"Motorola to boost 0.13µ PowerPCs with low-k dielectrics
Speculation that Motorola may soon cease to be a supplier of processors to Apple may be premature. The chip maker yesterday said it had successfully implemented low-k dielectric materials in its 0.18 micron silicon-on-insulator (SOI) processors, bringing an estimated 20 per cent speed bump to the PowerPC line.
Motorola expects to roll out the process on its 0.13 micron chips this month.
The company didn't mention specific models, but on the basis of the kind of chips it said it's planning to implement low-k dielectric materials in, candidate processors include the MPC7457, which has yet to ship but is set to take Motorola's G4 family beyond 1GHz.
"Our goal is to stay with a frequency doubling every 18 months or so, and get into the 2GHz range for PowerPC, but at very low power consumption of, say, 20W," said Dirk Wristers, director of device/integration for Motorola's MOS-13 wafer fab, according to an EE Times report. "The frequency could be higher if we were at higher power."
Low-k dielectric materials are used to improve the insulation between processor circuits. As chips get smaller and their circuit lines come closer together signals in nearby circuits increasingly interfere with each other, effectively slowing the ability of the component transistors to switch on and off. Getting the switch speed up involves raising the voltage and thus the power the transistor - and in turn the whole chip - consumes.
Reducing that cross-circuit interference with low-k dielectrics allows transistors to switch more quickly and draw less power. The trouble is, implementing low-k dielectrics has not proved easy - new materials require new methods, all of which must be thoroughly proved before they can be implemented in commercial products.
Wristers claimed that some chip vendors - he didn't name names - are now waiting for the switch from 0.13 micron fabrication processes to 0.09 micron (90nm) before implementing low-k dielectrics. The implication in his comment is that since Motorola can use the technology in its 0.13 micron chips, it will be able to really run with it when it makes the transition to 90nm.
With Apple expected to announce a shift to IBM's 64-bit PowerPC 970 processor in just a few weeks' time, the timing of Motorola's announcement takes on a new relevance. Speculation is already mounting that Apple will drop Motorola for IBM, ditching the G4 for a mix of 970s and Big Blue's rumoured AltiVec-enabled G3-class processors.
Motorola's top-end product is the 7457, announced earlier this year but not yet in volume production. The 7457 is essentially a 7455 with double the on-die L2 cache (512KB) and fabbed at 0.13 micron rather than 0.18 micron. Older Motorola roadmaps show the 7457 topping out at 1.3GHz. Implementing low-k dielectrics could lift that ceiling significantly.
Interestingly, Motorola said it had been delivering low-k dielectric 0.18 micron SOI processors for a full quarter. The 7455 is just such a chip - Motorola's claim may explain why Apple has had such success overclocking the 1GHz 0.18 micron MPC7455 to 1.42GHz in its Power Mac models..."
I have read some benchmarks comparing the dual G4 1,42 with the old dual 1,25 ghz. The dual 1,42 is only 3 % better. ... A 2 ghz G4 will not ouperform much the current dual, if the same lame bus is still there.
There is no future for the current G4, and the RM version disappeared from the roadmap.
I don't know what the bus speed means for embedded applications, but in the desktop area this brings the question: is Motorola beginning to resort to RDF in order to sell CPUs which have to wait half their time for the bus? Or is there a sound reason to bump the clock rate any further? Maybe, for IU and FPU it still has some sense, but doesn't it look like wasting time and resources?
That's why I'm pulling for Mot to overcome their problems and get back in the game. If Apple goes from single source to single source, their situation hasn't really improved - even if, in the short term, the top-end CPU available to them has.
...
The problem with IBM, of course, is that IBM is huge enough to do things entirely in its own interest, following its own compass, and so they ended up making smaller and smaller G3s rather than jump on the SIMD wagon because it met some internal goal. The best situation for Apple is multiple vendors - that's why they went with AIM in the first place, rather than AM or AI.
Good post.
It certainly is. All the frenzied IBM-olites take note, read, re-read, and read again untill these words have penetrated well past your current CPU frustrations. Apple needs Moto just as badly as they need IBM. A Moto-SMC partnership is reason for optimism regarding Moto. IBM's history is reason for skepticism regarding IBM. If the rumors of Apple v. Moto legal troubles are true, it can only mean that Apple understands their basic need for two CPU suppliers and is leveraging to get moto's g4/5 developments back on track, they know a long law suit would cost them more in the end, than a nice new agreement.
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Not exactly hidden, rather delayed. As long as Apple's and IBM's roads coincide, Apple will be safe and provided with good CPUs galore. But only as long.
It certainly is. All the frenzied IBM-olites take note, read, re-read, and read again untill these words have penetrated well past your current CPU frustrations. Apple needs Moto just as badly as they need IBM. A Moto-SMC partnership is reason for optimism regarding Moto. IBM's history is reason for skepticism regarding IBM. If the rumors of Apple v. Moto legal troubles are true, it can only mean that Apple understands their basic need for two CPU suppliers and is leveraging to get moto's g4/5 developments back on track, they know a long law suit would cost them more in the end, than a nice new agreement.
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Good post Matsu.
The problem here is that Motorola is heading down. Right now it seems that it may be able to produce some fairly useful chips (2ghz, 20 watts) that would be nice to have in the iBooks and other computers still stuck in G4 land. And I don't believe that all Apple products will jump ship in a few months. The G4 is going to be with us in some products for some time yet, and it's important that Motorola make them as good as possible.
Some sort of AIM-collaboration would be nice to see again, but I really don't see that happen. I'm sure Moto has had some very good processors in developement that have been axed. Even two years ago we heard talk of 2.4 ghz G5's. If they had collaborated with IBM then they may actually had seen the light of day. But Motorola seems to be having increadibly big problems actually producing the chips they are designing. I don't know what the problem is, but I guess that if IBM could have helped Motorola make the chips, there may have been some fairly fast PPC processors out there.
Another question that must be asked is this: Is moto actually interested in making chips for computers right now, or are they just making them because Steve's all over them??? If they don't want to, Apple should better leave them alone and go IBM as fast as possible.
Not exactly hidden, rather delayed. As long as Apple's and IBM's roads coincide, Apple will be safe and provided with good CPUs galore. But only as long.
You are absolutely right.
The good new is speaking of the computer industry, the road seems to be long.
IBM's history is reason for skepticism regarding IBM.
Which "history" are you refering to? The only thing I can think of is that IBM pulled out of AIM when Motorola and Apple wanted to go with AltiVec but IBM wanted to go with high clock rates and focus on scalar performance. Guess who turned out to succeed? And now they have the ability to add vector units to their fast design, which is going to save Apple's bacon. I think this pattern of one hardly qualifies as a track record and you have to compare it against a solid commitment to POWER since about '88 -- 14+ years. Is there something else you are comparing it to?
Which "history" are you refering to? The only thing I can think of is that IBM pulled out of AIM when Motorola and Apple wanted to go with AltiVec but IBM wanted to go with high clock rates and focus on scalar performance. Guess who turned out to succeed? And now they have the ability to add vector units to their fast design, which is going to save Apple's bacon. I think this pattern of one hardly qualifies as a track record and you have to compare it against a solid commitment to POWER since about '88 -- 14+ years. Is there something else you are comparing it to?
I wholeheartedly agree. People look at IBM leaving AIM as though IBM gave up on PPC or something like that. This isn't even close to the case. IBM left AIM because they had a differing opinion about the direction the PPC architecture should go. IF Apple had gone with IBM's vision for PPC, then we would probably have 3-4GHz PPC processors without Altivec. IF Apple had gone with IBM, then PPC would probably have been the first to 1 GHz. I like Altivec and think that it is some amazing hardware, but I think that I would trade it for a PPC running at 4GHz.
Folks, IBM is not Moto, and some of the arguments that cry out for more than one supplier of PPC chips are not the best ones. For example, IBM is not going to decide to abandon the high end CPU market for the embedded market. IBM does not view the PPC architecture as an architecture for embedded CPU's. IBM has been and will continue to position the PPC architecture against x86, x86-64, and IA-64. IBM knows how to make fast chips and has clean fabs (witness the speed of the Power 4 chips given the die size and transistor count!). If Apple goes with IBM as its single chip supplier, then at least they have the common goal of making fast desktop chips!
In my ideal world, Moto would update the bus for the G4, shrink the die size and become the portable chip producer for Apple until the 970's undergo a die shrink and are fully usable in portables. I don't hold out much hope for Moto to remain with Apple because they have abandoned the high end market and that is where Apple wants to play.
Why would you think that IBM would have gotten to 4 GHz by now if Apple went with them to begin with? From reading the IBM documents on PPC it seems that have been developing the new PPC chips continuously since they left AIM. It may be that Apple has realized what IBM was doing with the PPC chips over time and just waited until they were available.
Even if Mot was at 3 GHz right now, I think IBM would still be developing the PPC970, but it may not have had Altivec, which from what I have read and seen seems to be an add-on anyway. Stevie said awhile back that they were going to have options (and not x86 as a everyone seem to hope for).
I wholeheartedly agree. People look at IBM leaving AIM as though IBM gave up on PPC or something like that. This isn't even close to the case. IBM left AIM because they had a differing opinion about the direction the PPC architecture should go. IF Apple had gone with IBM's vision for PPC, then we would probably have 3-4GHz PPC processors without Altivec. IF Apple had gone with IBM, then PPC would probably have been the first to 1 GHz. I like Altivec and think that it is some amazing hardware, but I think that I would trade it for a PPC running at 4GHz.
I disagree with this -- I don't think we would have seen anything terribly different than the POWER4/970. This was IBM's direction and when they bailed on AIM all they did was take their toybox and go home.
I disagree with this -- I don't think we would have seen anything terribly different than the POWER4/970. This was IBM's direction and when they bailed on AIM all they did was take their toybox and go home.
Once Apple chose Moto and IBM left AIM, then IBM only had to make chips for IBM. This meant making high end server chips, hence the Power 4 with an architecture that was clearly meant for servers. Had Apple stayed with IBM, then IBM would have focused on desktop chips. I seem to recall roadmaps back in the day that indicated 2 GHz in 2002. Of course, all such roadmaps are subject to some fudging, but at least IBM wasn't going to fall off the roadmap because of dirty fabs.
Once Apple chose Moto and IBM left AIM, then IBM only had to make chips for IBM. This meant making high end server chips, hence the Power 4 with an architecture that was clearly meant for servers. Had Apple stayed with IBM, then IBM would have focused on desktop chips. I seem to recall roadmaps back in the day that indicated 2 GHz in 2002. Of course, all such roadmaps are subject to some fudging, but at least IBM wasn't going to fall off the roadmap because of dirty fabs.
Yes, but I think all being part of AIM would have really done is either accelerate the 970's timeline or give Apple/Moto the fab technology needed to deliver on the G5.
Yes, but I think all being part of AIM would have really done is either accelerate the 970's timeline or give Apple/Moto the fab technology needed to deliver on the G5.
And this is a bad thing?
Not to mention, had IBM stayed in AIM, Moto might have gotten a break on IBM's exorbitant contracting rates and been able to borrow the expertise of their researchers and (especially!) their nice, clean fabs. We might have had a PowerBook line powered by 7457s this spring, or even a G5.
i just want to see the gobi iBooks before the dang school year starts...or if they want to throw a 1.25 G4 in there and stay with moto for a while, i would be ok with that too....but get all apple computers over 1 g and get all to have altivec
g
this, of course, would rely on the PB getting 970 chips (which might not happen till early 2004) or the PB getting 1.6 G4 chips or above (which might not happen ever...plus they would likely cook your naughty bits if you used the laptop on your lap)
Folks, IBM is not Moto, and some of the arguments that cry out for more than one supplier of PPC chips are not the best ones. For example, IBM is not going to decide to abandon the high end CPU market for the embedded market. IBM does not view the PPC architecture as an architecture for embedded CPU's. IBM has been and will continue to position the PPC architecture against x86, x86-64, and IA-64.
What? The 970 is almost certainly aimed at the high-end embedded market. Look, even AMD, which is selling a hell of a lot more processors than the 970 ever will is losing money hand over fist. And, unlike AMD (which doesn't have much choice but is trying to find a way out), if IBM starts losing money on the 970, the line will be gone a month later. Every IBM'er I've talked to has pointed out that no line is allowed to lose money (at this point in time - if the economy picks up, there may be more slack for projects with "potential"). I'd say that the 970 doesn't have a life span of over two years unless one of the following occurs:
1) Apple at least triples its marketshare
2) IBM 970 blades start selling like hotcakes
3) The 970 embedded market takes off
Of the three, I'd say that (3) is by far the most probable scenario (and actually fairly likely at that). Here's hoping that IBM *does* go after the embedded market and does so successfully. Sure Apple won't end up the 970's number 1 customer, but better a shared use processor than no processor at all (or an effectively orphaned processor...)
Remember, this is the IBM that has pretty much left the personal computer market. If the money isn't there, neither is IBM.
Like everyone else, I've got high hopes for the 970, but the "Die Motorola, Die!" stuff is just plain stupid. Motorola has badly messed up the 180->130 transition. So have a *lot* of companies ("Die nVidia, Die!" oops). If the Ars Technica discussion is accurate (and I'm inclined to believe it is), they probably tried to leapfrog everyone else by using low-K at 130 (unlike everyone else) and it blew up in their faces. That combined with inefficient management means that Motorola has a bad few years. However, I can't see anyone who really has Apple's interests at heart wishing Motorola anything other than enormous success (while maintaining well-earned skepticism, of course).
Or is it now a case that it's better that Apple die than Motorola live? \
No, I think Steve Jobs is fixing to put his foot in Moto's ass real soon. Moto has been the bane of Apple's existence ever since the G4 was introduced, and unless Jobs has some masochistic tendencies I don't see him using the G4 one day longer than he must.
Lord knows, he's not one to put up with other's sh*t.
As it is, even if this rumors has some basis in fact, I think it's Moto's last gasp attempt to catch up. Apple and IBM however, are going to leave them in the dust. Their biggest mistake (if this is true) is not owning up to the design faults in the G4 architecture (can you say bus speed?) and fixing it first.
It never ceases to amaze me how old outdated technologies (G4) suddenly get a new lease of life when a newer technology (PPC970) comes along. We've had to pay through the nose for shitty little incremental speed bumps, but now that the technology's on it's way out, they start producing 2GHz parts.
Same thing happened with the CRT monitor. Apparently it was "impossible" to manufacture a completely flat CRT, and then TFT came along and suddenly everybody and their dog was pumping out flat CRTs. It wasn't until DVD hit the mainstream that all these super fast CD-RW drives started appearing.
Coincidence? I think not. Can anybody say D-R-I-P-F-E-E-D?
Motorola is dead, Apple has been burned by this company too much. I say good riddance. They've been screwing Mac users for almost 2 freakin' decades with delays and underperforming processors.
As an owner of a Powermac 6100 & B&W G3 whose about to purchase his next Mac...I couldn't have said it better. Heck, I've been supporting Motorola processors since I fired up my old Amiga 1000, but I could care less about their CPUs now.
I say all Apple funding should go towards supplying IBM with funds for the best consumer CPUs available.
I dont see what you mean.On all the processor intensive tests the 1.42 scores just about where it should relative to the 1.25.It does look like the drives are slower or something on the new machines.
As an owner of a Powermac 6100 & B&W G3 whose about to purchase his next Mac...I couldn't have said it better. Heck, I've been supporting Motorola processors since I fired up my old Amiga 1000, but I could care less about their CPUs now.
I say all Apple funding should go towards supplying IBM with funds for the best consumer CPUs available.
Bye Motorola.
At least some of the 6100s shipped with IBM processors. I remember looking in awe at the IBM logo inside an Apple Macintosh back in the day.
I drifted off topic, didn't I? Motorola can bite me.
Comments
http://www.theregister.co.uk/content/39/30993.html
"Motorola to boost 0.13µ PowerPCs with low-k dielectrics
Speculation that Motorola may soon cease to be a supplier of processors to Apple may be premature. The chip maker yesterday said it had successfully implemented low-k dielectric materials in its 0.18 micron silicon-on-insulator (SOI) processors, bringing an estimated 20 per cent speed bump to the PowerPC line.
Motorola expects to roll out the process on its 0.13 micron chips this month.
The company didn't mention specific models, but on the basis of the kind of chips it said it's planning to implement low-k dielectric materials in, candidate processors include the MPC7457, which has yet to ship but is set to take Motorola's G4 family beyond 1GHz.
"Our goal is to stay with a frequency doubling every 18 months or so, and get into the 2GHz range for PowerPC, but at very low power consumption of, say, 20W," said Dirk Wristers, director of device/integration for Motorola's MOS-13 wafer fab, according to an EE Times report. "The frequency could be higher if we were at higher power."
Low-k dielectric materials are used to improve the insulation between processor circuits. As chips get smaller and their circuit lines come closer together signals in nearby circuits increasingly interfere with each other, effectively slowing the ability of the component transistors to switch on and off. Getting the switch speed up involves raising the voltage and thus the power the transistor - and in turn the whole chip - consumes.
Reducing that cross-circuit interference with low-k dielectrics allows transistors to switch more quickly and draw less power. The trouble is, implementing low-k dielectrics has not proved easy - new materials require new methods, all of which must be thoroughly proved before they can be implemented in commercial products.
Wristers claimed that some chip vendors - he didn't name names - are now waiting for the switch from 0.13 micron fabrication processes to 0.09 micron (90nm) before implementing low-k dielectrics. The implication in his comment is that since Motorola can use the technology in its 0.13 micron chips, it will be able to really run with it when it makes the transition to 90nm.
With Apple expected to announce a shift to IBM's 64-bit PowerPC 970 processor in just a few weeks' time, the timing of Motorola's announcement takes on a new relevance. Speculation is already mounting that Apple will drop Motorola for IBM, ditching the G4 for a mix of 970s and Big Blue's rumoured AltiVec-enabled G3-class processors.
Motorola's top-end product is the 7457, announced earlier this year but not yet in volume production. The 7457 is essentially a 7455 with double the on-die L2 cache (512KB) and fabbed at 0.13 micron rather than 0.18 micron. Older Motorola roadmaps show the 7457 topping out at 1.3GHz. Implementing low-k dielectrics could lift that ceiling significantly.
Interestingly, Motorola said it had been delivering low-k dielectric 0.18 micron SOI processors for a full quarter. The 7455 is just such a chip - Motorola's claim may explain why Apple has had such success overclocking the 1GHz 0.18 micron MPC7455 to 1.42GHz in its Power Mac models..."
Originally posted by Powerdoc
I have read some benchmarks comparing the dual G4 1,42 with the old dual 1,25 ghz. The dual 1,42 is only 3 % better. ... A 2 ghz G4 will not ouperform much the current dual, if the same lame bus is still there.
There is no future for the current G4, and the RM version disappeared from the roadmap.
I don't know what the bus speed means for embedded applications, but in the desktop area this brings the question: is Motorola beginning to resort to RDF in order to sell CPUs which have to wait half their time for the bus? Or is there a sound reason to bump the clock rate any further? Maybe, for IU and FPU it still has some sense, but doesn't it look like wasting time and resources?
Originally posted by Amorph
That's why I'm pulling for Mot to overcome their problems and get back in the game. If Apple goes from single source to single source, their situation hasn't really improved - even if, in the short term, the top-end CPU available to them has.
...
The problem with IBM, of course, is that IBM is huge enough to do things entirely in its own interest, following its own compass, and so they ended up making smaller and smaller G3s rather than jump on the SIMD wagon because it met some internal goal. The best situation for Apple is multiple vendors - that's why they went with AIM in the first place, rather than AM or AI.
Good post.
It certainly is. All the frenzied IBM-olites take note, read, re-read, and read again untill these words have penetrated well past your current CPU frustrations. Apple needs Moto just as badly as they need IBM. A Moto-SMC partnership is reason for optimism regarding Moto. IBM's history is reason for skepticism regarding IBM. If the rumors of Apple v. Moto legal troubles are true, it can only mean that Apple understands their basic need for two CPU suppliers and is leveraging to get moto's g4/5 developments back on track, they know a long law suit would cost them more in the end, than a nice new agreement.
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Originally posted by Matsu
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Not exactly hidden, rather delayed. As long as Apple's and IBM's roads coincide, Apple will be safe and provided with good CPUs galore. But only as long.
Originally posted by Matsu
It certainly is. All the frenzied IBM-olites take note, read, re-read, and read again untill these words have penetrated well past your current CPU frustrations. Apple needs Moto just as badly as they need IBM. A Moto-SMC partnership is reason for optimism regarding Moto. IBM's history is reason for skepticism regarding IBM. If the rumors of Apple v. Moto legal troubles are true, it can only mean that Apple understands their basic need for two CPU suppliers and is leveraging to get moto's g4/5 developments back on track, they know a long law suit would cost them more in the end, than a nice new agreement.
Moto isn't gone, hopefully. If they are, then Apple is in just as much trouble today as they were yesterday, it's just temporarily hidden.
Good post Matsu.
The problem here is that Motorola is heading down. Right now it seems that it may be able to produce some fairly useful chips (2ghz, 20 watts) that would be nice to have in the iBooks and other computers still stuck in G4 land. And I don't believe that all Apple products will jump ship in a few months. The G4 is going to be with us in some products for some time yet, and it's important that Motorola make them as good as possible.
Some sort of AIM-collaboration would be nice to see again, but I really don't see that happen. I'm sure Moto has had some very good processors in developement that have been axed. Even two years ago we heard talk of 2.4 ghz G5's. If they had collaborated with IBM then they may actually had seen the light of day. But Motorola seems to be having increadibly big problems actually producing the chips they are designing. I don't know what the problem is, but I guess that if IBM could have helped Motorola make the chips, there may have been some fairly fast PPC processors out there.
Another question that must be asked is this: Is moto actually interested in making chips for computers right now, or are they just making them because Steve's all over them??? If they don't want to, Apple should better leave them alone and go IBM as fast as possible.
Originally posted by costique
Not exactly hidden, rather delayed. As long as Apple's and IBM's roads coincide, Apple will be safe and provided with good CPUs galore. But only as long.
You are absolutely right.
The good new is speaking of the computer industry, the road seems to be long.
Originally posted by Matsu
IBM's history is reason for skepticism regarding IBM.
Which "history" are you refering to? The only thing I can think of is that IBM pulled out of AIM when Motorola and Apple wanted to go with AltiVec but IBM wanted to go with high clock rates and focus on scalar performance. Guess who turned out to succeed? And now they have the ability to add vector units to their fast design, which is going to save Apple's bacon. I think this pattern of one hardly qualifies as a track record and you have to compare it against a solid commitment to POWER since about '88 -- 14+ years. Is there something else you are comparing it to?
Originally posted by Programmer
Which "history" are you refering to? The only thing I can think of is that IBM pulled out of AIM when Motorola and Apple wanted to go with AltiVec but IBM wanted to go with high clock rates and focus on scalar performance. Guess who turned out to succeed? And now they have the ability to add vector units to their fast design, which is going to save Apple's bacon. I think this pattern of one hardly qualifies as a track record and you have to compare it against a solid commitment to POWER since about '88 -- 14+ years. Is there something else you are comparing it to?
I wholeheartedly agree. People look at IBM leaving AIM as though IBM gave up on PPC or something like that. This isn't even close to the case. IBM left AIM because they had a differing opinion about the direction the PPC architecture should go. IF Apple had gone with IBM's vision for PPC, then we would probably have 3-4GHz PPC processors without Altivec. IF Apple had gone with IBM, then PPC would probably have been the first to 1 GHz. I like Altivec and think that it is some amazing hardware, but I think that I would trade it for a PPC running at 4GHz.
Folks, IBM is not Moto, and some of the arguments that cry out for more than one supplier of PPC chips are not the best ones. For example, IBM is not going to decide to abandon the high end CPU market for the embedded market. IBM does not view the PPC architecture as an architecture for embedded CPU's. IBM has been and will continue to position the PPC architecture against x86, x86-64, and IA-64. IBM knows how to make fast chips and has clean fabs (witness the speed of the Power 4 chips given the die size and transistor count!). If Apple goes with IBM as its single chip supplier, then at least they have the common goal of making fast desktop chips!
In my ideal world, Moto would update the bus for the G4, shrink the die size and become the portable chip producer for Apple until the 970's undergo a die shrink and are fully usable in portables. I don't hold out much hope for Moto to remain with Apple because they have abandoned the high end market and that is where Apple wants to play.
Even if Mot was at 3 GHz right now, I think IBM would still be developing the PPC970, but it may not have had Altivec, which from what I have read and seen seems to be an add-on anyway. Stevie said awhile back that they were going to have options (and not x86 as a everyone seem to hope for).
Originally posted by Yevgeny
I wholeheartedly agree. People look at IBM leaving AIM as though IBM gave up on PPC or something like that. This isn't even close to the case. IBM left AIM because they had a differing opinion about the direction the PPC architecture should go. IF Apple had gone with IBM's vision for PPC, then we would probably have 3-4GHz PPC processors without Altivec. IF Apple had gone with IBM, then PPC would probably have been the first to 1 GHz. I like Altivec and think that it is some amazing hardware, but I think that I would trade it for a PPC running at 4GHz.
I disagree with this -- I don't think we would have seen anything terribly different than the POWER4/970. This was IBM's direction and when they bailed on AIM all they did was take their toybox and go home.
Originally posted by Programmer
I disagree with this -- I don't think we would have seen anything terribly different than the POWER4/970. This was IBM's direction and when they bailed on AIM all they did was take their toybox and go home.
Once Apple chose Moto and IBM left AIM, then IBM only had to make chips for IBM. This meant making high end server chips, hence the Power 4 with an architecture that was clearly meant for servers. Had Apple stayed with IBM, then IBM would have focused on desktop chips. I seem to recall roadmaps back in the day that indicated 2 GHz in 2002. Of course, all such roadmaps are subject to some fudging, but at least IBM wasn't going to fall off the roadmap because of dirty fabs.
Originally posted by Yevgeny
Once Apple chose Moto and IBM left AIM, then IBM only had to make chips for IBM. This meant making high end server chips, hence the Power 4 with an architecture that was clearly meant for servers. Had Apple stayed with IBM, then IBM would have focused on desktop chips. I seem to recall roadmaps back in the day that indicated 2 GHz in 2002. Of course, all such roadmaps are subject to some fudging, but at least IBM wasn't going to fall off the roadmap because of dirty fabs.
Yes, but I think all being part of AIM would have really done is either accelerate the 970's timeline or give Apple/Moto the fab technology needed to deliver on the G5.
Originally posted by Programmer
Yes, but I think all being part of AIM would have really done is either accelerate the 970's timeline or give Apple/Moto the fab technology needed to deliver on the G5.
And this is a bad thing?
Not to mention, had IBM stayed in AIM, Moto might have gotten a break on IBM's exorbitant contracting rates and been able to borrow the expertise of their researchers and (especially!) their nice, clean fabs. We might have had a PowerBook line powered by 7457s this spring, or even a G5.
g
this, of course, would rely on the PB getting 970 chips (which might not happen till early 2004) or the PB getting 1.6 G4 chips or above (which might not happen ever...plus they would likely cook your naughty bits if you used the laptop on your lap)
Folks, IBM is not Moto, and some of the arguments that cry out for more than one supplier of PPC chips are not the best ones. For example, IBM is not going to decide to abandon the high end CPU market for the embedded market. IBM does not view the PPC architecture as an architecture for embedded CPU's. IBM has been and will continue to position the PPC architecture against x86, x86-64, and IA-64.
1) Apple at least triples its marketshare
2) IBM 970 blades start selling like hotcakes
3) The 970 embedded market takes off
Of the three, I'd say that (3) is by far the most probable scenario (and actually fairly likely at that). Here's hoping that IBM *does* go after the embedded market and does so successfully. Sure Apple won't end up the 970's number 1 customer, but better a shared use processor than no processor at all (or an effectively orphaned processor...)
Remember, this is the IBM that has pretty much left the personal computer market. If the money isn't there, neither is IBM.
Like everyone else, I've got high hopes for the 970, but the "Die Motorola, Die!" stuff is just plain stupid. Motorola has badly messed up the 180->130 transition. So have a *lot* of companies ("Die nVidia, Die!" oops). If the Ars Technica discussion is accurate (and I'm inclined to believe it is), they probably tried to leapfrog everyone else by using low-K at 130 (unlike everyone else) and it blew up in their faces. That combined with inefficient management means that Motorola has a bad few years. However, I can't see anyone who really has Apple's interests at heart wishing Motorola anything other than enormous success (while maintaining well-earned skepticism, of course).
Or is it now a case that it's better that Apple die than Motorola live?
No, I think Steve Jobs is fixing to put his foot in Moto's ass real soon. Moto has been the bane of Apple's existence ever since the G4 was introduced, and unless Jobs has some masochistic tendencies I don't see him using the G4 one day longer than he must.
Lord knows, he's not one to put up with other's sh*t.
As it is, even if this rumors has some basis in fact, I think it's Moto's last gasp attempt to catch up. Apple and IBM however, are going to leave them in the dust. Their biggest mistake (if this is true) is not owning up to the design faults in the G4 architecture (can you say bus speed?) and fixing it first.
As I said, Moto's last gasp.
-tsukurite
It never ceases to amaze me how old outdated technologies (G4) suddenly get a new lease of life when a newer technology (PPC970) comes along. We've had to pay through the nose for shitty little incremental speed bumps, but now that the technology's on it's way out, they start producing 2GHz parts.
Same thing happened with the CRT monitor. Apparently it was "impossible" to manufacture a completely flat CRT, and then TFT came along and suddenly everybody and their dog was pumping out flat CRTs. It wasn't until DVD hit the mainstream that all these super fast CD-RW drives started appearing.
Coincidence? I think not. Can anybody say D-R-I-P-F-E-E-D?
Originally posted by hmurchison
Motorola is dead, Apple has been burned by this company too much. I say good riddance. They've been screwing Mac users for almost 2 freakin' decades with delays and underperforming processors.
As an owner of a Powermac 6100 & B&W G3 whose about to purchase his next Mac...I couldn't have said it better. Heck, I've been supporting Motorola processors since I fired up my old Amiga 1000, but I could care less about their CPUs now.
I say all Apple funding should go towards supplying IBM with funds for the best consumer CPUs available.
Bye Motorola.
Originally posted by Bigc
DP1.42 is worse in some cases.
I dont see what you mean.On all the processor intensive tests the 1.42 scores just about where it should relative to the 1.25.It does look like the drives are slower or something on the new machines.
Originally posted by XB Roughneck
As an owner of a Powermac 6100 & B&W G3 whose about to purchase his next Mac...I couldn't have said it better. Heck, I've been supporting Motorola processors since I fired up my old Amiga 1000, but I could care less about their CPUs now.
I say all Apple funding should go towards supplying IBM with funds for the best consumer CPUs available.
Bye Motorola.
At least some of the 6100s shipped with IBM processors. I remember looking in awe at the IBM logo inside an Apple Macintosh back in the day.
I drifted off topic, didn't I? Motorola can bite me.