If the G4 actually reaches 1.8GHz around MWSF then it could be said that the G4 is scaling better than the P4...from 800MHz to 1.8GHz in ~18 months...225% increase in clock speed.
Yay, Phillips and STMicro can now have the honor of making up for Motorola's incompetence and actually getting their advanced process technologies to work. We had our fill and dumped Moto like a bad habit. <img src="graemlins/oyvey.gif" border="0" alt="[No]" />
<strong>Yay, Phillips and STMicro can now have the honor of making up for Motorola's incompetence and actually getting their advanced process technologies to work. We had our fill and dumped Moto like a bad habit. <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> </strong><hr></blockquote>
So does Moto and AMD no longer have a partnership?
Cell phones? I dont know. I thoght, in general, older fab technology was used for stuff like that. <hr></blockquote>
Cell phones is where the money is. With PC sales growth in the single digits (especially in the current economy), most companies feel that the market has matured. (There's still potential markets, like China, but let's leave that alone for the moment.)
Sales of cell phones have also declined in this economy, but it's still better than PCs. Plus, what's more disposable-- a cell phone or a PC? This is literally a<a href="http://www.ananova.com/news/story/sm_654158.html?menu=news.technology"; target="_blank">World + Dog</a> technology.
[edit: I keep hitting that <add reply> button in the middle of sentences. Must be time for my medica-
<strong>Actually, Mot's been running a .13 micron fab for months now. They just haven't moved the G4 onto it. Now we know why.</strong><hr></blockquote>
From what I gather, moving to .13 (and smaller) processes requires significant redesign in order to combat poorly understood effects in silicon at that level. But once you have a template that produces acceptable yields at .13, moving to .09 is (comparatively) trivial and can mostly be accomplished through lithographic means.
This has two implications:
1) Moving to .13 may not be cost effective for achieving incremental performance gains from current designs.
2) Given item 1 is true, then it makes economic sense to invest in slightly longer development times to jump to fabbing parts at .09 without going to .13 at all.
And what does that mean for the G4? It means that it is unlikely we will see a G4 on .09 process and more likely we will see something like a G4++ or other significant modification at .09 and that production level quantities of this new version are sometime away yet.
Because if you are going to have to spend the time to redesign just to work around the physics at that level, you might as well tweak the design for additional performance (beyond clock speed gains) because the development time is going to be significant in either case.
Steve Jobs is a master of playing one supplier against another. He can sit back and cherry-pick what components to use while the manufacturers kill themselves competing. Both Nvidai and ATi would kill to get an exclusive to do all the graphics for Apple, but the users benefit as each strives to make the next great Mac graphics card. I'm sure SJ will let Moto think they have a future with the Mac (and indeed they might) but they'll have to keep up with IBM. Personally, I don't believe they'll get to the 0.09 Micron process anytime in the next several years. They're idiots or just don't care enough about the PowerPC. Either way, screw 'em. They're done.
Steve Jobs is a master of playing one supplier against another. He can sit back and cherry-pick what components to use while the manufacturers kill themselves competing. Both Nvidai and ATi would kill to get an exclusive to do all the graphics for Apple, but the users benefit as each strives to make the next great Mac graphics card.
so how come the quadros and fireGls are not avaialble for Powermacs?
<strong>Personally, I don't believe they'll get to the 0.09 Micron process anytime in the next several years. They're idiots or just don't care enough about the PowerPC.</strong><hr></blockquote>
They do care, they just happen to be more focused on the embedded market than on the desktop one. But since power consumption is a lot more important in embedded systems than it is in your average PC, and shrinking feature size is one way to significantly reduce power consumption, there could actually be quite an incentive for Moto to move to a 90nm process rather sooner than later.
[quote] Early this week, Motorola, STMicroelectronics and Philips jointly unveiled a design platform to build chips based on 90-nanometer circuitry, versus the current 130-nanometer standard.<hr></blockquote>
YAY! another alliance that will fall apart!!!
seriously though...
anyone know how much on-chip L2 they could fit on a .9nm G3/G4?
Don't believe all the marketing hype you hear about the so-called professional cards.</strong><hr></blockquote>
Oh, no kidding. I used to work at Evans & Sutherland (the grandaddy of all VR firms, and most of the graphics innovations to boot) and they almost lost the farm on trying to produce $2k-5k 'pro' boards. If Jupiter had shipped on time, it would have been amazing. But when it slipped a year, nVidia had already delivered 90% of the performance for 10% of the price. I don't think they're even in the business of such cards anymore.
I've read all the speculation about the Power4 and I think this is the way forward for Apple, but I am unconvinced that it will make it to Quicksilver/mirrors or whatever we call them now. I suspect that they will be headed for a new machine above the current top end aimed directly at the pro-video market.
How does the above statement relate to this thread? Well the above statement may well be wrong, but what it clear is that any Power4 derivative would be unsuitable for a portable machine. Therefore it is quite concievable that Moto has a bright future in Powerbooks. Current Powerbooks seem stuck at or near there current levels. The only way to reduce the heat is to move to a better fab and what could be better then 0.09!
I'm not so sure a POWER 'derivative' wouldn't be suitable. I know nothing about this stuff but when you strip out the huge data caches and all the transistors needed to make that work, mix in a process shrink, and modify other assorted bits, you probably end up with a P4 sized chip, and even that goes into laptops.
<strong>I've read all the speculation about the Power4 and I think this is the way forward for Apple, but I am unconvinced that it will make it to Quicksilver/mirrors or whatever we call them now. I suspect that they will be headed for a new machine above the current top end aimed directly at the pro-video market.</strong><hr></blockquote>
Two words: Heat Sink.
Either the '1.25 GHz G4' is Apple playing with overclocking, or there's something a lot hotter than normal aimed at that spot. A 10cm cube of Al pegs with a 120mm fan aimed straight into it is a heck of a lot of heat dissipation.
I mean, '7 lbs' is insane - but I can sure see where that idea originated. 10x10x10cm is just extreme overkill... for a G4.
I do know the mobo (new one not released yet) was blowing up in the towers because of the heat. That's why they made the holes on the front of the mirrored QSs. So the current case may be similar to the next gen new mobo towers. They were having some issues with the new mobo and that's adding to the delay. I don't think this new mobo is the G5 or power4, I think it;s the next gen G4 or maybe just the top pro tower leaving the power4 or G5 as a workstation.
<strong>This means nothing. For all we know it's in reference to the G5 aimed at communications (8400?).
I highly doubt Moto is going to fab advanced desktop G4s on a 90 nm process. That's just way too far out there for a loser company like Motorola.
IF Moto were to fab a G4 on such a process, then it would probably still only clock to 1.4 GHz or so, because Moto sucks. And of course it would be limited to a 166 MHz FSB. Forget RIO on the G4, it would be too much work to redesign the G4 to use RIO. RIO is for the G5 aimed at the communications market.</strong><hr></blockquote>
Funny! What for is MOT to blame? Was it stupid to cut down investment on an uncertain single customer Apple? Or to not trust SJ who generated a $95 mil. loss instead of letting MOT become the number Mac producer. BTW SJ did not gain any marketshare since him came back 1997 so why should MOT burn their money for Apple?
<strong>Funny! What for is MOT to blame?</strong><hr></blockquote>
For years of bafflingly incompetent management frittering away the company's money, talent and discipline. What finally turned them around was a shareholder revolt.
It's not Apple's fault they're cutting costs by turning the fans off in their fabs. Apple's cancelling the clones cost them a relatively small amount of money, and stopped a program that Mot hadn't made a dime from. Mismanagement has cost Mot billions of dollars and losses in market share in all segments.
[quote]<strong>Was it stupid to cut down investment on an uncertain single customer Apple? Or to not trust SJ who generated a $95 mil. loss instead of letting MOT become the number Mac producer. BTW SJ did not gain any marketshare since him came back 1997 so why should MOT burn their money for Apple?</strong><hr></blockquote>
Mot SPS is in no position to turn down customers - literally - and Apple is in one of their big ones. There are plenty of other customers (Cisco, the telcos) who will buy processors as powerful as Mot can make them. Mot's design priorities still dovetail with Apple's in three out of their four computer lines. Given that the SPS is currently trying to turn a profit in order to prevent being axed by Mot's newly refocused management, and management is trying to prevent the shareholders from calling for their heads, a $100 million dollar mistake from several years ago is just going to vanish. CHRP/PREP is dead. Everyone knows that now.
Mot SPS is in no position to turn down customers - literally - and Apple is in one of their big ones. There are plenty of other customers (Cisco, the telcos) who will buy processors as powerful as Mot can make them. Mot's design priorities still dovetail with Apple's in three out of their four computer lines. Given that the SPS is currently trying to turn a profit in order to prevent being axed by Mot's newly refocused management, and management is trying to prevent the shareholders from calling for their heads, a $100 million dollar mistake from several years ago is just going to vanish. CHRP/PREP is dead. Everyone knows that now.</strong><hr></blockquote>
1.) I think you overestimate the value of Apple for MOT SPS. They selling a lot more PPCs to other customers in total. At some time Cisco was buying more PPCs than Apple. All actions of the last years clearly show that Apple isn't that important to MOT as some think (wish).
2.) Regardless how stupid MOT's management was (which it surely was, no argue 'bout that), their decision to cut down investment on the desktop PPCs was right. Look at Apple's current marketshare. It is still not clear if Apple will ever be able to return to it's former share.
3.) Everyone knew that MOT would become the number 1 mac producer, so in reality MOT not lose only $95 mil but whole opportunities tide with it. And SJ pissed off MOT big time, believe it or not.
Comments
Knowing Motorola, look for 1.6GHz.
<strong>Yay, Phillips and STMicro can now have the honor of making up for Motorola's incompetence and actually getting their advanced process technologies to work. We had our fill and dumped Moto like a bad habit. <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> </strong><hr></blockquote>
So does Moto and AMD no longer have a partnership?
communication could be a G4..
Cell phones? I dont know. I thoght, in general, older fab technology was used for stuff like that.
Cell phones is where the money is. With PC sales growth in the single digits (especially in the current economy), most companies feel that the market has matured. (There's still potential markets, like China, but let's leave that alone for the moment.)
Sales of cell phones have also declined in this economy, but it's still better than PCs. Plus, what's more disposable-- a cell phone or a PC? This is literally a<a href="http://www.ananova.com/news/story/sm_654158.html?menu=news.technology"; target="_blank">World + Dog</a> technology.
[edit: I keep hitting that <add reply> button in the middle of sentences. Must be time for my medica-
[ 09-02-2002: Message edited by: GardenOfEarthlyDelights ]</p>
<strong>We had our fill and dumped Moto like a bad habit. <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> </strong><hr></blockquote>
Bad habits are hard to dump though...they always creep back into your routine subconciously.
<strong>Actually, Mot's been running a .13 micron fab for months now. They just haven't moved the G4 onto it. Now we know why.</strong><hr></blockquote>
From what I gather, moving to .13 (and smaller) processes requires significant redesign in order to combat poorly understood effects in silicon at that level. But once you have a template that produces acceptable yields at .13, moving to .09 is (comparatively) trivial and can mostly be accomplished through lithographic means.
This has two implications:
1) Moving to .13 may not be cost effective for achieving incremental performance gains from current designs.
2) Given item 1 is true, then it makes economic sense to invest in slightly longer development times to jump to fabbing parts at .09 without going to .13 at all.
And what does that mean for the G4? It means that it is unlikely we will see a G4 on .09 process and more likely we will see something like a G4++ or other significant modification at .09 and that production level quantities of this new version are sometime away yet.
Because if you are going to have to spend the time to redesign just to work around the physics at that level, you might as well tweak the design for additional performance (beyond clock speed gains) because the development time is going to be significant in either case.
so how come the quadros and fireGls are not avaialble for Powermacs?
[ 09-02-2002: Message edited by: spooky ]</p>
<strong>Personally, I don't believe they'll get to the 0.09 Micron process anytime in the next several years. They're idiots or just don't care enough about the PowerPC.</strong><hr></blockquote>
They do care, they just happen to be more focused on the embedded market than on the desktop one. But since power consumption is a lot more important in embedded systems than it is in your average PC, and shrinking feature size is one way to significantly reduce power consumption, there could actually be quite an incentive for Moto to move to a 90nm process rather sooner than later.
Bye,
RazzFazz
YAY! another alliance that will fall apart!!!
seriously though...
anyone know how much on-chip L2 they could fit on a .9nm G3/G4?
but yes, it appears we have a .9nm race...
<a href="http://www.anandtech.com/cpu/showdoc.html?i=1677"; target="_blank">http://www.anandtech.com/cpu/showdoc.html?i=1677</a>;
[ 09-04-2002: Message edited by: gumby5647 ]</p>
<strong>
Don't believe all the marketing hype you hear about the so-called professional cards.</strong><hr></blockquote>
Oh, no kidding. I used to work at Evans & Sutherland (the grandaddy of all VR firms, and most of the graphics innovations to boot) and they almost lost the farm on trying to produce $2k-5k 'pro' boards. If Jupiter had shipped on time, it would have been amazing. But when it slipped a year, nVidia had already delivered 90% of the performance for 10% of the price. I don't think they're even in the business of such cards anymore.
How does the above statement relate to this thread? Well the above statement may well be wrong, but what it clear is that any Power4 derivative would be unsuitable for a portable machine. Therefore it is quite concievable that Moto has a bright future in Powerbooks. Current Powerbooks seem stuck at or near there current levels. The only way to reduce the heat is to move to a better fab and what could be better then 0.09!
<strong>I've read all the speculation about the Power4 and I think this is the way forward for Apple, but I am unconvinced that it will make it to Quicksilver/mirrors or whatever we call them now. I suspect that they will be headed for a new machine above the current top end aimed directly at the pro-video market.</strong><hr></blockquote>
Two words: Heat Sink.
Either the '1.25 GHz G4' is Apple playing with overclocking, or there's something a lot hotter than normal aimed at that spot. A 10cm cube of Al pegs with a 120mm fan aimed straight into it is a heck of a lot of heat dissipation.
I mean, '7 lbs' is insane - but I can sure see where that idea originated. 10x10x10cm is just extreme overkill... for a G4.
<strong>This means nothing. For all we know it's in reference to the G5 aimed at communications (8400?).
I highly doubt Moto is going to fab advanced desktop G4s on a 90 nm process. That's just way too far out there for a loser company like Motorola.
IF Moto were to fab a G4 on such a process, then it would probably still only clock to 1.4 GHz or so, because Moto sucks. And of course it would be limited to a 166 MHz FSB. Forget RIO on the G4, it would be too much work to redesign the G4 to use RIO. RIO is for the G5 aimed at the communications market.</strong><hr></blockquote>
Funny! What for is MOT to blame? Was it stupid to cut down investment on an uncertain single customer Apple? Or to not trust SJ who generated a $95 mil. loss instead of letting MOT become the number Mac producer. BTW SJ did not gain any marketshare since him came back 1997 so why should MOT burn their money for Apple?
End of Line
<strong>Funny! What for is MOT to blame?</strong><hr></blockquote>
For years of bafflingly incompetent management frittering away the company's money, talent and discipline. What finally turned them around was a shareholder revolt.
It's not Apple's fault they're cutting costs by turning the fans off in their fabs. Apple's cancelling the clones cost them a relatively small amount of money, and stopped a program that Mot hadn't made a dime from. Mismanagement has cost Mot billions of dollars and losses in market share in all segments.
[quote]<strong>Was it stupid to cut down investment on an uncertain single customer Apple? Or to not trust SJ who generated a $95 mil. loss instead of letting MOT become the number Mac producer. BTW SJ did not gain any marketshare since him came back 1997 so why should MOT burn their money for Apple?</strong><hr></blockquote>
Mot SPS is in no position to turn down customers - literally - and Apple is in one of their big ones. There are plenty of other customers (Cisco, the telcos) who will buy processors as powerful as Mot can make them. Mot's design priorities still dovetail with Apple's in three out of their four computer lines. Given that the SPS is currently trying to turn a profit in order to prevent being axed by Mot's newly refocused management, and management is trying to prevent the shareholders from calling for their heads, a $100 million dollar mistake from several years ago is just going to vanish. CHRP/PREP is dead. Everyone knows that now.
<strong>
Mot SPS is in no position to turn down customers - literally - and Apple is in one of their big ones. There are plenty of other customers (Cisco, the telcos) who will buy processors as powerful as Mot can make them. Mot's design priorities still dovetail with Apple's in three out of their four computer lines. Given that the SPS is currently trying to turn a profit in order to prevent being axed by Mot's newly refocused management, and management is trying to prevent the shareholders from calling for their heads, a $100 million dollar mistake from several years ago is just going to vanish. CHRP/PREP is dead. Everyone knows that now.</strong><hr></blockquote>
1.) I think you overestimate the value of Apple for MOT SPS. They selling a lot more PPCs to other customers in total. At some time Cisco was buying more PPCs than Apple. All actions of the last years clearly show that Apple isn't that important to MOT as some think (wish).
2.) Regardless how stupid MOT's management was (which it surely was, no argue 'bout that), their decision to cut down investment on the desktop PPCs was right. Look at Apple's current marketshare. It is still not clear if Apple will ever be able to return to it's former share.
3.) Everyone knew that MOT would become the number 1 mac producer, so in reality MOT not lose only $95 mil but whole opportunities tide with it. And SJ pissed off MOT big time, believe it or not.
End of Line