With a quick transition to 0.09 970s in 04...I don't see how Moto' on current/past performance can stay in the game for longer than what? Another 9 months?
To me, it looks like desktop G5 all the way in 2004. Laptops too. By next March? A Powerbook G5? And are Apple going to have a G4 in their iMac 2 if the Powerbook has a G5? A pathetic looking 1.3? 1.6 G4 next to a 4 gig Prescott? iMac2 sales are less than amazing. Still selling the same crappy card, bus and geriatric G4 for another 13 months won't ignite sales in my opinion. The original iMac had tiered 'cheapness' on its side (and it wasn't really THAT cheap at that in some ways...) The iMac2 needs something special inside to go with something special that it has on the outside. An update is overdue for it already.
Sorry, but I think IBM have been brought in to help Apple play hardball again. And pathetic 1.3 G4s in 2004 won't do. Not for me.
A desperate company may surprise you. Apple did once upon a time... 0.09 G4 from Moto'? I can't get my head round it. They're so far behind, they'd have to do something they haven't been able to do so far even to close the gap.
A consumer chip looks more likely to come from IBM at the moment.
Multiple suppliers is good if they can both produce the goods. Once ATi wasn't doing too well. But the competition from Nvidia did Apple and Ati the power of good.
I tend to gravitate towards Dr. Boar's opinion on this one.
But Motorola can't be ruled out entirely. But more likey would be some low power G4 at 1.3 gig for the Pro laptops. That could go in the iMac2 and eMac for the 2nd half of this year. Not spectacular. Especially with the current bus. Pretty pathetic, really.
But with the cooling requirements of the G5 at .13...it looks like its G5 at the top and Moto very much at the bottom.
A 0.09 dual core G4 upto 2 gig? Believe it when I see it.
An IBM G3+ by 2004. By then Moto' and Apple's relationship will be on its death throws...
I dunno. I still see Moto' on their way out 'huge'(ly).
With a quick transition to 0.09 970s in 04...I don't see how Moto' on current/past performance can stay in the game for longer than what? Another 9 months?
Crolles is good for 90nm, 65nm, and one more process shrink beyond that. When you're forking out billions of dollars (or, in this case, Euros) for a fab, you want to get a few years of service out of it.
Remember, Mot isn't going it alone this time. STM is the largest contract fabricator in Europe, and Philips is, well, Philips.
I couldn't help noticing the shift from Mot's official rhetoric a year or so ago ("hoping", "planning", etc.) to the most recent articles where they talk about doing.
I understand the frustration with Motorola - anyone who isn't aghast at the last three years hasn't been paying attention - but if you're wishing for one of Apple's two CPU suppliers to get lost, be mindful that you might get what you wish for.
I think Motorola's 7457 holds some promise in the near term, over the next year perhaps, in giving a boost to Apple's portables and low-to-midrange desktop systems.
But for Motorola to really be a serious proposition going forward, they need a new, faster FSB. The problem is, if they come out with a chip with new FSB technology, it'll no longer be a drop-in replacement for the old G4.
So, if Apple has to redesign their motherboards for a new chip, why not redesign for the G5, something they're already doing, instead of redesigning for yet another new chip?
If Motorola doesn't improve their bus speed soon, Apple with have to drop them from the entire product line eventually. But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
Crolles is good for 90nm, 65nm, and one more process shrink beyond that. When you're forking out billions of dollars (or, in this case, Euros) for a fab, you want to get a few years of service out of it.
Remember, Mot isn't going it alone this time. STM is the largest contract fabricator in Europe, and Philips is, well, Philips.
I couldn't help noticing the shift from Mot's official rhetoric a year or so ago ("hoping", "planning", etc.) to the most recent articles where they talk about doing.
I understand the frustration with Motorola - anyone who isn't aghast at the last three years hasn't been paying attention - but if you're wishing for one of Apple's two CPU suppliers to get lost, be mindful that you might get what you wish for.
Actually, I think that it would be better, for Apple anyway, to renue the continuity in chip design that was around when AIM was young, and the 604 reigned supriem. The idea of a divergent path in chips from consumer to pro lines seams to me like it would add to R&D cost of both systems and could make optemising software harder in the long run.
I know this isnt going to happen. I doubt that IBM and Moto will ever work together on a desktop chip again. Apple will have to make the best of what they can get, which from IBM is pretty good today. I would hope that Apple takes to heart the MP platform, I think that for the "hub" idea it will be the winner in 2-4 years as home systems are required to do more parellel tasks. For the developers sake I think it would be great if Apple also moved the whole line to 64 bit, so that it is more likely that software will be optamized for it.
If Motorola doesn't improve their bus speed soon, Apple with have to drop them from the entire product line eventually. But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
The bus doesn't matter if they bring the memory controller on-chip, which is what I expect Motorola to do with their PowerPC line (assuming it has a future). This works well in the embedded market where Motorola spends its effort, and their designs are typically low power for the same reason. If we continue to see Motorola chips in Apple machines they'll be in the portables and consumer desktops.
Matsu, the iMac needs to be fast. It needs to get a G5 as soon as possible. The iMac is the high-end consumer machine, and it needs to be high-end. It needs a great graphics card too. Let the dollar-conscious folks buy an eMac with a G4.
NO, absolutely not. This an illusion created by the lack of a credible high end pro machine. Ironically, it limited the ability of consumer machines, not so much to get faster, since the pro chips were there, but for them to get cheaper, lest they make PM values seem more abysmal than ever.
This problem is now solved, convincingly, by the G5.
The consumer machines are now free to get a whole lot cheaper because that will not impact the relative value of powerhouse pro towers at all, prior to the G5, it would have.
Apple has stated a commitment (with the continuation of the PMG4) to a middle tier, and that will certainly get a G5 before the iMac ever does.
The eMac appears to be a better value, but the CRT is hard to take at this point. It was 2 years too late, and it's days are numbered unless Apple is willing to take it into commodity priced territory.
It is much more important to get the iMac down to 1299 for a 17" and 899-999 for a 15" than it is to chase a G5 that could be sold to better effect in more pro duals, a middle tier machine (now occupied by the PMG4) and ultimately Powerbooks.
No no. The iMac has a LONG G4 based career ahead of it.
The bus doesn't matter if they bring the memory controller on-chip, which is what I expect Motorola to do with their PowerPC line (assuming it has a future). This works well in the embedded market where Motorola spends its effort, and their designs are typically low power for the same reason. If we continue to see Motorola chips in Apple machines they'll be in the portables and consumer desktops.
IBM appears to serve the embedded market too, from what I read, but Gobi does not have a memory controller. I wonder whether Mojave might have a memory controller, as well as the AltiVec? From what you say, it looks like chips for Apple's low end desktop and laptop Macs have a lot in common with the embedded market chips. Very low power, low cost and reasonably good performance.
NO, absolutely not. This an illusion created by the lack of a credible high end pro machine. Ironically, it limited the ability of consumer machines, not so much to get faster, since the pro chips were there, but for them to get cheaper, lest they make PM values seem more abysmal than ever.
This problem is now solved, convincingly, by the G5.
The consumer machines are now free to get a whole lot cheaper because that will not impact the relative value of powerhouse pro towers at all, prior to the G5, it would have.
Apple has stated a commitment (with the continuation of the PMG4) to a middle tier, and that will certainly get a G5 before the iMac ever does.
The eMac appears to be a better value, but the CRT is hard to take at this point. It was 2 years too late, and it's days are numbered unless Apple is willing to take it into commodity priced territory.
It is much more important to get the iMac down to 1299 for a 17" and 899-999 for a 15" than it is to chase a G5 that could be sold to better effect in more pro duals, a middle tier machine (now occupied by the PMG4) and ultimately Powerbooks.
No no. The iMac has a LONG G4 based career ahead of it.
In fact, IBM made a great favour with G5 to MOTO!
If the mid and low line of computers is to implement the G4, than MOTO will sell to APPLE more CPU in future than in the past.
But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
There will be a G4/5 division at Apple for more than one year.
And I suspect that after the G4 passes, you might even see a G5lite from Moto before you see an ALL IBM lineup.
Just watch. In the end it makes sense for Apple to have two suppliers. IBM has a big plant to pay for, but they also have a lot of potential clientele as the silicon industry goes through some major changes. Fabs aree shaping up to no longer be things that even relatively wealthy chip makers can afford. I suspect fishkill will get a lot of "fab" work, cranking out chips under contract for other people's designs. AMD, for one, Apple for the first (with their custom controller) and anyone else who can pay for the design work but just can afford to have a modern fab of their own.
In the end, people making top end small process fab may shrink to IBM, Intel, and SMC-Moto-phillips.
After 12 months the G4 to G5 distinction will look like this:
G5:
Powermacs, Xserves, a headless middle tier "cube" like machine to replace the PMG4, and Powerbooks.
G4:
iMacs, eMacs, iBooks.
And that division will hold untill basically untill the G6 is almost ready, or a G5Lite is ressurected at Moto.
What moto has done at .18u is rather amazing in terms of power consumption, at .13u they'd clearly be ahead of anyone, and at .09u nobody will have more performance per watt/dollar than Moto.
What moto has done at .18u is rather amazing in terms of power consumption, at .13u they'd clearly be ahead of anyone, and at .09u nobody will have more performance per watt/dollar than Moto.
I think the watt/speed issue is overrated. In towers it does not matter at all, the user want speed and do not care about heat. Also the eMac and iMac can take a lot of heat ,so to speak.
Even in portables power consumption only matter to a point. Many times the porteble is actully connected to power and then it hardly matters. And even on the road it is the [U]total consumption including, hard disk, optical drive, memory, screen and so on that matters. If the CPU is say 20W and the rest is 80W then getting a zero W CPU only increase battery time by 25%
I think it time to forget the crap about "elegant implementation of the G4" or the technical skill of making a 500 MHz CPU with pipeline of only 4 stage. If it fast enough and does not run too hot it is good. We have been shafted by dog slow G4s since 1999, it is time to pull out
That was then, this is now. If, IF, the FSB can be fixed, I'm willing to bet that the G4 can put on a very good show, it's the only real design flaw, the rest has been a fabrication nightmare, not an inherent failing of the G4 PPC.
They pulled the stages out to 7, and made process enhancements along the way. Clearly, since that time, the demands of a high-end desktop chip have moved well beyond what makes a capable laptop chip or even a part for a small AIO enclosure. And there is also the issue of cost, small process G4's will most certainly cost less than PPC970.
When I wrote that, the G4 was extremely limited and stretched way too thin. Apple had to use it to supply high and low end needs, and it was just spread too thin -- hindered in the consumer product because it was not allowed to approach the speeds of the Pro offerings, and inadequate in the pro offerings because it simply couldn't achieve adequate performance for machines of their cost. Back then we were contemplating the power consumption of G4's and even more pitiful G3's, a true fossil in the CPU world. The G5 has moved the mark enough, that if the best of the G4 line (process shrunk, modern FSB, and cheaper cost) were to be used throughout the consumer line-up then it would allow for competitive performance and small AFFORDABLE packages.
Would you really balk at using 1.3-2Ghz G4's with a modern FSB in a 799-1299 range of e/iMac and iBook products? Mebbe then you protest merely out of habit?
I favor MUCH LOWER prices on the consumer machines. Don't believe the cheap PPC970 line for a second. How could that be? The prices went up on the low end, the heat sincs are huge. No, the PPC970 costs more, and IS hot. Why would IBM still be tinkering with an expanded G3 performance envelope if they could simply down-clock the G5 into the appropriate heat and cost parameters? Simple answer, they can't, not yet.
Everything I argued is still true of towers, build it first, find a way to make it fit afterwards, even by nine fans if need be.
Matsu, you're right. Though I hoped that a low-end G5 would not be over $1500, I agree that those are machines of a higher level. One thing is not so clear, though: 2x2GHz is not an average consumer PC, nor is a 3GHz P4; what about 1.6GHz G5 vs. 2GHz P4? The low-end G5 seems a bit overpriced and that really suggests that consumer Macs and G5s will get further apart in prices.
One day, a 5GHz Mac will cost $599, but not now. As long as there are computers to help make money, their makers will make money selling them.
The 970 is what was said to be cheaper, not the G5. The support chips (and there are several of them) plus the mobo capable of carrying the 1 GHz FSB, plus the case are what drive the price of the Apple machine up.
The 970 is also supposed to be cheaper than the 7455, not the 7457. If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
I don't think Moto will "fix" the FSB of the G4 any time soon, although it might be improved by moving to 200 MHz. MPX won't go DDR, they'll go to RIO first and (hopefully) bring the memory controller onto the processor. That will improve price, performance, integration, and reduce power consumption further.
If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
From all their current docs, it sure seems that Motorola is considering the 7457 for the embedded market. It is so hard to gauge their direction for future Apple chips.
The 970 is what was said to be cheaper, not the G5. The support chips (and there are several of them) plus the mobo capable of carrying the 1 GHz FSB, plus the case are what drive the price of the Apple machine up.
True, but the G4 needs expensive L3 cache to boost performance, let alone any speculation on yield issues.
Quote:
Originally posted by Programmer
The 970 is also supposed to be cheaper than the 7455, not the 7457. If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
But, as you stated, they're still left using @ best a 200MGz FSB, thus aggravating an already excruciatingly, painfully, woefully choked FSB. Their answer, support for even more, up to 4MB/processor, very expensive L3 cache.
Quote:
Originally posted by Programmer
I don't think Moto will "fix" the FSB of the G4 any time soon, although it might be improved by moving to 200 MHz. MPX won't go DDR, they'll go to RIO first and (hopefully) bring the memory controller onto the processor. That will improve price, performance, integration, and reduce power consumption further.
You hit the nail on the head with the statement,"any time soon". The last information from Motorola had the MPC7457-rm(re: using Rapid I/O) as only "proposed".
If and when this cpu appears it won't be before 2005, where will IBM be in this time frame. I like Motorola, and they have apparantly done what they could for Apple in getting the MPC7455 up to an astonishing 1.42GHz, but I hold no hope for miracles from them. Maybe collaboration w/ Phillips and SMT will speed any development, but the future of Apple's consumer machines looks fairly grim if they stay w/ Motorola cpu's.
Oh, I forgot to say I'm going to be buying an Apple computer with a G5 in it as soon as they hit the stores. Looks to be either the low end tower or mid-range tower at this moment, but hope springs eternal for additional introductions between now and the August time frame.
Comments
To me, it looks like desktop G5 all the way in 2004. Laptops too. By next March? A Powerbook G5? And are Apple going to have a G4 in their iMac 2 if the Powerbook has a G5? A pathetic looking 1.3? 1.6 G4 next to a 4 gig Prescott? iMac2 sales are less than amazing. Still selling the same crappy card, bus and geriatric G4 for another 13 months won't ignite sales in my opinion. The original iMac had tiered 'cheapness' on its side (and it wasn't really THAT cheap at that in some ways...) The iMac2 needs something special inside to go with something special that it has on the outside. An update is overdue for it already.
Sorry, but I think IBM have been brought in to help Apple play hardball again. And pathetic 1.3 G4s in 2004 won't do. Not for me.
A desperate company may surprise you. Apple did once upon a time... 0.09 G4 from Moto'? I can't get my head round it. They're so far behind, they'd have to do something they haven't been able to do so far even to close the gap.
A consumer chip looks more likely to come from IBM at the moment.
Multiple suppliers is good if they can both produce the goods. Once ATi wasn't doing too well. But the competition from Nvidia did Apple and Ati the power of good.
I tend to gravitate towards Dr. Boar's opinion on this one.
But Motorola can't be ruled out entirely. But more likey would be some low power G4 at 1.3 gig for the Pro laptops. That could go in the iMac2 and eMac for the 2nd half of this year. Not spectacular. Especially with the current bus. Pretty pathetic, really.
But with the cooling requirements of the G5 at .13...it looks like its G5 at the top and Moto very much at the bottom.
A 0.09 dual core G4 upto 2 gig? Believe it when I see it.
An IBM G3+ by 2004. By then Moto' and Apple's relationship will be on its death throws...
I dunno. I still see Moto' on their way out 'huge'(ly).
Lemon Bon Bon
4 years later.
Lemon Bon Bon
I B M
Originally posted by Lemon Bon Bon
With a quick transition to 0.09 970s in 04...I don't see how Moto' on current/past performance can stay in the game for longer than what? Another 9 months?
Crolles is good for 90nm, 65nm, and one more process shrink beyond that. When you're forking out billions of dollars (or, in this case, Euros) for a fab, you want to get a few years of service out of it.
Remember, Mot isn't going it alone this time. STM is the largest contract fabricator in Europe, and Philips is, well, Philips.
I couldn't help noticing the shift from Mot's official rhetoric a year or so ago ("hoping", "planning", etc.) to the most recent articles where they talk about doing.
I understand the frustration with Motorola - anyone who isn't aghast at the last three years hasn't been paying attention - but if you're wishing for one of Apple's two CPU suppliers to get lost, be mindful that you might get what you wish for.
But for Motorola to really be a serious proposition going forward, they need a new, faster FSB. The problem is, if they come out with a chip with new FSB technology, it'll no longer be a drop-in replacement for the old G4.
So, if Apple has to redesign their motherboards for a new chip, why not redesign for the G5, something they're already doing, instead of redesigning for yet another new chip?
If Motorola doesn't improve their bus speed soon, Apple with have to drop them from the entire product line eventually. But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
Originally posted by Amorph
Crolles is good for 90nm, 65nm, and one more process shrink beyond that. When you're forking out billions of dollars (or, in this case, Euros) for a fab, you want to get a few years of service out of it.
Remember, Mot isn't going it alone this time. STM is the largest contract fabricator in Europe, and Philips is, well, Philips.
I couldn't help noticing the shift from Mot's official rhetoric a year or so ago ("hoping", "planning", etc.) to the most recent articles where they talk about doing.
I understand the frustration with Motorola - anyone who isn't aghast at the last three years hasn't been paying attention - but if you're wishing for one of Apple's two CPU suppliers to get lost, be mindful that you might get what you wish for.
Actually, I think that it would be better, for Apple anyway, to renue the continuity in chip design that was around when AIM was young, and the 604 reigned supriem. The idea of a divergent path in chips from consumer to pro lines seams to me like it would add to R&D cost of both systems and could make optemising software harder in the long run.
I know this isnt going to happen. I doubt that IBM and Moto will ever work together on a desktop chip again. Apple will have to make the best of what they can get, which from IBM is pretty good today. I would hope that Apple takes to heart the MP platform, I think that for the "hub" idea it will be the winner in 2-4 years as home systems are required to do more parellel tasks. For the developers sake I think it would be great if Apple also moved the whole line to 64 bit, so that it is more likely that software will be optamized for it.
Originally posted by shetline
If Motorola doesn't improve their bus speed soon, Apple with have to drop them from the entire product line eventually. But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
The bus doesn't matter if they bring the memory controller on-chip, which is what I expect Motorola to do with their PowerPC line (assuming it has a future). This works well in the embedded market where Motorola spends its effort, and their designs are typically low power for the same reason. If we continue to see Motorola chips in Apple machines they'll be in the portables and consumer desktops.
This problem is now solved, convincingly, by the G5.
The consumer machines are now free to get a whole lot cheaper because that will not impact the relative value of powerhouse pro towers at all, prior to the G5, it would have.
Apple has stated a commitment (with the continuation of the PMG4) to a middle tier, and that will certainly get a G5 before the iMac ever does.
The eMac appears to be a better value, but the CRT is hard to take at this point. It was 2 years too late, and it's days are numbered unless Apple is willing to take it into commodity priced territory.
It is much more important to get the iMac down to 1299 for a 17" and 899-999 for a 15" than it is to chase a G5 that could be sold to better effect in more pro duals, a middle tier machine (now occupied by the PMG4) and ultimately Powerbooks.
No no. The iMac has a LONG G4 based career ahead of it.
Originally posted by Programmer
The bus doesn't matter if they bring the memory controller on-chip, which is what I expect Motorola to do with their PowerPC line (assuming it has a future). This works well in the embedded market where Motorola spends its effort, and their designs are typically low power for the same reason. If we continue to see Motorola chips in Apple machines they'll be in the portables and consumer desktops.
IBM appears to serve the embedded market too, from what I read, but Gobi does not have a memory controller. I wonder whether Mojave might have a memory controller, as well as the AltiVec? From what you say, it looks like chips for Apple's low end desktop and laptop Macs have a lot in common with the embedded market chips. Very low power, low cost and reasonably good performance.
Originally posted by Matsu
NO, absolutely not. This an illusion created by the lack of a credible high end pro machine. Ironically, it limited the ability of consumer machines, not so much to get faster, since the pro chips were there, but for them to get cheaper, lest they make PM values seem more abysmal than ever.
This problem is now solved, convincingly, by the G5.
The consumer machines are now free to get a whole lot cheaper because that will not impact the relative value of powerhouse pro towers at all, prior to the G5, it would have.
Apple has stated a commitment (with the continuation of the PMG4) to a middle tier, and that will certainly get a G5 before the iMac ever does.
The eMac appears to be a better value, but the CRT is hard to take at this point. It was 2 years too late, and it's days are numbered unless Apple is willing to take it into commodity priced territory.
It is much more important to get the iMac down to 1299 for a 17" and 899-999 for a 15" than it is to chase a G5 that could be sold to better effect in more pro duals, a middle tier machine (now occupied by the PMG4) and ultimately Powerbooks.
No no. The iMac has a LONG G4 based career ahead of it.
In fact, IBM made a great favour with G5 to MOTO!
If the mid and low line of computers is to implement the G4, than MOTO will sell to APPLE more CPU in future than in the past.
Rooster
But even if Motorola does speed up their G4 bus, they'll still have to come out with something very compelling, something that's especially power efficient and cool, or simply especially cheap, for it to be worthwhile for Apple to carry on with Motorola rather than use the cheaper, faster, and cooler-running G5s from IBM that will almost certainly be here in a year's time.
Eloquently said.
Lemon Bon Bon
And I suspect that after the G4 passes, you might even see a G5lite from Moto before you see an ALL IBM lineup.
Just watch. In the end it makes sense for Apple to have two suppliers. IBM has a big plant to pay for, but they also have a lot of potential clientele as the silicon industry goes through some major changes. Fabs aree shaping up to no longer be things that even relatively wealthy chip makers can afford. I suspect fishkill will get a lot of "fab" work, cranking out chips under contract for other people's designs. AMD, for one, Apple for the first (with their custom controller) and anyone else who can pay for the design work but just can afford to have a modern fab of their own.
In the end, people making top end small process fab may shrink to IBM, Intel, and SMC-Moto-phillips.
After 12 months the G4 to G5 distinction will look like this:
G5:
Powermacs, Xserves, a headless middle tier "cube" like machine to replace the PMG4, and Powerbooks.
G4:
iMacs, eMacs, iBooks.
And that division will hold untill basically untill the G6 is almost ready, or a G5Lite is ressurected at Moto.
What moto has done at .18u is rather amazing in terms of power consumption, at .13u they'd clearly be ahead of anyone, and at .09u nobody will have more performance per watt/dollar than Moto.
What moto has done at .18u is rather amazing in terms of power consumption, at .13u they'd clearly be ahead of anyone, and at .09u nobody will have more performance per watt/dollar than Moto.
------------------------------------------------------------------------
By Matsu
I think the watt/speed issue is overrated. In towers it does not matter at all, the user want speed and do not care about heat. Also the eMac and iMac can take a lot of heat ,so to speak.
Even in portables power consumption only matter to a point. Many times the porteble is actully connected to power and then it hardly matters. And even on the road it is the [U]total consumption including, hard disk, optical drive, memory, screen and so on that matters. If the CPU is say 20W and the rest is 80W then getting a zero W CPU only increase battery time by 25%
I think it time to forget the crap about "elegant implementation of the G4" or the technical skill of making a 500 MHz CPU with pipeline of only 4 stage. If it fast enough and does not run too hot it is good. We have been shafted by dog slow G4s since 1999, it is time to pull out
Originally posted by LiquidR
What's the possibility of Apple planning a consumer box around the new 2Ghz Moto chip?
What is the new 2Ghz Moto chip?!
They pulled the stages out to 7, and made process enhancements along the way. Clearly, since that time, the demands of a high-end desktop chip have moved well beyond what makes a capable laptop chip or even a part for a small AIO enclosure. And there is also the issue of cost, small process G4's will most certainly cost less than PPC970.
When I wrote that, the G4 was extremely limited and stretched way too thin. Apple had to use it to supply high and low end needs, and it was just spread too thin -- hindered in the consumer product because it was not allowed to approach the speeds of the Pro offerings, and inadequate in the pro offerings because it simply couldn't achieve adequate performance for machines of their cost. Back then we were contemplating the power consumption of G4's and even more pitiful G3's, a true fossil in the CPU world. The G5 has moved the mark enough, that if the best of the G4 line (process shrunk, modern FSB, and cheaper cost) were to be used throughout the consumer line-up then it would allow for competitive performance and small AFFORDABLE packages.
Would you really balk at using 1.3-2Ghz G4's with a modern FSB in a 799-1299 range of e/iMac and iBook products? Mebbe then you protest merely out of habit?
I favor MUCH LOWER prices on the consumer machines. Don't believe the cheap PPC970 line for a second. How could that be? The prices went up on the low end, the heat sincs are huge. No, the PPC970 costs more, and IS hot. Why would IBM still be tinkering with an expanded G3 performance envelope if they could simply down-clock the G5 into the appropriate heat and cost parameters? Simple answer, they can't, not yet.
Everything I argued is still true of towers, build it first, find a way to make it fit afterwards, even by nine fans if need be.
One day, a 5GHz Mac will cost $599, but not now. As long as there are computers to help make money, their makers will make money selling them.
The 970 is also supposed to be cheaper than the 7455, not the 7457. If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
I don't think Moto will "fix" the FSB of the G4 any time soon, although it might be improved by moving to 200 MHz. MPX won't go DDR, they'll go to RIO first and (hopefully) bring the memory controller onto the processor. That will improve price, performance, integration, and reduce power consumption further.
Originally posted by Programmer
If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
From all their current docs, it sure seems that Motorola is considering the 7457 for the embedded market. It is so hard to gauge their direction for future Apple chips.
Originally posted by Programmer
The 970 is what was said to be cheaper, not the G5. The support chips (and there are several of them) plus the mobo capable of carrying the 1 GHz FSB, plus the case are what drive the price of the Apple machine up.
True, but the G4 needs expensive L3 cache to boost performance, let alone any speculation on yield issues.
Originally posted by Programmer
The 970 is also supposed to be cheaper than the 7455, not the 7457. If Moto can get their 0.13 fab working (or use their partner's 0.09 fab) then the 7457 will decrease costs, improve performance, and reduce power consumption / heat production. The reduced power consumption is very important in the portable & consumer machines because affects the hardware design significantly for heat reasons. At some point the power consumption itself isn't as big an issue because relative to the display, memory, GPU, hard disk, etc the processor isn't such a big consumer of power... but I don't think the G4 is quite at that point yet. With the 7457 it might be.
But, as you stated, they're still left using @ best a 200MGz FSB, thus aggravating an already excruciatingly, painfully, woefully choked FSB. Their answer, support for even more, up to 4MB/processor, very expensive L3 cache.
Originally posted by Programmer
I don't think Moto will "fix" the FSB of the G4 any time soon, although it might be improved by moving to 200 MHz. MPX won't go DDR, they'll go to RIO first and (hopefully) bring the memory controller onto the processor. That will improve price, performance, integration, and reduce power consumption further.
You hit the nail on the head with the statement,"any time soon". The last information from Motorola had the MPC7457-rm(re: using Rapid I/O) as only "proposed".
If and when this cpu appears it won't be before 2005, where will IBM be in this time frame. I like Motorola, and they have apparantly done what they could for Apple in getting the MPC7455 up to an astonishing 1.42GHz, but I hold no hope for miracles from them. Maybe collaboration w/ Phillips and SMT will speed any development, but the future of Apple's consumer machines looks fairly grim if they stay w/ Motorola cpu's.
Oh, I forgot to say I'm going to be buying an Apple computer with a G5 in it as soon as they hit the stores. Looks to be either the low end tower or mid-range tower at this moment, but hope springs eternal for additional introductions between now and the August time frame.