has the 970 bus IO and a RIO port io connecting to the peripheral/PCI/AGP controller on the motherboard. Each processor would need it's own companion chip but they would connect to the same peripheral controller. The main controller on the motherboard could have 2 or even 4 (for quad processor machines) RIO ports.
<strong>Don't deny that IBM IS working on a memory controller...</strong><hr></blockquote>
I'd say IBM is working on it since the whole concept of 970. They can't make a machine with 970 without a memory controller, and they will definitely make such machines. Would be weird otherwise.
I'd say IBM is working on it since the whole concept of 970. They can't make a machine with 970 without a memory controller, and they will definitely make such machines. Would be weird otherwise.</strong><hr></blockquote>
Exactly. Maybe the 970 will be sold as part of a bundle with its companion chip.
First off, my postings have been in response to an absolute certaintly found on these boards that
(a) Apple will/should wholesale adopt the 970,
(b) The 970 will completely trash anything Motorola has out there, and
(c) The 970 will give the P4 a run for its money.
(1) I think Apple must diversify its CPU supplier if possible, and it should definitely not depend solely on IBM producing a product that in the end may or may not be viable in IBM's eyes.
(2) I'm not sure that Apple has the technical wherewithal to manage a board that supports the 970 effectively and at a comparable pricing structure. The machine has *got* to be relatively inexpensive.
(3) Steve's portable comments and Apple's drive to bring sales to portables are a pretty clear indication to me that Apple's CPU woes are not short lived.
(4) Given that Motorola is losing money on SPS hand over fist, we should be grateful that they haven't just closed the whole division down.
(5) Apple is not "doomed" if it doesn't close the performance gap. It would certainly help out, but I doubt anyone is buying Macintosh because its faster. Anyone who feels the need to have the fastest machine and owns a Macintosh is either really stupid, or more likely, really unhappy. Just face that fact that Intel can outspend the rest of the CPU sector 10 to 1 and face the reality that for anyone besides Intel, processor improvements are slow and steady. I cannot think of any company outside of AMD that is even close to competetive with Intel. MOST IMPORTANTLY OF ALL - IT DOESN'T REALLY MATTER. It's why a lot of Intel shareholders are really mad that Intel has spent huge amounts their money on R+D when it could go into shareholder pockets. Speed improvements are not really paying for themselves anymore...
Also let me say that
(1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel. It may not be substantially faster than Motorola solutions when it does come out and it may be subtantially more expensive to implement 970 based solutions.
Now to address points:
[quote]And who says Apple can't make a memory controller for it, or that they haven't already got one?<hr></blockquote>
Nobody. On the other hand, there's no guarantee
that they can do so effectively and at the right price point. Also, Apple's needs are a little different from IBM's. If the general feeling becomes that Apple's future depends on the 970 and the 970 doesn't pan out, Apple's in big trouble. Far better for the 970 to be one part of their strategy, and not a vital one.
[quote]Moto's semi business represents a big part of their revenue picture even in the current economic climate. And Apple's commitment to the G4 is a large part of that despite Moto's embedded business.<hr></blockquote>
From the 2001 annual reports:
Semiconductor Products Segment:
Sales: 4,936 million out of a total 30,004 million or about 16.5% of sales.
Earnings: -2,142 million out of a total of -5,511 million or about 38.8% of total losses.
By my calculations, Motorola is losing about 6 millions dollars every day it elects to keep the division open. In that situation, a CEO would *have* to be evaluating whether to keep it alive, or risk being shoved out the door by its stock holders. Certainly in such a situation, a CEO *could* be influenced by the fact that a major customer is constantly ragging on him and sending signals that they're going to abandon the company as soon as possible...
[quote]...My IBM dropping OS/2 and Desktops comment...<hr></blockquote>
[quote]Neither of your examples would support the idea that IBM would kill the PPC970 before it had a chance to prove itself.<hr></blockquote>
As I also said, barring catastrophe at IBM, the 970 will come out. Of course, catastrophe is not out of the question if a bean counter decides that IBM can make more on the necessary additional investment elsewhere. Bean counters are hyperaware of the dangers of throwing good money after bad. You do not look at the investment so far, only at the investment that is still required and the returns it will bring. Still, I also get the impression that the 970 was developed on the cheap, which is good. If IBM had to spend on the 970 more than 1/1000th of what Intel spent on the P4, there's no way that the 970 will succeed. The market is just too small.
[quote]If they had had development problems they never would have announced the chip at MPF. It's a done deal, IBM will manufacture this chip and it will ship this year.<hr></blockquote>
I'm certain the chip design is a done deal. However, that may only be 90% of the problem. There's still the other 90% - manufacturing. Ask Motorola...
[quote]You mean 0% of IBM's sales since they haven't shipped yet. But when they do ship, what makes you think only Apple will be buying them? How do you know what IBM's plans are?<hr></blockquote>
That's the critical question, isn't it. If Apple is the only major 970 customer, then I doubt the chip will last. The question is where will the other 970 customers come from?
[quote]Worse and worse. Cyberdog and Newton, huh? Oh yeah, the 16 Newton developers who were going to get rich off it were dissapointed, hardly anyone else was.<hr></blockquote>
I'd have to say that the comparison is apt. Apple's size in comparison to IBM is probably about the Newton developer's size compared to Apple :-). In both cases, their fate would not be a factor in the decision to continue/discontinue the product.
Manufacturing. Quite likely a vastly larger expense than the original design.
[quote]At this point they've invested too much NOT to go forward no matter what.<hr></blockquote>
First fallacy here. Good money after bad. Bean counters go after that in a flash. The *only* factor is will the additional investment (manufacturing costs, salaries) make *more* money than if the money was invested elsewhere.
[quote]And if IBM is such a huge behemoth, how is it they couldn't afford this teensy-weeny, itsy-bitsy project no matter how bad their bottom line?<hr></blockquote>
Second fallacy. Of course they *could* afford it. But why on earth do you think they'd spend their money subsidizing Apple?
[quote]Power4? PPC architecture? RISC? Have you heard any of these terms? Do you understand that the PPC970 is just one whistle stop on a roadmap that extends much farther than your limited ability to comprehend it?<hr></blockquote>
That's certainly the idea. And boy do I hope that it will pan out (for Apple's sake). However, if you think that IBM will not kill the project if it doesn't meet expectations (which I hope are very low indeed), you are deluding yourself.
[quote]Oh yeah, that's right. They stopped development of the Hip7 process and forget about all those announcements about 300mm fabs and HiP8 and all that, that's just a smokescreen to hide the fact that Moto isn't investing any R&D in semiconductors anymore. Uh huh.<hr></blockquote>
Sorry for being sloppy. The general consensus is that Motorola has severely cut the level of R+D at SPS. This may be why we have not seen Hip7 yet. With a loss of 2 billion+, I imagine that even if they dropped every penny of R+D, they'd be losing their shirts just on manufacturing costs...
[quote]This is an observation that is so stunning in it's sheer paucity of any sense whatsoever that it defies comment.
"Even the Intel processor market is looking weak for anyone but Intel". That's just inspired. I'm going to have to remember that one.<hr></blockquote>
Eek, grammar nazi :-).
Okay, how about
"Even the Intel-compatible processor market is looking weak for anyone but Intel"
or
"Even the x86-compatible processor market is looking weak for anyone but Intel"
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel
Well, since you've reversed yourself on everything of meaningfull importance...we are in agreement.
But you're still confused about Motorola's position vis-a-vis the G4. Essentially, the expensive part of designing the 7455 and its ilk has been done. Every 7455 that Motorola can sell Apple is PURE gravy. So Motorola is more than happy to sell 7455s to Apple FOR AS LONG AS APPLE BUYS THEM. There is absolutely no chance that Motorola will simply abandon making the 7455. The same is essentially true for the 7457 (basically a tweaked '55) for which the VAST MAJORITY of costs have been "sunk". That means they are effectively ignored when deciding whether to can the project or not.
Put another way... Motorola could announce they will no longer develop NEW chips...but they will sure as hell continue to make the chips they've already completed.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel[/i]
<hr></blockquote>
[quote]Well, since you've reversed yourself on everything of meaningful importance...we are in agreement.<hr></blockquote>
My original post was that there was a "considerable chance" that Apple does not release a 970. I'd put it at about 25% (a considerable chance) that Apple cannot make a 970-based machine that is cost effective enough or that Motorola will have something that is performance competitive (not necessarily superior) by the time the 970 would show up (in a year or so). That also makes a 75% chance that Apple does come out with a machine. I'd call that "very likely", perhaps just "likely"...
It was in response to the fact that apparently "everyone knows that an Apple 970 machine is a certainty" and that if that belief becomes wide spread and the 970 doesn't appear, Apple could be in trouble. Disappointed expectations are a killer. Certainly worse than a satisfactory, if somewhat mediocre slow increase in chip speed.
I'm not certain how this became "the 970 will never appear in an Apple machine".
Likewise, I'm not certain how my concern for the 970's long term existence turned into "IBM will knife it before it's first release".
I've reread my posts, and I don't see it.
All in all, it seems a somewhat overwhelming response to the *considerable possibility* that the 970 is not necessarily the answer to all of Apple's problems...
As for Motorola, I don't know how their manufacturing lines run, but you often have to close entire factories, losing a profitable product line along with the vast majority of unprofitable product lines. I'm certainly hoping that's not the case, but I have seen a lot of factories cut seemingly profitable lines because the fixed costs of being in the business at all were way too high.
I'll admit my fear is influenced by a number of companies that have closed factories and branches locally because they were "only" making millions of dollars a year, and companies felt their efforts and money were better placed elsewhere. Unemployment anyone?
(b) The 970 will completely trash anything Motorola has out there, and
(c) The 970 will give the P4 a run for its money.
(1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel. It may not be substantially faster than Motorola solutions when it does come out and it may be subtantially more expensive to implement 970 based solutions.<hr></blockquote>
Also let me say that
(a) your (2) response answers that question.
(b) What has been usually said is refering to the comments made that the 970 @ the same MHz will be about 2X's faster than a G4. Based on the # and length of the 970's pipelines, this seems reasonable. It should trash a G4 as they now stand, seems like there was some confusion on our part understanding your comments.
(c)Yes it should. And it will be 64 bit
(1) agreed.
(2) see above.
(3) agreed.
I think most of the arguments based on your comments were a combination of misunderstandings of some of your comments and the forcefullness of your statements.
That said I do tend to disagree with this statement.
[quote]Tom West
(3) Steve's portable comments and Apple's drive to bring sales to portables are a pretty clear indication to me that Apple's CPU woes are not short lived.<hr></blockquote>
Actually, I think his comments should take into account the setting - he just introduced 2 new laptops. The fact that laptops are becoming more popular is also important. But to infer Apple focused on laptops @ MWSF because "CPU woes are not short lived" may be a slight stretch. And what would you consider "not short lived" - months years decades
Actually, I get the feeling that most people here are saying that Apple should drop Motorola as fast as humanly possible. Motorola ought to be punished for not increasing the G4's speed fast enough.
I find that attitude dangerous and destructive. I also find the "IBM will be our saviour" highly worrisome. The 970 should be a nice chip. I anxiously look forward to how it benchmarks. But I really doubt that it will catch up with Intel, despite expectations. A few billion dollars R+D buys you a lot...
[quote]What has been usually said is refering to the comments made that the 970 @ the same MHz will be about 2X's faster than a G4. Based on the # and length of the 970's pipelines, this seems reasonable.<hr></blockquote>
I'll admit to being completely amazed if it can actually do that in real world benchmarks. There are so many other factors, even excluding disk, etc., that it seems unlikely. If the P4 is any indication, long pipelines lower performance. Although I will admit to hoping that somehow magic does occur, I don't expect it.
I like that. Apple, a world leader in Portable computers, won't be able to design a motherboard to cope with the 970. What a crock of crap.
Any two bit x86 motherboard maker are pumping out m/bs for whatever AMD/Intel cpu happens along.
Gee, somehow Apple won't be able to. :eek:
IBM themselves have said the 970 will (and these are 'conservative' estimates...) give twice the performance of any G4 at any given clock.
So, let's give Motorola the benefit of the doubt (stop sniggering at the back of class...) and say that they ship (heh) a 1.8 gig processor (G4, right?) by the end of the year...on their crappy motherboard.
IBM ship a 1.8 gig 970 on their rather NOT crappy m/board.
Yes. I can see how it will be hard for the 970 to be twice as fast. What, with extra fpu, more instructions per clock, bandwidth that swamps the current Moto' bus...
Given that the 970 does so well in classically Intel biased spec tests I'd take that to mean in real world tests? The 970 will be more than a match for any Intel this September.
1.8 970. Twice the performance of the G4. That makes a G4 at 3.6 gig if you will. And there's the better motherboard, instructions per clock a many and a superior fpu. In theory it stacks up well with IBM's 'conservative' estimates. In real world tests? We may be pleasantly surprised. Add in the fact that a few commentators have said the 970 will play better at multiprocessing than the G4. And the Pentium 4 is a single processor cpu?
Worried about Motorola or Intel? I don't think IBM, who designed the Power4 afterall and have world class cpu R&D are worried.
IBM have a market for the 970. Starting with their own. AND unlike 'last times', IBM need to play for keep. I don't thin they'll pull the plug before they see what it can really do. Then there's PPC. Which has been around a while. If the 970 doesn't tag 'them'. The follow in probably will.
'Considerable'. Yep. That is precisely the word I'd use for Apple to get a world class cpu into it's 'power'Mac line. Cause it doesn't look like it's coming from Motorola.
Motorola punished? Well, like Ati found out. One day, Apple will wake up and decide that having 2nd rate, uncompetitive product isn't good enough for their Mac users...or Apple's profits.
So Motorola will, for the short term, be relegated to the consumer line. And if they think they can come back at IBM with a G6, then good luck to them. They'll need it.
I have to agree that any talk of "punishing" Motorola is absurd. It's not Mot SPS's fault that Chris Galvin isn't half the CEO that his father was, or that he only acted to turn the company around when the shareholders revolted and threatened to take control of the company from the Galvin family. They didn't launch $9 billion in satellites into the air with no plan for what to do next, or abandon Six Sigma (Mot's vaunted - and successful - quality control program, which only tolerated a defect rate in parts per billion) in favor of general chaos.
Right now, SPS is in survival mode. They're shutting down all older fabs and partnering for new fabs. They're sticking with processor lines that are already paid for. The processors they have are excellent for what they are designed for, and Apple should use them to the extent that they are appropriate (e.g., in laptops and AIOs).
I am not expecting anything high end out of Motorola until they address more immediate concerns, like profitability, and until Motorola turns their whole big ship back around (they have brought back Six Sigma). Until then, I don't think anyone can blame them for cutting back to their core competencies and focusing on those.
It's important to remember that Mot is pretty much what Apple was in '97: A great company with a proud history going through a rough patch - except that Motorola goes a lot farther back. Apple would be foolish to burn any bridges here; in fact it would be wise to help Mot get through their current trials to the extent possible, not least because Mot SPS's processor design skills are highly respected, and useful to Apple.
Even if there's a 100% chance that Apple will roll out a 970-based PowerMac, they'd have to be crazy to go with a single chip vendor for very long. Especially a fundamentally self-interested partner such as IBM (I mean self-interested in the sense that IBM Semiconductor's #1 customer is, and has always been, IBM itself. This is true for the 970 as well).
Well I have been rather busy lately and have not gotten a chance to post. I am quite amazed at the last posts to my thread and hesitate to try to compete. All I will add is that I agree with both views being discussed in part. I reckon that there is a chance that the 970 will not make it into Macs and I also believe that IBM is the only other option apple has right now. I mean if apple does not use the 970 what chip are they going to use? And don't say AMD or Intel for it wont happen.
The only thing really holding Apple back right now is, as I have said before, the bad price verse performance ratio. Apple has beautiful computers and a great OS, much more stable than windows and much better for programing. As well they have great software. The apple Stores are making a profit and I believe they are a great thing for apple. If only they had a faster chip Like the 970 I believe they would take off. I think Steve knows this and I think we will see the 970 sooner rather than later. For if we do not apple will only fall further behind in market size.
<strong>Actually, I get the feeling that most people here are saying that Apple should drop Motorola as fast as humanly possible. Motorola ought to be punished for not increasing the G4's speed fast enough.
I find that attitude dangerous and destructive. </strong><hr></blockquote>
In part it comes from the past relationships between Apple & Motorola. From the killing of the clones (which lost Mot $100 million) to the G4 @ 500 MHz intro, to two major delays enroute from 500MHz to 1.25 GHz... with 1.25GHz parts that people claim are marked 'X' (Mot speak for 'Preproduction/Experimental').... Meaning we've been stuck at 1GHz in production quality G4s for how long?
If you are in an abusive relationship, it is time to leave. Period. I do agree that this won't help Mot/Apple relations, and I don't see that the Mot design teams themselves haven't operated in basically good faith - but it DOES happen that people let emotions overrule dollars in business decisions.
If I felt Mot would give me a fair shake on future work, or at least access to other products (say G3s), I'd try to stay on as good terms as I could... Note well that Fred Anderson didn't say "Our powermac sales are down 24% year-on-year because the performance we're able to get from Motorolla's abysmal G4 doesn't stack up..." He probably could say something that drastic...legally. But that's setting fire to the house you are currently in -> not the plan.
It isn't that Apple needs to 'punish' Mot - I'm pretty confident Apple would buy any really compelling product Mot ever comes up with. Just that there's no point in sticking around at the moment. A 1.4 GHz G4 isn't compelling. A 1.4 Ghz G5, which the real difference is the bus can now handle 200MHz is also not so compelling. If there was a serious power requirement drop or something.. maybe.
But 5x the FP, more if you need double precision is pretty freaking compelling to me.
Comments
<strong>Obviously Apple is *meant* to use this chip. That doesn't necessarily that it will be *able* to effectively use the chip. If Apple can't design the appropriate memory controllers, it'll have to give the chip a pass, or at least a big delay.</strong><hr></blockquote>
And who says Apple can't make a memory controller for it, or that they haven't already got one?
[quote]<strong>Was that top 5 client or top 5 microprocessor client? If Apple is really a top 5 client, then you are right and its unlikely the Motorola is going to just stop selling G4's. On the other hand, there are continued rumors about Motorola just selling the whole division and getting out while it can. Not something you do if the division is making money...</strong><hr></blockquote>
And not something that has happened either. Moto's semi business represents a big part of their revenue picture even in the current economic climate. And Apple's commitment to the G4 is a large part of that despite Moto's embedded business.
[quote]<strong>Garbage. IBM is so diverse that there is *nothing* that is a huge part of their total sales. It's why they're so resilient. Think OS/2. They put a *huge* effort into it (probably about 100 times what they're putting into the 970). It only had hundreds of thousands of paying customers, so it was effectively killed.</strong><hr></blockquote>
No, it effectively died. Partly because IBM relied heavily on MS to assist in development when MS was more interested in developing a competing product.
[quote]<strong>For crying out loud, IBM has pretty much stopped making desktops because the return wasn't good enough! This is a company that can rationally look at a market and say "We're not making 10% annually on our investment, kill the project and move on."</strong><hr></blockquote>
Neither of your examples would support the idea that IBM would kill the PPC970 before it had a chance to prove itself.
[quote]<strong>Of course, IBM doesn't *intend to fail*. However, unlike Apple, if they run into development trouble or slow sales, they will ditch it.</strong><hr></blockquote>
You really don't have a clue do you? If they had had development problems they never would have announced the chip at MPF. It's a done deal, IBM will manufacture this chip and it will ship this year.
[quote]<strong>Apple doesn't have that choice, but they'd be fools to depend entirely on a product that is 0.1% of IBM's sales.</strong><hr></blockquote>
You mean 0% of IBM's sales since they haven't shipped yet. But when they do ship, what makes you think only Apple will be buying them? How do you know what IBM's plans are?
[quote]<strong>Or does noone remember document-centric computing anymore? Or Newton? Or...</strong><hr></blockquote>
Worse and worse. Cyberdog and Newton, huh? Oh yeah, the 16 Newton developers who were going to get rich off it were dissapointed, hardly anyone else was. And object oriented operating systems were a passing fad that no one could explain to users much less sell them.
[quote]<strong>I think it almost a certainty that the 970 will be produced unless IBM has a incredibly bad 2003, in which case they may just kill all sorts of projects as "too much additional investment required". </strong><hr></blockquote>
What additional investment? At this point they've invested too much NOT to go forward no matter what. And if IBM is such a huge behemoth, how is it they couldn't afford this teensy-weeny, itsy-bitsy project no matter how bad their bottom line?
[quote]<strong>However, the chance that it (and its follow-ons) are in production in 5 years is about 75% in my opinion.</strong><hr></blockquote>
Given the level of ignorance you demonstrate in regard to these matters, why should we concern ourselves with your opinion?
[quote]<strong>Would I throw my company completely around a chip with a 25% (my guess) chance of disappearance? Not in your life.</strong><hr></blockquote>
Again, I should care... why?
[quote]<strong>And I doubt IBM would be willing to guarantee 10 years of multimillion dollar investments...</strong><hr></blockquote>
You really have no clue, do you? Power4? PPC architecture? RISC? Have you heard any of these terms? Do you understand that the PPC970 is just one whistle stop on a roadmap that extends much farther than your limited ability to comprehend it?
[quote]<strong>Only because if the microprocessor division was making money, they wouldn't have essentially stopped putting R+D money into it.</strong><hr></blockquote>
Oh yeah, that's right. They stopped development of the Hip7 process and forget about all those announcements about 300mm fabs and HiP8 and all that, that's just a smokescreen to hide the fact that Moto isn't investing any R&D in semiconductors anymore. Uh huh.
[quote]<strong>I doubt that anyone is going to fight for a market that's already marginal. Even the Intel processor market is looking weak for anyone but Intel.</strong><hr></blockquote>
This is an observation that is so stunning in it's sheer paucity of any sense whatsoever that it defies comment.
"Even the Intel processor market is looking weak for anyone but Intel". That's just inspired. I'm going to have to remember that one.
[ 01-13-2003: Message edited by: Tomb of the Unknown ]</p>
Try Blade Servers--its a fairly low powered chip now (compared to Intel/AMD offerings) and a move to .09u and its an ideal blade server CPU. Perfect with Linux.
<strong>I think we all can agree that the PPC970 is coming some time this year. We just don't know how soon.
... depending on how motherboard design goes. The 970 is *not* just a plug in replacement for the G4 and Apple is (from reports) having a hell of a time with the board design. There is a considerable chance that that Apple never produces a 970 based machine. (Want a more technical description of why not? Check Ars Technica's Macintosh board.)
</strong><hr></blockquote>
If Apple really can't design support chips then they would need to buy them. IBM is undoubtably designing support chips and quite possibly one or two third parties as well(that already design PPC support chips).
Apple has had years to work on board design and is likely at a point where initial costs are not much of an issue--they must put out faster machines.
Also, I read Ars every day and your characterization of the threads there is very deceptive: practically every one there--including people within the industry--assume Apple is going to use the 970. They'd be insane not to.
<strong>Moto's semi business</strong><hr></blockquote>
The quote of the day!
[quote]Originally posted by Outsider:
<strong>Don't deny that IBM IS working on a memory controller...</strong><hr></blockquote>
I'd say IBM is working on it since the whole concept of 970. They can't make a machine with 970 without a memory controller, and they will definitely make such machines. Would be weird otherwise.
<strong>
I'd say IBM is working on it since the whole concept of 970. They can't make a machine with 970 without a memory controller, and they will definitely make such machines. Would be weird otherwise.</strong><hr></blockquote>
Exactly.
(a) Apple will/should wholesale adopt the 970,
(b) The 970 will completely trash anything Motorola has out there, and
(c) The 970 will give the P4 a run for its money.
(1) I think Apple must diversify its CPU supplier if possible, and it should definitely not depend solely on IBM producing a product that in the end may or may not be viable in IBM's eyes.
(2) I'm not sure that Apple has the technical wherewithal to manage a board that supports the 970 effectively and at a comparable pricing structure. The machine has *got* to be relatively inexpensive.
(3) Steve's portable comments and Apple's drive to bring sales to portables are a pretty clear indication to me that Apple's CPU woes are not short lived.
(4) Given that Motorola is losing money on SPS hand over fist, we should be grateful that they haven't just closed the whole division down.
(5) Apple is not "doomed" if it doesn't close the performance gap. It would certainly help out, but I doubt anyone is buying Macintosh because its faster. Anyone who feels the need to have the fastest machine and owns a Macintosh is either really stupid, or more likely, really unhappy. Just face that fact that Intel can outspend the rest of the CPU sector 10 to 1 and face the reality that for anyone besides Intel, processor improvements are slow and steady. I cannot think of any company outside of AMD that is even close to competetive with Intel. MOST IMPORTANTLY OF ALL - IT DOESN'T REALLY MATTER. It's why a lot of Intel shareholders are really mad that Intel has spent huge amounts their money on R+D when it could go into shareholder pockets. Speed improvements are not really paying for themselves anymore...
Also let me say that
(1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel. It may not be substantially faster than Motorola solutions when it does come out and it may be subtantially more expensive to implement 970 based solutions.
Now to address points:
[quote]And who says Apple can't make a memory controller for it, or that they haven't already got one?<hr></blockquote>
Nobody. On the other hand, there's no guarantee
that they can do so effectively and at the right price point. Also, Apple's needs are a little different from IBM's. If the general feeling becomes that Apple's future depends on the 970 and the 970 doesn't pan out, Apple's in big trouble. Far better for the 970 to be one part of their strategy, and not a vital one.
[quote]Moto's semi business represents a big part of their revenue picture even in the current economic climate. And Apple's commitment to the G4 is a large part of that despite Moto's embedded business.<hr></blockquote>
From the 2001 annual reports:
Semiconductor Products Segment:
Sales: 4,936 million out of a total 30,004 million or about 16.5% of sales.
Earnings: -2,142 million out of a total of -5,511 million or about 38.8% of total losses.
By my calculations, Motorola is losing about 6 millions dollars every day it elects to keep the division open. In that situation, a CEO would *have* to be evaluating whether to keep it alive, or risk being shoved out the door by its stock holders. Certainly in such a situation, a CEO *could* be influenced by the fact that a major customer is constantly ragging on him and sending signals that they're going to abandon the company as soon as possible...
[quote]...My IBM dropping OS/2 and Desktops comment...<hr></blockquote>
[quote]Neither of your examples would support the idea that IBM would kill the PPC970 before it had a chance to prove itself.<hr></blockquote>
As I also said, barring catastrophe at IBM, the 970 will come out. Of course, catastrophe is not out of the question if a bean counter decides that IBM can make more on the necessary additional investment elsewhere. Bean counters are hyperaware of the dangers of throwing good money after bad. You do not look at the investment so far, only at the investment that is still required and the returns it will bring. Still, I also get the impression that the 970 was developed on the cheap, which is good. If IBM had to spend on the 970 more than 1/1000th of what Intel spent on the P4, there's no way that the 970 will succeed. The market is just too small.
[quote]If they had had development problems they never would have announced the chip at MPF. It's a done deal, IBM will manufacture this chip and it will ship this year.<hr></blockquote>
I'm certain the chip design is a done deal. However, that may only be 90% of the problem. There's still the other 90% - manufacturing. Ask Motorola...
[quote]You mean 0% of IBM's sales since they haven't shipped yet. But when they do ship, what makes you think only Apple will be buying them? How do you know what IBM's plans are?<hr></blockquote>
That's the critical question, isn't it. If Apple is the only major 970 customer, then I doubt the chip will last. The question is where will the other 970 customers come from?
[quote]Worse and worse. Cyberdog and Newton, huh? Oh yeah, the 16 Newton developers who were going to get rich off it were dissapointed, hardly anyone else was.<hr></blockquote>
I'd have to say that the comparison is apt. Apple's size in comparison to IBM is probably about the Newton developer's size compared to Apple :-). In both cases, their fate would not be a factor in the decision to continue/discontinue the product.
[quote]What additional investment?<hr></blockquote>
Manufacturing. Quite likely a vastly larger expense than the original design.
[quote]At this point they've invested too much NOT to go forward no matter what.<hr></blockquote>
First fallacy here. Good money after bad. Bean counters go after that in a flash. The *only* factor is will the additional investment (manufacturing costs, salaries) make *more* money than if the money was invested elsewhere.
[quote]And if IBM is such a huge behemoth, how is it they couldn't afford this teensy-weeny, itsy-bitsy project no matter how bad their bottom line?<hr></blockquote>
Second fallacy. Of course they *could* afford it. But why on earth do you think they'd spend their money subsidizing Apple?
[quote]Power4? PPC architecture? RISC? Have you heard any of these terms? Do you understand that the PPC970 is just one whistle stop on a roadmap that extends much farther than your limited ability to comprehend it?<hr></blockquote>
That's certainly the idea. And boy do I hope that it will pan out (for Apple's sake). However, if you think that IBM will not kill the project if it doesn't meet expectations (which I hope are very low indeed), you are deluding yourself.
[quote]Oh yeah, that's right. They stopped development of the Hip7 process and forget about all those announcements about 300mm fabs and HiP8 and all that, that's just a smokescreen to hide the fact that Moto isn't investing any R&D in semiconductors anymore. Uh huh.<hr></blockquote>
Sorry for being sloppy. The general consensus is that Motorola has severely cut the level of R+D at SPS. This may be why we have not seen Hip7 yet. With a loss of 2 billion+, I imagine that even if they dropped every penny of R+D, they'd be losing their shirts just on manufacturing costs...
[quote]This is an observation that is so stunning in it's sheer paucity of any sense whatsoever that it defies comment.
"Even the Intel processor market is looking weak for anyone but Intel". That's just inspired. I'm going to have to remember that one.<hr></blockquote>
Eek, grammar nazi :-).
Okay, how about
"Even the Intel-compatible processor market is looking weak for anyone but Intel"
or
"Even the x86-compatible processor market is looking weak for anyone but Intel"
Feel better?
[ 01-14-2003: Message edited by: Tom West ]</p>
((1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel
Well, since you've reversed yourself on everything of meaningfull importance...we are in agreement.
But you're still confused about Motorola's position vis-a-vis the G4. Essentially, the expensive part of designing the 7455 and its ilk has been done. Every 7455 that Motorola can sell Apple is PURE gravy. So Motorola is more than happy to sell 7455s to Apple FOR AS LONG AS APPLE BUYS THEM. There is absolutely no chance that Motorola will simply abandon making the 7455. The same is essentially true for the 7457 (basically a tweaked '55) for which the VAST MAJORITY of costs have been "sunk". That means they are effectively ignored when deciding whether to can the project or not.
Put another way... Motorola could announce they will no longer develop NEW chips...but they will sure as hell continue to make the chips they've already completed.
(1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel[/i]
<hr></blockquote>
[quote]Well, since you've reversed yourself on everything of meaningful importance...we are in agreement.<hr></blockquote>
My original post was that there was a "considerable chance" that Apple does not release a 970. I'd put it at about 25% (a considerable chance) that Apple cannot make a 970-based machine that is cost effective enough or that Motorola will have something that is performance competitive (not necessarily superior) by the time the 970 would show up (in a year or so). That also makes a 75% chance that Apple does come out with a machine. I'd call that "very likely", perhaps just "likely"...
It was in response to the fact that apparently "everyone knows that an Apple 970 machine is a certainty" and that if that belief becomes wide spread and the 970 doesn't appear, Apple could be in trouble. Disappointed expectations are a killer. Certainly worse than a satisfactory, if somewhat mediocre slow increase in chip speed.
I'm not certain how this became "the 970 will never appear in an Apple machine".
Likewise, I'm not certain how my concern for the 970's long term existence turned into "IBM will knife it before it's first release".
I've reread my posts, and I don't see it.
All in all, it seems a somewhat overwhelming response to the *considerable possibility* that the 970 is not necessarily the answer to all of Apple's problems...
As for Motorola, I don't know how their manufacturing lines run, but you often have to close entire factories, losing a profitable product line along with the vast majority of unprofitable product lines. I'm certainly hoping that's not the case, but I have seen a lot of factories cut seemingly profitable lines because the fixed costs of being in the business at all were way too high.
I'll admit my fear is influenced by a number of companies that have closed factories and branches locally because they were "only" making millions of dollars a year, and companies felt their efforts and money were better placed elsewhere. Unemployment anyone?
[ 01-14-2003: Message edited by: Tom West ]</p>
(a) Apple will/should wholesale adopt the 970,
(b) The 970 will completely trash anything Motorola has out there, and
(c) The 970 will give the P4 a run for its money.
(1) The 970 is *extremely* likely to come out.
(2) Apple is very likely to use the 970 in some products.
(3) In all likelihood, the 970 will shrink the performance gap between Mac and Intel. It may not be substantially faster than Motorola solutions when it does come out and it may be subtantially more expensive to implement 970 based solutions.<hr></blockquote>
Also let me say that
(a) your (2) response answers that question.
(b) What has been usually said is refering to the comments made that the 970 @ the same MHz will be about 2X's faster than a G4. Based on the # and length of the 970's pipelines, this seems reasonable. It should trash a G4 as they now stand, seems like there was some confusion on our part understanding your comments.
(c)Yes it should. And it will be 64 bit
(1) agreed.
(2) see above.
(3) agreed.
I think most of the arguments based on your comments were a combination of misunderstandings of some of your comments and the forcefullness of your statements.
That said I do tend to disagree with this statement.
[quote]Tom West
(3) Steve's portable comments and Apple's drive to bring sales to portables are a pretty clear indication to me that Apple's CPU woes are not short lived.<hr></blockquote>
Actually, I think his comments should take into account the setting - he just introduced 2 new laptops. The fact that laptops are becoming more popular is also important. But to infer Apple focused on laptops @ MWSF because "CPU woes are not short lived" may be a slight stretch. And what would you consider "not short lived" - months years decades
I find that attitude dangerous and destructive. I also find the "IBM will be our saviour" highly worrisome. The 970 should be a nice chip. I anxiously look forward to how it benchmarks. But I really doubt that it will catch up with Intel, despite expectations. A few billion dollars R+D buys you a lot...
[quote]What has been usually said is refering to the comments made that the 970 @ the same MHz will be about 2X's faster than a G4. Based on the # and length of the 970's pipelines, this seems reasonable.<hr></blockquote>
I'll admit to being completely amazed if it can actually do that in real world benchmarks. There are so many other factors, even excluding disk, etc., that it seems unlikely. If the P4 is any indication, long pipelines lower performance. Although I will admit to hoping that somehow magic does occur, I don't expect it.
Any two bit x86 motherboard maker are pumping out m/bs for whatever AMD/Intel cpu happens along.
Gee, somehow Apple won't be able to. :eek:
IBM themselves have said the 970 will (and these are 'conservative' estimates...) give twice the performance of any G4 at any given clock.
So, let's give Motorola the benefit of the doubt (stop sniggering at the back of class...) and say that they ship (heh) a 1.8 gig processor (G4, right?) by the end of the year...on their crappy motherboard.
IBM ship a 1.8 gig 970 on their rather NOT crappy m/board.
Yes. I can see how it will be hard for the 970 to be twice as fast. What, with extra fpu, more instructions per clock, bandwidth that swamps the current Moto' bus...
Given that the 970 does so well in classically Intel biased spec tests I'd take that to mean in real world tests? The 970 will be more than a match for any Intel this September.
1.8 970. Twice the performance of the G4. That makes a G4 at 3.6 gig if you will. And there's the better motherboard, instructions per clock a many and a superior fpu. In theory it stacks up well with IBM's 'conservative' estimates. In real world tests? We may be pleasantly surprised. Add in the fact that a few commentators have said the 970 will play better at multiprocessing than the G4. And the Pentium 4 is a single processor cpu?
Worried about Motorola or Intel? I don't think IBM, who designed the Power4 afterall and have world class cpu R&D are worried.
IBM have a market for the 970. Starting with their own. AND unlike 'last times', IBM need to play for keep. I don't thin they'll pull the plug before they see what it can really do. Then there's PPC. Which has been around a while. If the 970 doesn't tag 'them'. The follow in probably will.
'Considerable'. Yep. That is precisely the word I'd use for Apple to get a world class cpu into it's 'power'Mac line. Cause it doesn't look like it's coming from Motorola.
Motorola punished? Well, like Ati found out. One day, Apple will wake up and decide that having 2nd rate, uncompetitive product isn't good enough for their Mac users...or Apple's profits.
So Motorola will, for the short term, be relegated to the consumer line. And if they think they can come back at IBM with a G6, then good luck to them. They'll need it.
Lemon Bon Bon
Right now, SPS is in survival mode. They're shutting down all older fabs and partnering for new fabs. They're sticking with processor lines that are already paid for. The processors they have are excellent for what they are designed for, and Apple should use them to the extent that they are appropriate (e.g., in laptops and AIOs).
I am not expecting anything high end out of Motorola until they address more immediate concerns, like profitability, and until Motorola turns their whole big ship back around (they have brought back Six Sigma). Until then, I don't think anyone can blame them for cutting back to their core competencies and focusing on those.
It's important to remember that Mot is pretty much what Apple was in '97: A great company with a proud history going through a rough patch - except that Motorola goes a lot farther back. Apple would be foolish to burn any bridges here; in fact it would be wise to help Mot get through their current trials to the extent possible, not least because Mot SPS's processor design skills are highly respected, and useful to Apple.
Even if there's a 100% chance that Apple will roll out a 970-based PowerMac, they'd have to be crazy to go with a single chip vendor for very long. Especially a fundamentally self-interested partner such as IBM (I mean self-interested in the sense that IBM Semiconductor's #1 customer is, and has always been, IBM itself. This is true for the 970 as well).
The only thing really holding Apple back right now is, as I have said before, the bad price verse performance ratio. Apple has beautiful computers and a great OS, much more stable than windows and much better for programing. As well they have great software. The apple Stores are making a profit and I believe they are a great thing for apple. If only they had a faster chip Like the 970 I believe they would take off. I think Steve knows this and I think we will see the 970 sooner rather than later. For if we do not apple will only fall further behind in market size.
[ 01-14-2003: Message edited by: Algol ]</p>
I could help but notice the irony of that comment. Especially if you replace 'IBM' with Motorola.
Heh.
<img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
Lemon Bon bon
<strong>Actually, I get the feeling that most people here are saying that Apple should drop Motorola as fast as humanly possible. Motorola ought to be punished for not increasing the G4's speed fast enough.
I find that attitude dangerous and destructive. </strong><hr></blockquote>
In part it comes from the past relationships between Apple & Motorola. From the killing of the clones (which lost Mot $100 million) to the G4 @ 500 MHz intro, to two major delays enroute from 500MHz to 1.25 GHz... with 1.25GHz parts that people claim are marked 'X' (Mot speak for 'Preproduction/Experimental').... Meaning we've been stuck at 1GHz in production quality G4s for how long?
If you are in an abusive relationship, it is time to leave. Period. I do agree that this won't help Mot/Apple relations, and I don't see that the Mot design teams themselves haven't operated in basically good faith - but it DOES happen that people let emotions overrule dollars in business decisions.
If I felt Mot would give me a fair shake on future work, or at least access to other products (say G3s), I'd try to stay on as good terms as I could... Note well that Fred Anderson didn't say "Our powermac sales are down 24% year-on-year because the performance we're able to get from Motorolla's abysmal G4 doesn't stack up..." He probably could say something that drastic...legally. But that's setting fire to the house you are currently in -> not the plan.
It isn't that Apple needs to 'punish' Mot - I'm pretty confident Apple would buy any really compelling product Mot ever comes up with. Just that there's no point in sticking around at the moment. A 1.4 GHz G4 isn't compelling. A 1.4 Ghz G5, which the real difference is the bus can now handle 200MHz is also not so compelling. If there was a serious power requirement drop or something.. maybe.
But 5x the FP, more if you need double precision is pretty freaking compelling to me.
"they'd have to be crazy to go with a single chip vendor for very long. Especially a fundamentally self-interested partner such as IBM"
<strong>I could help but notice the irony of that comment.</strong><hr></blockquote>
That's not irony, that's evidence.
[ 01-15-2003: Message edited by: Amorph ]</p>