Why? Wouldn't DDR be atttractive for migration to the iMac when the G5 comes out?
just asking</strong><hr></blockquote>
I threw in a caveat about that at the end:
"(of course, it could also mean that they want to go DDR in the PB and iMac, so it may be irrelevant - isn't crystal ball gazing fun? )."
From what I gathered from Applenut's thread, adding DDR compatibility to the G4 is no small matter. So, if Apple wants to move on to the G5 fairly soon, I don't think they would encourage Moto to add it (since the G5 would have it already). That also assumes Apple would be content with SDR in the iMacs and PBs until they can be moved to G5s - which seems reasonable, since they are getting very good performance from the dual 1-GHz PM with SDR.
On the other hand, coming out with a DDR compatible G4 means that Apple did indeed push Moto to make those major revisions to the G4. This seems senseless to me if they are about to go to a G5 in the PMs.
It seems unlikely that Apple would sweat bullets about getting DDR into an iMac or PB when they can get sufficient performance with SDR for these systems. The iMac is not meant to be a professional-level butt-kickin' system (but is still very good, of course, and a great value). A DDR G4 and mobo combo may be rather power-hungry in a PB for the first few iterations (I don't know, but if I had to bet it would be on higher power consumption rather than less) - so it may not fly for the PB either, at least early on. So developing a DDR G4 JUST for the iMac/PB/iB doesn't seem likely. They would most likely get it only if it was already done for the PM and they could get it as a spin-off.
Of course, there is always the caveat that I may be totally full of it, but I've tried to reason things out as best I could. All this leads me to the conclusion that a G4/DDR combo in a PM means no G5 for a while. Your mileage, as always, may vary.
A DDR G4 and mobo combo may be rather power-hungry in a PB for the first few iterations (I don't know, but if I had to bet it would be on higher power consumption rather than less)
</strong><hr></blockquote>
Given DDR-RAM commonly uses only 2.5V rather than the 3.3V found on SDR modules, some people argue that moving to DDR might actually reduce energy consumption.
Given DDR-RAM commonly uses only 2.5V rather than the 3.3V found on SDR modules, some people argue that moving to DDR might actually reduce energy consumption.
Bye,
RazzFazz</strong><hr></blockquote>
It's not so much the RAM that would concern me, but the processor itself. I think the first out would be the "quick and dirty" version, so it would be designed to be easy to fab with minimal production line changes. Energy efficiency would probably not be one of the first design criteria. Optimization for power consumption tends to come in the 2nd or 3rd rev, generally (the low-power Apollo chips have yet to show, for example). Again, I have no direct knowledge, just making an educated guess.
On the other hand, if the DDR RAM consumes significantly less power than SDR, perhaps that would ameliorate some of the increased power consumption of a revised processor. On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption. Anybody out there know anything about this? It's got me curious, now.
Ok, so after "some months", there will be new CPUs.
"G4 has lots of life left in it"
Obviously. If the G4 superscedes the G3 as Apple's consumer CPU, then it sure as hell better have lots of life left in it, at least 2-3 years worth. The iMac is going to use a G4 for at least 1-3 years, and if x86 is at 3 GHz in a few years, then the iMac cannot be wallowing at < 1 GHz (not unless Apple plans on using them as paper weights).
I don't understand why 3 pages of discussion on this? The meaning of these statements seems obvious....and it clearly was said to quench the G5 rumors swirling around the internet.
My guess is that the new Powermacs aren't selling well at all, and Apple is desperate to pump up sales, so this announcement was made with the hope that it would spark some Powermac sales. In other words, Apple is in deep sh!t because their pro desktops aren't selling, and there is nothing that can be done about it short of offering a Powermac with competitive performance.
There has been so much information leaked about the G5 recently that it must exist, and in fact it must be in the field testing phase of development. It will not appear for "some months". I believe it will appear between MWNY and MWSF, no earlier, and no later, because any later and Apple would be in over their heads. 1 GHz up against a 2.5-3.0 GHz Pentium spells doom for Apple, and I'm sure that management will not let this happen even if they must throw insane amounts of money at G5 R&D.
Current hardware speaks volumes more about future Macs than any half-baked announcement does. Consumer Macs = fast G4. Powermacs = stopgap measure. Taken together, the current iMacs and Powermacs suggest that the Pro will migrate to the G5 architecture.
There will not be quad CPU Powermacs, or sexta or octa powermacs. No dodecahedron Powermacs with 10 CPUs, no 16 CPU Powermacs running at 50 teraflops. But there will be G5 powermacs, probably in at least one dual configuration, and they will rock.
Now I'm gonna go beat off so I can forget about this thread. Urghgh.
On the other hand, if the DDR RAM consumes significantly less power than SDR, perhaps that would ameliorate some of the increased power consumption of a revised processor. On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption. Anybody out there know anything about this? It's got me curious, now.</strong><hr></blockquote>
DDR-RAM works by moving data at the top and bottom of its cycle, as opposed to just one side of the cycle in SDR-RAM. This mean that 266MHz DDR-RAM is a 133MHz clock sending two data pulses / clock as opposed to one pulse / clock in SDR-RAM. So, if those voltages are comparing PC133 vs. PC266DDR, then the DDR would use less energy.
On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption.</strong><hr></blockquote>
Well, the point is, contrary to popular belief, DDR266 RAM runs only at 133MHz, exactly like PC133.
... But the questions remain. What the heck can Apple do? I don't know, we'll see what Mr. Jobs does. But it is getting pretty ludicrous that by the time the G5 or whatever the next generation chip Apple uses from Motorola/IBM, Intel and AMD will be pushing, if not past 3GHz. I'm also sure that Intel/AMD aren't standing still on developing faster tech like RapidI/O, Hypertransport, whatever.
If any one has the answer please send it to Mr. Jobs via email, regular mail, Fed X, UPS or pony express, soon.<hr></blockquote>
The answer has been here for years. ~ MULTIPLE PROCESSORS ~ In Architosh's survey of Apple professional users more than half (55%) state that what they most want is QUAD G4's, not G5's. AND they felt that $4999 would be a price point they would be willing to pay.
This is so obvious a next step (after all Apple has been there before) that it has to have heavy consideration in Apple R&D labs. AND it would be within the scope of Shiller's statement of no new CPU's for "some months" if they announced QUAD's at Seybold, MWTY, or tommorow.
That most certainly answers the question for high end users, at least those that use software designed for multiprocessor set ups.
What about the vast majority of users? Will software developers begin recoding all software intended for use on Apple computers to be multiprocessor(or should I say multithreaded here?) optimized?
This solution wouldn't bother me in the least, if the software were available.
The answer has been here for years. ~ MULTIPLE PROCESSORS ~ In Architosh's survey of Apple professional users more than half (55%) state that what they most want is QUAD G4's, not G5's. AND they felt that $4999 would be a price point they would be willing to pay.
This is so obvious a next step (after all Apple has been there before) that it has to have heavy consideration in Apple R&D labs. AND it would be within the scope of Shiller's statement of no new CPU's for "some months" if they announced QUAD's at Seybold, MWTY, or tommorow.</strong><hr></blockquote>
Heh, basing the next Apple machine on the results of a web poll is just silliness. Most users don't know what the heck they want, except that it should be fast. They think 4 must be better than 1 so of course they vote for it.
Until the memory bandwidth issues are addressed (i.e. at least DDR266) then the benefit of a quad processor is very dubious. Even when DDR shows up we're more likely to see a dual 1.5 GHz machine than a quad processor -- and that machine will be faster than the hypothetical quad.
"Until the memory bandwidth issues are addressed (i.e. at least DDR266) then the benefit of a quad processor is very dubious. Even when DDR shows up we're more likely to see a dual 1.5 GHz machine than a quad processor -- and that machine will be faster than the hypothetical quad."<hr></blockquote>
Then I'm back to where I started."But the questions remain. What the heck can Apple do?"
Let's see, Atltivec is great, no question.
Only Motorola seems interested in Altivec.
Motorola seems to give not even one whit about the desktop market.
No way Apple buys Motorola's SPS sector, no way.
No way Apple goes to the dark side, x86.
IBM could probably use Altivec, part of AIM and all that.
But IBM is actually behind Motorola in MHz. and also seems immune to the allure of the desktop market.
That leaves what, Transmeta and AMD? Who else? And how they could get the rights to Altivec? AMD buy Motorola's SPS, very doubtful.
I used to have a fantasy that Apple would buyout the Alpha project or at least its engineering team. No go obviously.
Such a route adds some definite ugliness to Apple's roadmap. Transitioning to OSX is plenty enough for quite a while. Big developers already seem annoyed, can you imagine if they had to deal with a completely different chip architecture as well?
DDR-RAM works by moving data at the top and bottom of its cycle, as opposed to just one side of the cycle in SDR-RAM. This mean that 266MHz DDR-RAM is a 133MHz clock sending two data pulses / clock as opposed to one pulse / clock in SDR-RAM. So, if those voltages are comparing PC133 vs. PC266DDR, then the DDR would use less energy.</strong><hr></blockquote>
Ah, but what is that actually consumes energy in RAM? I am under the impression that most of it comes from flipping the transistor gates on and off, with most of the rest coming from resistance in the circuits. So if the gates are getting flipped twice as often (once on the uptick and again on the downtick of the clock signal) and signals are moving through the circuits twice as often, doesn't that translate into net higher power consumption? I can't imagine that the second "push" per clock cycle is a freebie, energy-wise.
Comments
"If we get one[motherboard w/ DDR] with the G4, I think it will be awhile before we see the G5"<hr></blockquote>
Why? Wouldn't DDR be atttractive for migration to the iMac when the G5 comes out?
just asking
<strong>
Why? Wouldn't DDR be atttractive for migration to the iMac when the G5 comes out?
just asking</strong><hr></blockquote>
I threw in a caveat about that at the end:
"(of course, it could also mean that they want to go DDR in the PB and iMac, so it may be irrelevant - isn't crystal ball gazing fun? )."
From what I gathered from Applenut's thread, adding DDR compatibility to the G4 is no small matter. So, if Apple wants to move on to the G5 fairly soon, I don't think they would encourage Moto to add it (since the G5 would have it already). That also assumes Apple would be content with SDR in the iMacs and PBs until they can be moved to G5s - which seems reasonable, since they are getting very good performance from the dual 1-GHz PM with SDR.
On the other hand, coming out with a DDR compatible G4 means that Apple did indeed push Moto to make those major revisions to the G4. This seems senseless to me if they are about to go to a G5 in the PMs.
It seems unlikely that Apple would sweat bullets about getting DDR into an iMac or PB when they can get sufficient performance with SDR for these systems. The iMac is not meant to be a professional-level butt-kickin' system (but is still very good, of course, and a great value). A DDR G4 and mobo combo may be rather power-hungry in a PB for the first few iterations (I don't know, but if I had to bet it would be on higher power consumption rather than less) - so it may not fly for the PB either, at least early on. So developing a DDR G4 JUST for the iMac/PB/iB doesn't seem likely. They would most likely get it only if it was already done for the PM and they could get it as a spin-off.
Of course, there is always the caveat that I may be totally full of it, but I've tried to reason things out as best I could. All this leads me to the conclusion that a G4/DDR combo in a PM means no G5 for a while. Your mileage, as always, may vary.
[ 02-13-2002: Message edited by: TJM ]</p>
<strong>
A DDR G4 and mobo combo may be rather power-hungry in a PB for the first few iterations (I don't know, but if I had to bet it would be on higher power consumption rather than less)
</strong><hr></blockquote>
Given DDR-RAM commonly uses only 2.5V rather than the 3.3V found on SDR modules, some people argue that moving to DDR might actually reduce energy consumption.
Bye,
RazzFazz
<strong>
Given DDR-RAM commonly uses only 2.5V rather than the 3.3V found on SDR modules, some people argue that moving to DDR might actually reduce energy consumption.
Bye,
RazzFazz</strong><hr></blockquote>
It's not so much the RAM that would concern me, but the processor itself. I think the first out would be the "quick and dirty" version, so it would be designed to be easy to fab with minimal production line changes. Energy efficiency would probably not be one of the first design criteria. Optimization for power consumption tends to come in the 2nd or 3rd rev, generally (the low-power Apollo chips have yet to show, for example). Again, I have no direct knowledge, just making an educated guess.
On the other hand, if the DDR RAM consumes significantly less power than SDR, perhaps that would ameliorate some of the increased power consumption of a revised processor. On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption. Anybody out there know anything about this? It's got me curious, now.
Ok, so after "some months", there will be new CPUs.
"G4 has lots of life left in it"
Obviously. If the G4 superscedes the G3 as Apple's consumer CPU, then it sure as hell better have lots of life left in it, at least 2-3 years worth. The iMac is going to use a G4 for at least 1-3 years, and if x86 is at 3 GHz in a few years, then the iMac cannot be wallowing at < 1 GHz (not unless Apple plans on using them as paper weights).
I don't understand why 3 pages of discussion on this? The meaning of these statements seems obvious....and it clearly was said to quench the G5 rumors swirling around the internet.
My guess is that the new Powermacs aren't selling well at all, and Apple is desperate to pump up sales, so this announcement was made with the hope that it would spark some Powermac sales. In other words, Apple is in deep sh!t because their pro desktops aren't selling, and there is nothing that can be done about it short of offering a Powermac with competitive performance.
There has been so much information leaked about the G5 recently that it must exist, and in fact it must be in the field testing phase of development. It will not appear for "some months". I believe it will appear between MWNY and MWSF, no earlier, and no later, because any later and Apple would be in over their heads. 1 GHz up against a 2.5-3.0 GHz Pentium spells doom for Apple, and I'm sure that management will not let this happen even if they must throw insane amounts of money at G5 R&D.
Current hardware speaks volumes more about future Macs than any half-baked announcement does. Consumer Macs = fast G4. Powermacs = stopgap measure. Taken together, the current iMacs and Powermacs suggest that the Pro will migrate to the G5 architecture.
There will not be quad CPU Powermacs, or sexta or octa powermacs. No dodecahedron Powermacs with 10 CPUs, no 16 CPU Powermacs running at 50 teraflops. But there will be G5 powermacs, probably in at least one dual configuration, and they will rock.
Now I'm gonna go beat off so I can forget about this thread. Urghgh.
<strong>
[SNIP]
On the other hand, if the DDR RAM consumes significantly less power than SDR, perhaps that would ameliorate some of the increased power consumption of a revised processor. On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption. Anybody out there know anything about this? It's got me curious, now.</strong><hr></blockquote>
DDR-RAM works by moving data at the top and bottom of its cycle, as opposed to just one side of the cycle in SDR-RAM. This mean that 266MHz DDR-RAM is a 133MHz clock sending two data pulses / clock as opposed to one pulse / clock in SDR-RAM. So, if those voltages are comparing PC133 vs. PC266DDR, then the DDR would use less energy.
<strong>I'm guessing nothing new until May. Or at least until they remove the "NEW" logos from the Apple Store.</strong><hr></blockquote>
"New" has been removed from the PBooks on the Apple Store!
<strong>It sounds a bit like a '"channel-clearing statement"
Anyway, I know to the hour when the TiBook is getting a CPU/GPU bump.....
Precisely one hour after I finally buy the old one.</strong><hr></blockquote>
Yup... CONFIRMED!! G5 Release Date!! One week after I finally stop waiting and buy the G4
<strong>
On the other other hand, DDR may consume less power per clock cycle due to its lower voltage, but being run at twice the speed may result in a net increase in power consumption.</strong><hr></blockquote>
Well, the point is, contrary to popular belief, DDR266 RAM runs only at 133MHz, exactly like PC133.
Bye,
RazzFazz
... But the questions remain. What the heck can Apple do? I don't know, we'll see what Mr. Jobs does. But it is getting pretty ludicrous that by the time the G5 or whatever the next generation chip Apple uses from Motorola/IBM, Intel and AMD will be pushing, if not past 3GHz. I'm also sure that Intel/AMD aren't standing still on developing faster tech like RapidI/O, Hypertransport, whatever.
If any one has the answer please send it to Mr. Jobs via email, regular mail, Fed X, UPS or pony express, soon.<hr></blockquote>
The answer has been here for years. ~ MULTIPLE PROCESSORS ~ In Architosh's survey of Apple professional users more than half (55%) state that what they most want is QUAD G4's, not G5's. AND they felt that $4999 would be a price point they would be willing to pay.
This is so obvious a next step (after all Apple has been there before) that it has to have heavy consideration in Apple R&D labs. AND it would be within the scope of Shiller's statement of no new CPU's for "some months" if they announced QUAD's at Seybold, MWTY, or tommorow.
"~ MULTIPLE PROCESSORS ~"<hr></blockquote>
That most certainly answers the question for high end users, at least those that use software designed for multiprocessor set ups.
What about the vast majority of users? Will software developers begin recoding all software intended for use on Apple computers to be multiprocessor(or should I say multithreaded here?) optimized?
This solution wouldn't bother me in the least, if the software were available.
<strong>
The answer has been here for years. ~ MULTIPLE PROCESSORS ~ In Architosh's survey of Apple professional users more than half (55%) state that what they most want is QUAD G4's, not G5's. AND they felt that $4999 would be a price point they would be willing to pay.
This is so obvious a next step (after all Apple has been there before) that it has to have heavy consideration in Apple R&D labs. AND it would be within the scope of Shiller's statement of no new CPU's for "some months" if they announced QUAD's at Seybold, MWTY, or tommorow.</strong><hr></blockquote>
Heh, basing the next Apple machine on the results of a web poll is just silliness. Most users don't know what the heck they want, except that it should be fast. They think 4 must be better than 1 so of course they vote for it.
Until the memory bandwidth issues are addressed (i.e. at least DDR266) then the benefit of a quad processor is very dubious. Even when DDR shows up we're more likely to see a dual 1.5 GHz machine than a quad processor -- and that machine will be faster than the hypothetical quad.
"Until the memory bandwidth issues are addressed (i.e. at least DDR266) then the benefit of a quad processor is very dubious. Even when DDR shows up we're more likely to see a dual 1.5 GHz machine than a quad processor -- and that machine will be faster than the hypothetical quad."<hr></blockquote>
Then I'm back to where I started."But the questions remain. What the heck can Apple do?"
Let's see, Atltivec is great, no question.
Only Motorola seems interested in Altivec.
Motorola seems to give not even one whit about the desktop market.
No way Apple buys Motorola's SPS sector, no way.
No way Apple goes to the dark side, x86.
IBM could probably use Altivec, part of AIM and all that.
But IBM is actually behind Motorola in MHz. and also seems immune to the allure of the desktop market.
That leaves what, Transmeta and AMD? Who else? And how they could get the rights to Altivec? AMD buy Motorola's SPS, very doubtful.
Such a route adds some definite ugliness to Apple's roadmap. Transitioning to OSX is plenty enough for quite a while. Big developers already seem annoyed, can you imagine if they had to deal with a completely different chip architecture as well?
[ 02-14-2002: Message edited by: MemeTransport ]</p>
<strong>
DDR-RAM works by moving data at the top and bottom of its cycle, as opposed to just one side of the cycle in SDR-RAM. This mean that 266MHz DDR-RAM is a 133MHz clock sending two data pulses / clock as opposed to one pulse / clock in SDR-RAM. So, if those voltages are comparing PC133 vs. PC266DDR, then the DDR would use less energy.</strong><hr></blockquote>
Ah, but what is that actually consumes energy in RAM? I am under the impression that most of it comes from flipping the transistor gates on and off, with most of the rest coming from resistance in the circuits. So if the gates are getting flipped twice as often (once on the uptick and again on the downtick of the clock signal) and signals are moving through the circuits twice as often, doesn't that translate into net higher power consumption? I can't imagine that the second "push" per clock cycle is a freebie, energy-wise.