do you think IBM or Freescale and Apple have been working on a new chip for the laptops?
Of course they have been and they are working on that. Do you see anything for the next Powerbook update? The Powerbooks use now the highest clocked G4 and there is nothing better right now that could go into the Powerbook for the next update (late this year or early next year). Nothing visible to us. This means there is certainly something in the works. The question is who is going to produce it, IBM or Freescale and what it is exactly.
Quote:
I mean, if its going to take a while to develop, I hope they dont count on the G4 to last a few more years in the powerbooks in its current state.
If a new chip comes from Freescale, it will be a G4. Hopefully, not the G4 we know today. See for example this one. It is a short description of the upcoming e600/700 architectures.
Apple has publically said there will not be a G5 PowerBook for sometime. If one reasonably precludes the possibility of a split PowerBook lineup (all Powerbooks not using the same processor core), one can logically conclude by Apple's statements that future G5s are reserved for the PowerBook, thus negating the possibility of an e600 PowerBook.
The iBook's and the eMac's futures lie with the e600.
... one can logically conclude by Apple's statements that future G5s are reserved for the PowerBook, thus negating the possibility of an e600 PowerBook.
I believe that they would call G5 in the Powerbook, anything drastically different from the current G4, even if it is not a 64-bit chip. And let us not forget here that the e700 has 64-bit support, like a G5 from IBM.
I believe that they would call G5 in the Powerbook, anything drastically different from the current G4, even if it is not a 64-bit chip. And let us not forget here that the e700 has 64-bit support, like a G5 from IBM.
I doubt that. Being 64 bit is one of the things that defines the G5. If they use the e600 in a powerbook, maybe they'll break from the Gx naming scheme. After all, there's nothing really sacrosanct about it.
Aside from the G5, you mean? That is a pretty significant achievement in anybody's book. The vapourware isn't really vapour either, it was rumours for the most part and its hardly fair to hold them to rumours.
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
IBM releases technology press releases all the time, that doesn't mean they are intended to make a product out of it. Often it is just to gauge response from potential OEMs. Quite likely nobody even cared about a 2 GHz G3 so they just dropped it. They also never said anything about a G3 with AltiVec, although on one roadmap they had "SIMD extensions" listed next to the last generation G3 -- again, this is hardly a product announcement and the rumours from March said that Apple had told IBM they didn't want it so IBM dropped it. Why should they build something that nobody is going to use? That would be dumb.
Now IBM is clearly invested in their Power architecture, which is primarily focused on POWER4-derived technologies. I'm dubious about when that will reach notebooks and expect that Freescale will have that end of the solution. There might be a SoC 9xx eventually though, I suppose.
I think what has caught everyone by surprise is that 90nm is not only really hard, but its payoff is nowhere near what previous transitions have brought in terms of benefits. 90nm parts are smaller, and therefore cheaper... if you can get the yields up to 130nm levels. From a power consumption and clock rate perspective, however, it isn't significantly better than 130nm and it increases the heat density problem to epic proportions.
The ramifications of this for Freescale will be interesting. If their partnership at Crolles works out well, and they can actually deliver a 90nm G4-class part (say with on-chip memory controller), what will its heat / clock characteristics be like? According to IBM the fab & design teams will need to work much more closely together at 90nm and below, which may not bode well for Freescale using an external fab. I'd place my bet with IBM being able to make better progress than the as-yet unproven Freescale/Crolles combo.
Motorola's design strategy, and their Quixotic quest for 130nm, might actually help them here. I remember when the 7457 came out everyone was baffled by how huge it was. The contemporary G3 packed nearly as many transistors into a much smaller space, even after you normalized for process. And sure enough, the die photo of the 7457 showed that it was laid out like a Midwestern town. Why?
Well, I'll bet it shrinks down well. The design is absolutely paranoid about preventing "hot spots" and so there's all kinds of space. It's an approach that nobody else is taking, as far as I know — I've even seen it held that it's nothing more than the result of junior engineers out of their depth — but it just might work.
Certainly, by now, Mot/Freescale is quite used to dealing with difficult processes and low yields.
As far as the gap between process & design teams go, Crolles will be using Motorola's process, built around Black Diamond with all the trimmings, so that should help. I can only imagine where IBM would be if they hadn't had the design and process engineers for the 970 series all ensconced in Fishkill.
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
power 4 power 4+ power 5 (basis for all 97x designs)
Apple has publically said there will not be a G5 PowerBook for sometime. If one reasonably precludes the possibility of a split PowerBook lineup (all Powerbooks not using the same processor core), one can logically conclude by Apple's statements that future G5s are reserved for the PowerBook, thus negating the possibility of an e600 PowerBook.
so if there is going to be no G5 in the PB for awhile, and the e600 wont be used, then for the next year or so they are going to keep using the current g4s?
And sure enough, the die photo of the 7457 showed that it was laid out like a Midwestern town. Why?
Well, I'll bet it shrinks down well. The design is absolutely paranoid about preventing "hot spots" and so there's all kinds of space. It's an approach that nobody else is taking, as far as I know — I've even seen it held that it's nothing more than the result of junior engineers out of their depth — but it just might work.
None at all. As I've said before, it may be feasible to put a 1.6GHZ 90nm G5 in a PowerBook (though the system controller likely makes it a bit stickier), but are their any real benefits to it, and will it be faster than the 1.5GHZ G4? I think that after the FSB is reduced greatly in speed the G5 begins to lose some of its advantage because bandwidth is reduced.
Actually, I think Motorola chips do quite well when you consider their bus speed and lack of true DDR support. With some architectural improvements, better cache access (and increased cache), and a faster FSB (up to 400MHZ?) with true DDR implementation - I think Moto would have a big time winner here. Heck, I think they could build a chip that would rival Centrino - lower clock rate, but plenty fast because of intelligent design. Add dual core to that and the PowerBooks would be in great shape.
To move on to other posts, the biggest question with Apple's naming strategy is whether or not they will use the e600 in desktops. It is possible that Apple could put the G5 in all of its desktops (which could make better use of it because they would allow for the high bus speeds, etc) and the e600 in all laptops. If this is the case, maybe it would be best to call the e600 that "M5" and begin to clearly delineate the mobile chips from the desktop chips. The biggest problem is that it shouldn't be called a G5 and you can't really call it a G6 - and G4 doesn't resonate well if everyone in the Mac community. Besides, I don't think the e700 would really be a G4 and you can't call it a G6. Could be an interesting problem, or Apple might just stick with G4 for a long time.
so if there is going to be no G5 in the PB for awhile, and the e600 wont be used, then for the next year or so they are going to keep using the current g4s?
We first have to admit as fact that the G5 won't go to the Powerbook for quite some time yet. Now, if Freescale delivers the e600 chip in time (say this autumn), then Apple will use it for sure. Apple has no other choice and the e600 is going to be very nice (at least the descriptions leave to think so).
What happens if Freescale fails? Now, this would be interesting. It would leave perhaps the Powerbook line for a whole year or more without CPU updates, unless they are able to push a little more the clock speed of the old (but still in use) G4. But I am not sure if it makes sense to do that, since already the clock and the bus frequencies are at 9:1 ratio. And if it makes sense, on the other side there is already the Dothan chip, running at 2 GHz and sporting architectural improvements that would challenge even a G5. Hehe, fail is not an option now for Freescale.
What gives me faith and hope, is the move of the iBook line to the G4. The gap between the two portable lines is very narrow now. I don't think Apple would do this if there was not something really good coming for the Powerbook. And we know this is not a G5.
We first have to admit as fact that the G5 won't go to the Powerbook for quite some time yet.
yeah good point, i know i want more powerful pb's and would enjoy a pb that was g5 like in power, but if stuffing a g5 into a pb means 2 hours of battery life, 2 inch thick case, loud fans and hot metal then i will pass. just hoping something more pentium m like comes out that is geared for laptops. thanks
e600 is definitely a G4, and e700 could simply be called a G5, due to its 64-bitness.
Apple should have called the G5 a G6 - we waited long enough for two processor revs, and the name is just marketing anyway. If they did that, they'd have room to grow a "consumer" line of processors that could be 32-bit but shed the stigma that the G4 is getting.
With that in mind, why not call the consumer chips "i5" and the pro chips "G5"? That way, whether it's freescale or IBM who makes the consumer chips, the "G5" brand is protected, but everyone can upgrade to a "5"
i5 -> 32 bit but >1.5 GHZ, optimized for consumer requirements like low power and small form factor (think Centrino)
G5 -> high power reqs but 64-bit, optimized for the sort of things I don't do on my home computer.
Then they'd just need a G-something for the powerbooks. Some kind of stopgap solution.
The added benefit of this is that right now, "Power" means "dual" and there's no hope of the PowerBook even getting one G5, much less 2. If the "Power" and "i" lines used different processors, this problem goes away...
Actually, I think Motorola chips do quite well when you consider their bus speed and lack of true DDR support. With some architectural improvements, better cache access (and increased cache), and a faster FSB (up to 400MHZ?) with true DDR implementation - I think Moto would have a big time winner here. Heck, I think they could build a chip that would rival Centrino - lower clock rate, but plenty fast because of intelligent design. Add dual core to that and the PowerBooks would be in great shape.
The next significant G4 revision will most likely do away with the FSB completely, interface directly to RAM and have an I/O bus (RIO or HT). If Apple is really lucky some of the I/O devices will move directly onto the processor as well. That will be a massive boost in performance for the G4, even without a clock speed bump. At 90nm it ought to stack up reasonably well against Centrino.
The next significant G4 revision will most likely do away with the FSB completely, interface directly to RAM and have an I/O bus (RIO or HT). If Apple is really lucky some of the I/O devices will move directly onto the processor as well. That will be a massive boost in performance for the G4, even without a clock speed bump. At 90nm it ought to stack up reasonably well against Centrino.
I had forgotten that SoC would allow for this (I'm not a chip guru, by a long shot). I think Moto tends to lean towards Rapid I/O. Not only would there be a major speed boost, but there'd also be a reduction in overall system power consumption. RAM would still be a bottleneck, but only because it can't be utilized at the processor's speed. However, the removal of the FSB would remove a hughe bottleneck. Coupled with improved caching ability and a 1MB L2 cache, and this would be a very quick, efficient little chip. I'm rooting hard for FreeScale, because if they deliver this in the relatively near future, it's great news for Apple and MacHeads in general.
Comments
Originally posted by Peter North
do you think IBM or Freescale and Apple have been working on a new chip for the laptops?
Of course they have been and they are working on that. Do you see anything for the next Powerbook update? The Powerbooks use now the highest clocked G4 and there is nothing better right now that could go into the Powerbook for the next update (late this year or early next year). Nothing visible to us. This means there is certainly something in the works. The question is who is going to produce it, IBM or Freescale and what it is exactly.
I mean, if its going to take a while to develop, I hope they dont count on the G4 to last a few more years in the powerbooks in its current state.
If a new chip comes from Freescale, it will be a G4. Hopefully, not the G4 we know today. See for example this one. It is a short description of the upcoming e600/700 architectures.
The iBook's and the eMac's futures lie with the e600.
Originally posted by Existence
... one can logically conclude by Apple's statements that future G5s are reserved for the PowerBook, thus negating the possibility of an e600 PowerBook.
I believe that they would call G5 in the Powerbook, anything drastically different from the current G4, even if it is not a 64-bit chip. And let us not forget here that the e700 has 64-bit support, like a G5 from IBM.
Originally posted by PB
I believe that they would call G5 in the Powerbook, anything drastically different from the current G4, even if it is not a 64-bit chip. And let us not forget here that the e700 has 64-bit support, like a G5 from IBM.
I doubt that. Being 64 bit is one of the things that defines the G5. If they use the e600 in a powerbook, maybe they'll break from the Gx naming scheme. After all, there's nothing really sacrosanct about it.
Originally posted by Programmer
Aside from the G5, you mean? That is a pretty significant achievement in anybody's book. The vapourware isn't really vapour either, it was rumours for the most part and its hardly fair to hold them to rumours.
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
Originally posted by Matsu
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
IBM releases technology press releases all the time, that doesn't mean they are intended to make a product out of it. Often it is just to gauge response from potential OEMs. Quite likely nobody even cared about a 2 GHz G3 so they just dropped it. They also never said anything about a G3 with AltiVec, although on one roadmap they had "SIMD extensions" listed next to the last generation G3 -- again, this is hardly a product announcement and the rumours from March said that Apple had told IBM they didn't want it so IBM dropped it. Why should they build something that nobody is going to use? That would be dumb.
Now IBM is clearly invested in their Power architecture, which is primarily focused on POWER4-derived technologies. I'm dubious about when that will reach notebooks and expect that Freescale will have that end of the solution. There might be a SoC 9xx eventually though, I suppose.
Originally posted by Programmer
I think what has caught everyone by surprise is that 90nm is not only really hard, but its payoff is nowhere near what previous transitions have brought in terms of benefits. 90nm parts are smaller, and therefore cheaper... if you can get the yields up to 130nm levels. From a power consumption and clock rate perspective, however, it isn't significantly better than 130nm and it increases the heat density problem to epic proportions.
The ramifications of this for Freescale will be interesting. If their partnership at Crolles works out well, and they can actually deliver a 90nm G4-class part (say with on-chip memory controller), what will its heat / clock characteristics be like? According to IBM the fab & design teams will need to work much more closely together at 90nm and below, which may not bode well for Freescale using an external fab. I'd place my bet with IBM being able to make better progress than the as-yet unproven Freescale/Crolles combo.
Motorola's design strategy, and their Quixotic quest for 130nm, might actually help them here. I remember when the 7457 came out everyone was baffled by how huge it was. The contemporary G3 packed nearly as many transistors into a much smaller space, even after you normalized for process. And sure enough, the die photo of the 7457 showed that it was laid out like a Midwestern town. Why?
Well, I'll bet it shrinks down well. The design is absolutely paranoid about preventing "hot spots" and so there's all kinds of space. It's an approach that nobody else is taking, as far as I know — I've even seen it held that it's nothing more than the result of junior engineers out of their depth — but it just might work.
Certainly, by now, Mot/Freescale is quite used to dealing with difficult processes and low yields.
As far as the gap between process & design teams go, Crolles will be using Motorola's process, built around Black Diamond with all the trimmings, so that should help. I can only imagine where IBM would be if they hadn't had the design and process engineers for the 970 series all ensconced in Fishkill.
Originally posted by Matsu
Yes, except the G5, my bad. As for the G3 class vapourware, I have to disagree. IBM had press releases saying they'd tested PPC's up to 2Ghz, and spoke of the last G3 as having altivec like speed enhancements. Perhaps not strictly vapourware, but the clear implication was that they could build a faster PPC than moto; they never did, of course, untill now.
power 4 power 4+ power 5 (basis for all 97x designs)
Originally posted by windowsblowsass
poer 4 power 4+ ppoer 5
Either your keyboard has issues or you are trying to say something
Apple has publically said there will not be a G5 PowerBook for sometime. If one reasonably precludes the possibility of a split PowerBook lineup (all Powerbooks not using the same processor core), one can logically conclude by Apple's statements that future G5s are reserved for the PowerBook, thus negating the possibility of an e600 PowerBook.
so if there is going to be no G5 in the PB for awhile, and the e600 wont be used, then for the next year or so they are going to keep using the current g4s?
Originally posted by Amorph
And sure enough, the die photo of the 7457 showed that it was laid out like a Midwestern town. Why?
Well, I'll bet it shrinks down well. The design is absolutely paranoid about preventing "hot spots" and so there's all kinds of space. It's an approach that nobody else is taking, as far as I know — I've even seen it held that it's nothing more than the result of junior engineers out of their depth — but it just might work.
PPC7447 - 21.3W at 1.33GHz
PPC7447A - 20.0W at 1.42GHz
PPC 970 - 65W at 2.0GHz
PPC 970FX - 50W at 2.5GHz
Any doubt why Motorola chips are in laptops?
Originally posted by Rhumgod
PPC7447 - 21.3W at 1.33GHz
PPC7447A - 20.0W at 1.42GHz
PPC 970 - 65W at 2.0GHz
PPC 970FX - 50W at 2.5GHz
Any doubt why Motorola chips are in laptops?
None at all. As I've said before, it may be feasible to put a 1.6GHZ 90nm G5 in a PowerBook (though the system controller likely makes it a bit stickier), but are their any real benefits to it, and will it be faster than the 1.5GHZ G4? I think that after the FSB is reduced greatly in speed the G5 begins to lose some of its advantage because bandwidth is reduced.
Actually, I think Motorola chips do quite well when you consider their bus speed and lack of true DDR support. With some architectural improvements, better cache access (and increased cache), and a faster FSB (up to 400MHZ?) with true DDR implementation - I think Moto would have a big time winner here. Heck, I think they could build a chip that would rival Centrino - lower clock rate, but plenty fast because of intelligent design. Add dual core to that and the PowerBooks would be in great shape.
To move on to other posts, the biggest question with Apple's naming strategy is whether or not they will use the e600 in desktops. It is possible that Apple could put the G5 in all of its desktops (which could make better use of it because they would allow for the high bus speeds, etc) and the e600 in all laptops. If this is the case, maybe it would be best to call the e600 that "M5" and begin to clearly delineate the mobile chips from the desktop chips. The biggest problem is that it shouldn't be called a G5 and you can't really call it a G6 - and G4 doesn't resonate well if everyone in the Mac community. Besides, I don't think the e700 would really be a G4 and you can't call it a G6. Could be an interesting problem, or Apple might just stick with G4 for a long time.
Originally posted by Peter North
so if there is going to be no G5 in the PB for awhile, and the e600 wont be used, then for the next year or so they are going to keep using the current g4s?
We first have to admit as fact that the G5 won't go to the Powerbook for quite some time yet. Now, if Freescale delivers the e600 chip in time (say this autumn), then Apple will use it for sure. Apple has no other choice and the e600 is going to be very nice (at least the descriptions leave to think so).
What happens if Freescale fails? Now, this would be interesting. It would leave perhaps the Powerbook line for a whole year or more without CPU updates, unless they are able to push a little more the clock speed of the old (but still in use) G4. But I am not sure if it makes sense to do that, since already the clock and the bus frequencies are at 9:1 ratio. And if it makes sense, on the other side there is already the Dothan chip, running at 2 GHz and sporting architectural improvements that would challenge even a G5. Hehe, fail is not an option now for Freescale.
What gives me faith and hope, is the move of the iBook line to the G4. The gap between the two portable lines is very narrow now. I don't think Apple would do this if there was not something really good coming for the Powerbook. And we know this is not a G5.
We first have to admit as fact that the G5 won't go to the Powerbook for quite some time yet.
yeah good point, i know i want more powerful pb's and would enjoy a pb that was g5 like in power, but if stuffing a g5 into a pb means 2 hours of battery life, 2 inch thick case, loud fans and hot metal then i will pass. just hoping something more pentium m like comes out that is geared for laptops. thanks
Originally posted by BRussell
e600 is definitely a G4, and e700 could simply be called a G5, due to its 64-bitness.
Apple should have called the G5 a G6 - we waited long enough for two processor revs, and the name is just marketing anyway. If they did that, they'd have room to grow a "consumer" line of processors that could be 32-bit but shed the stigma that the G4 is getting.
With that in mind, why not call the consumer chips "i5" and the pro chips "G5"? That way, whether it's freescale or IBM who makes the consumer chips, the "G5" brand is protected, but everyone can upgrade to a "5"
i5 -> 32 bit but >1.5 GHZ, optimized for consumer requirements like low power and small form factor (think Centrino)
G5 -> high power reqs but 64-bit, optimized for the sort of things I don't do on my home computer.
Then they'd just need a G-something for the powerbooks. Some kind of stopgap solution.
The added benefit of this is that right now, "Power" means "dual" and there's no hope of the PowerBook even getting one G5, much less 2. If the "Power" and "i" lines used different processors, this problem goes away...
Originally posted by Mr. MacPhisto
Actually, I think Motorola chips do quite well when you consider their bus speed and lack of true DDR support. With some architectural improvements, better cache access (and increased cache), and a faster FSB (up to 400MHZ?) with true DDR implementation - I think Moto would have a big time winner here. Heck, I think they could build a chip that would rival Centrino - lower clock rate, but plenty fast because of intelligent design. Add dual core to that and the PowerBooks would be in great shape.
The next significant G4 revision will most likely do away with the FSB completely, interface directly to RAM and have an I/O bus (RIO or HT). If Apple is really lucky some of the I/O devices will move directly onto the processor as well. That will be a massive boost in performance for the G4, even without a clock speed bump. At 90nm it ought to stack up reasonably well against Centrino.
Originally posted by Programmer
The next significant G4 revision will most likely do away with the FSB completely, interface directly to RAM and have an I/O bus (RIO or HT). If Apple is really lucky some of the I/O devices will move directly onto the processor as well. That will be a massive boost in performance for the G4, even without a clock speed bump. At 90nm it ought to stack up reasonably well against Centrino.
I had forgotten that SoC would allow for this (I'm not a chip guru, by a long shot). I think Moto tends to lean towards Rapid I/O. Not only would there be a major speed boost, but there'd also be a reduction in overall system power consumption. RAM would still be a bottleneck, but only because it can't be utilized at the processor's speed. However, the removal of the FSB would remove a hughe bottleneck. Coupled with improved caching ability and a 1MB L2 cache, and this would be a very quick, efficient little chip. I'm rooting hard for FreeScale, because if they deliver this in the relatively near future, it's great news for Apple and MacHeads in general.
RAM would still be a bottleneck, but only because it can't be utilized at the processor's speed.
any solutions or remedies?
Originally posted by Rhumgod
PPC7447 - 21.3W at 1.33GHz
PPC7447A - 20.0W at 1.42GHz
PPC 970 - 65W at 2.0GHz
PPC 970FX - 50W at 2.5GHz
Any doubt why Motorola chips are in laptops?
PPC 970FX @ 1.5 or 1.6GHz == ??? Watts ???