Oh good, another PowerBook thread. Now there are only 14 for me to wade through. Thanks!
This IGM "rumor" is just the Mac Doobie rumor that's already been beaten to death in the other threads.
Mods, help!
Get over it. Think of the intolerably high number of PB threads as being something to do until the new PB's are released.
What would you prefer? One big PB thread with five different discussions in it? I for one would not read such a thread because it has no focus and it would be very difficult to actually discuss one issue in particular (imagine trying to discuss a point while four other points are being discussed!).
The rumor provided is a valid reason for a new PB thread- it is talking about why the PB's are delayed. If you don't like having so many PB threads, then don't read them.
I think the exclusion of the PowerPC 970 in the next Powerbooks because of heat dissipation and power consumption is just a convenient excuse for rumorologists hedging their bets. Most simply don't know and are looking at what is the most reasonable.
IBM estimates a 1.1V 1.2 GHz PPC 970 has 19 Watts. The real numbers won't be more than 10% off that. If they are, someone is doing their math wrong. Really really wrong. Possible, but it's something I find really hard to believe. 19 Watts is a perfectly acceptable power consumption number for notebooks. Along with frequency cycling, which the PowerMac already does, I don't see anything holding it back. The key for a good notebook CPU is a combination of voltage cycling and frequency cycling, so hopefully the 970 will be able to do voltage cycling too (maybe it does, I don't know).
19W + bus power for a 600MHz FSB + DDR power. If DDR is running at 333, then that is nothing that isn't already being done. Otherwise, the bus is going to eat power and dissipate heat. One option would be to run the bus at a 3:1 ratio (it seems that this can be done, see the Ars PPC 970 article)
Something to bear in mind is that a 1.2 GHz 970 would only be faster at bus intensive and FP intensive apps, not integer apps. A 1.3 G4 would actually be faster. I think that once the first die shrink is done and the 970 is smaller, you will see 1.8GHz G5 laptops and that is what apple is waiting for. I'd love to be disproven.
19W + bus power for a 600MHz FSB + DDR power. If DDR is running at 333, then that is nothing that isn't already being done. Otherwise, the bus is going to eat power and dissipate heat. One option would be to run the bus at a 3:1 ratio (it seems that this can be done, see the Ars PPC 970 article)
Is this a legitimate issue? A 1.2 GHz 970, with 2:1 cpu-to-bus ratio, would have a 333 MHz bus that has half the pins of the current parallel bus systems. So it's twice the clock but about half the pins to power, a wash. Maybe the 970 elastic bus requires higher voltages to power, but I don't think so.
Heck if would I heard right about the PowerMac G5's CPU module socket is true, that it only has 300 pins, than I don't think it's that much of an issue.
Quote:
Something to bear in mind is that a 1.2 GHz 970 would only be faster at bus intensive and FP intensive apps, not integer apps. A 1.3 G4 would actually be faster.
I think the 970 will be equivalent to the 7457 at integer codes. Maybe a little bit less, but not that much, and that's assuming some handy tuning for apps on the 7457. The PPC 970 has better out-of-order execution but less integer resources (1 full, but both do mul/add) while the 7457 has poor OOOE but more integer resources (1 full, 3 add). So, I think in typical usages, the differences will wash out.
Quote:
I think that once the first die shrink is done and the 970 is smaller, you will see 1.8GHz G5 laptops and that is what apple is waiting for. I'd love to be disproven.
The power issues for a 1.8 GHz 90 nm PPC 970 will be the same as a 1.2 GHz 130 nm PPC 970. The power consumption would be the same. The capactance of the transistors would reduce around 30 to 50% (not sure of the exact figure) for such process improvements, thereby reducing power, but the clock rate is also increasing by 50% which increases the power consumption.
What we should be looking for in a 970+ is the ability to cycle frequency and cycle voltages at the same time. It can reduce power consumption very large amounts (80+ percent).
What would you prefer? One big PB thread with five different discussions in it? I for one would not read such a thread because it has no focus and it would be very difficult to actually discuss one issue in particular (imagine trying to discuss a point while four other points are being discussed!).
Your "separate threads for separate PB rumors/issues" argument might make sense if we actually had that. We do not. We have five threads with five stupid titles that each end up covering the same issues over and over.
As long as that's the case it might as well be in all one thread. At least that way it would be easier to ignore Matsu's and Clive's posts.
A IBM 970 at 1.2 GHz should have a SPEC200 integer score of 624 a score a Motorola G4 attains at 2.1 GHz
Even if the real differenece would be less due to nonoptimal compilers and that a 970 with a higer bus ratio thant 1:2 say 1:3 or 1:4 might be slowed down it really have to loose a lot to sink down to the performance of a 1.3 GHz G4.
Even if they had to make an iMac where the dome is cast aluminium and the whole thing is used as a heat sink, that is not a big problem. Perhaps the layout of the fast bus is sensitive geomentry but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook
Maybe the best reason for not using it is because Apple simply didn't want to, or didn't have the resources, to design a Powerbook G5 system at the same time as the PowerMac G5. The more I think about it, PPC 970 systems at equivalent clock rates as 7457 systems make a whole more sense from a cost, performance and manufacturing standpoint.
on top of those reasons, maybe existing contracts with Motorola?
Quote:
DrBoar
...but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook.
Touche, bring on the iMac, Powerbook and all new minitower G5's.
A IBM 970 at 1.2 GHz should have a SPEC200 integer score of 624 a score a Motorola G4 attains at 2.1 GHz
Even if the real differenece would be less due to nonoptimal compilers and that a 970 with a higer bus ratio thant 1:2 say 1:3 or 1:4 might be slowed down it really have to loose a lot to sink down to the performance of a 1.3 GHz G4.
Even if they had to make an iMac where the dome is cast aluminium and the whole thing is used as a heat sink, that is not a big problem. Perhaps the layout of the fast bus is sensitive geomentry but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook
My sentiments exactly, I want to see 1.2G G5 PB's now, and leave the 7457's for the iBooks!. For crying out loud even the iMacs are looking anaemic at 1GHz, they could well do with a 1.33-1.42 G4.
Of course, the G5 WILL need a process update in order for the NEXT update, but we dont need that for another 6 months. Unless, IBM cannot get 0.09um 970's in time, then we are back to square one again, jamming forums with posts like these! \
Comments
Originally posted by Ensign Pulver
Oh good, another PowerBook thread. Now there are only 14 for me to wade through. Thanks!
This IGM "rumor" is just the Mac Doobie rumor that's already been beaten to death in the other threads.
Mods, help!
Get over it. Think of the intolerably high number of PB threads as being something to do until the new PB's are released.
What would you prefer? One big PB thread with five different discussions in it? I for one would not read such a thread because it has no focus and it would be very difficult to actually discuss one issue in particular (imagine trying to discuss a point while four other points are being discussed!).
The rumor provided is a valid reason for a new PB thread- it is talking about why the PB's are delayed. If you don't like having so many PB threads, then don't read them.
Originally posted by THT
I think the exclusion of the PowerPC 970 in the next Powerbooks because of heat dissipation and power consumption is just a convenient excuse for rumorologists hedging their bets. Most simply don't know and are looking at what is the most reasonable.
IBM estimates a 1.1V 1.2 GHz PPC 970 has 19 Watts. The real numbers won't be more than 10% off that. If they are, someone is doing their math wrong. Really really wrong. Possible, but it's something I find really hard to believe. 19 Watts is a perfectly acceptable power consumption number for notebooks. Along with frequency cycling, which the PowerMac already does, I don't see anything holding it back. The key for a good notebook CPU is a combination of voltage cycling and frequency cycling, so hopefully the 970 will be able to do voltage cycling too (maybe it does, I don't know).
19W + bus power for a 600MHz FSB + DDR power. If DDR is running at 333, then that is nothing that isn't already being done. Otherwise, the bus is going to eat power and dissipate heat. One option would be to run the bus at a 3:1 ratio (it seems that this can be done, see the Ars PPC 970 article)
Something to bear in mind is that a 1.2 GHz 970 would only be faster at bus intensive and FP intensive apps, not integer apps. A 1.3 G4 would actually be faster. I think that once the first die shrink is done and the 970 is smaller, you will see 1.8GHz G5 laptops and that is what apple is waiting for. I'd love to be disproven.
Originally posted by Yevgeny
19W + bus power for a 600MHz FSB + DDR power. If DDR is running at 333, then that is nothing that isn't already being done. Otherwise, the bus is going to eat power and dissipate heat. One option would be to run the bus at a 3:1 ratio (it seems that this can be done, see the Ars PPC 970 article)
Is this a legitimate issue? A 1.2 GHz 970, with 2:1 cpu-to-bus ratio, would have a 333 MHz bus that has half the pins of the current parallel bus systems. So it's twice the clock but about half the pins to power, a wash. Maybe the 970 elastic bus requires higher voltages to power, but I don't think so.
Heck if would I heard right about the PowerMac G5's CPU module socket is true, that it only has 300 pins, than I don't think it's that much of an issue.
Something to bear in mind is that a 1.2 GHz 970 would only be faster at bus intensive and FP intensive apps, not integer apps. A 1.3 G4 would actually be faster.
I think the 970 will be equivalent to the 7457 at integer codes. Maybe a little bit less, but not that much, and that's assuming some handy tuning for apps on the 7457. The PPC 970 has better out-of-order execution but less integer resources (1 full, but both do mul/add) while the 7457 has poor OOOE but more integer resources (1 full, 3 add). So, I think in typical usages, the differences will wash out.
I think that once the first die shrink is done and the 970 is smaller, you will see 1.8GHz G5 laptops and that is what apple is waiting for. I'd love to be disproven.
The power issues for a 1.8 GHz 90 nm PPC 970 will be the same as a 1.2 GHz 130 nm PPC 970. The power consumption would be the same. The capactance of the transistors would reduce around 30 to 50% (not sure of the exact figure) for such process improvements, thereby reducing power, but the clock rate is also increasing by 50% which increases the power consumption.
What we should be looking for in a 970+ is the ability to cycle frequency and cycle voltages at the same time. It can reduce power consumption very large amounts (80+ percent).
Originally posted by Yevgeny
What would you prefer? One big PB thread with five different discussions in it? I for one would not read such a thread because it has no focus and it would be very difficult to actually discuss one issue in particular (imagine trying to discuss a point while four other points are being discussed!).
Your "separate threads for separate PB rumors/issues" argument might make sense if we actually had that. We do not. We have five threads with five stupid titles that each end up covering the same issues over and over.
As long as that's the case it might as well be in all one thread. At least that way it would be easier to ignore Matsu's and Clive's posts.
A IBM 970 at 1.2 GHz should have a SPEC200 integer score of 624 a score a Motorola G4 attains at 2.1 GHz
Even if the real differenece would be less due to nonoptimal compilers and that a 970 with a higer bus ratio thant 1:2 say 1:3 or 1:4 might be slowed down it really have to loose a lot to sink down to the performance of a 1.3 GHz G4.
Even if they had to make an iMac where the dome is cast aluminium and the whole thing is used as a heat sink, that is not a big problem. Perhaps the layout of the fast bus is sensitive geomentry but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook
THT
Maybe the best reason for not using it is because Apple simply didn't want to, or didn't have the resources, to design a Powerbook G5 system at the same time as the PowerMac G5. The more I think about it, PPC 970 systems at equivalent clock rates as 7457 systems make a whole more sense from a cost, performance and manufacturing standpoint.
on top of those reasons, maybe existing contracts with Motorola?
DrBoar
...but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook.
Touche, bring on the iMac, Powerbook and all new minitower G5's.
Originally posted by DrBoar
G4 better integer than G5?
A IBM 970 at 1.2 GHz should have a SPEC200 integer score of 624 a score a Motorola G4 attains at 2.1 GHz
Even if the real differenece would be less due to nonoptimal compilers and that a 970 with a higer bus ratio thant 1:2 say 1:3 or 1:4 might be slowed down it really have to loose a lot to sink down to the performance of a 1.3 GHz G4.
Even if they had to make an iMac where the dome is cast aluminium and the whole thing is used as a heat sink, that is not a big problem. Perhaps the layout of the fast bus is sensitive geomentry but if Dell can make notebooks with a 2.6 GHz CPU and a 400 MHz bus Apple should be able to cram a 1.2 GHz CPU and a 400 MHz bus in an iBook
My sentiments exactly, I want to see 1.2G G5 PB's now, and leave the 7457's for the iBooks!. For crying out loud even the iMacs are looking anaemic at 1GHz, they could well do with a 1.33-1.42 G4.
Of course, the G5 WILL need a process update in order for the NEXT update, but we dont need that for another 6 months. Unless, IBM cannot get 0.09um 970's in time, then we are back to square one again, jamming forums with posts like these!