PDA

View Full Version : IBM's PowerTune Technology


MacRumors
Feb 16, 2004, 09:15 AM
TechWorld offers (http://www.techworld.com/news/index.cfm?fuseaction=displaynews&NewsID=1035) some early details of IBM's new PowerTune technology in the PowerPC 970FX.

The company claims that the chip's unique power-tuning technique allows it to vary power consumption thousands of times per second, reducing the overall power consumption compared with other chips that consume a constant amount of power.

According to a senior technical staff member at IBM, "customers will be able to use the chip in everything from PCs to notebooks to networking equipment".

The new processor is being used in Apple's Xserve.

CmdrLaForge
Feb 16, 2004, 09:23 AM
That sounds great. PB G5s can come !

tazznb
Feb 16, 2004, 09:25 AM
This will greatly decrease the size factor, if need be. :)

idkew
Feb 16, 2004, 09:25 AM
Just give me my PB G5!

GetSome681
Feb 16, 2004, 09:28 AM
I think people are missing the point of this. Sure, powerbook g5. Think harder. Imagine if you're doing simple stuff (browsing web, word processor) and using like 15-25% cpu...imagine the battery life you're going to end up with!

CmdrLaForge
Feb 16, 2004, 09:31 AM
Originally posted by GetSome681
I think people are missing the point of this. Sure, powerbook g5. Think harder. Imagine if you're doing simple stuff (browsing web, word processor) and using like 15-25% cpu...imagine the battery life you're going to end up with!

Yes - you are right. Even better. Powerbook G5 with 20 hours battery life :D

xtekdiver
Feb 16, 2004, 09:31 AM
I wonder what else Apple has in mind for this chip? How about my new iTv? If M$ puts it in their Xbox what do you think the chances are that someone will hack it and be able to run OS X on it? Oh, that would be rich! A $200 Mac from Microsoft.

a17inchFuture
Feb 16, 2004, 09:33 AM
good point "getsome". But that is just another reason why Apple shoudl get the ball rolling on this thing asap. I hope we can see the processor make its way into powerbooks by the summer. This seems to be a slightly less powerful g5, so just skip the redesigning of the pb's comepltely, and stick the chip in the present models, or at least similar ones, ones that won't require 8 months of designs and prototypes.

virividox
Feb 16, 2004, 09:34 AM
here t comes!!! pb g5

dongmin
Feb 16, 2004, 09:36 AM
Originally posted by Macrumors
According to a senior technical staff member at IBM, "customers will be able to use the chip in everything from PCs to notebooks to networking equipment." My guess is that the CPU issue was resolved a long time ago. It's the other components that are holding Apple back. In any event, this bodes well for PB consumers.

gwuMACaddict
Feb 16, 2004, 09:39 AM
sounds really neat! and yeah, looks like it could make for longer battery life in a PB

Mr. Anderson
Feb 16, 2004, 09:42 AM
This is a great. Does anybody know if its already incorporated into the XServes?

D

a17inchFuture
Feb 16, 2004, 09:46 AM
Does anyone think that there could be powerbook g5's before the fall???

Is there any way they could have been designing the book and just waiting for the chip, or something along those lines? These are less power consuming, and less heat producing, so what would the obstacles be that are still holding the new g5 pb from coming out?

jholzner
Feb 16, 2004, 09:53 AM
Originally posted by Mr. Anderson
This is a great. Does anybody know if its already incorporated into the XServes?

D

Yes, it is.

elmimmo
Feb 16, 2004, 09:55 AM
Hype to the rescue!

silvergunuk
Feb 16, 2004, 09:55 AM
I think well see G5 powerbooks if not next month then possibly wwdc with a totally new designed case.

Lanbrown
Feb 16, 2004, 09:56 AM
Originally posted by a17inchFuture
good point "getsome". But that is just another reason why Apple shoudl get the ball rolling on this thing asap. I hope we can see the processor make its way into powerbooks by the summer. This seems to be a slightly less powerful g5, so just skip the redesigning of the pb's comepltely, and stick the chip in the present models, or at least similar ones, ones that won't require 8 months of designs and prototypes.

The G5 and G4 processor are totally different. They cannot put a G5 in a G4 mobo. A new board needs to be designed. Apple has already had these chips for sometime, how do you think they have been able to show the new Xserve.

Benjamin
Feb 16, 2004, 09:59 AM
how did i know that the first reply would be basically "PB G5!" lol

anyway this is looking to be a great step to that dream of many of the mac community.

Lanbrown
Feb 16, 2004, 10:00 AM
Originally posted by Mr. Anderson
This is a great. Does anybody know if its already incorporated into the XServes?

D

I would say that the Xserve is not using it. It can impact performance and heat is not an issue in servers. A properly designed server will be able to handle the chip running at full power indefinitely. The fans would increase their speed to match the heat output. Some tasks could push the CPU to 100% for a brief amount of time and then that process would be complete. A lot of processes that complete in a short amount of time could hurt performance as the chip is increasing and decreasing it's power.

a17inchFuture
Feb 16, 2004, 10:01 AM
Originally posted by Lanbrown
The G5 and G4 processor are totally different. They cannot put a G5 in a G4 mobo. A new board needs to be designed. Apple has already had these chips for sometime, how do you think they have been able to show the new Xserve.

Yeah I understand reconfiguring the board, but I stil don't understand why if they have "had the chips for somet time" and if they're already in Xserve G5's, Aren't we smart enough consumers to absolutely demand that it be put in PB's soon, and instead of a modest revision to the no-longer-worth-it G4 line.

P.S. I know thats a controversial comment I jsut made about no longer being worth it, but anyone who would look at this news and the probable next few months for apple would be kinda crazy to buy a G4 now or even to get one of the modest updates rumored to be released soon.

Of course, I still respect that the present G4 PB's are incredible.

Lanbrown
Feb 16, 2004, 10:05 AM
Originally posted by jholzner
Yes, it is.

Based upon what? Just because the processor supports it, doesn't mean that Apple is using it. I could not find anything on that mention PowerTune on the new Xserve. Got a link?

Lanbrown
Feb 16, 2004, 10:13 AM
Originally posted by a17inchFuture
Yeah I understand reconfiguring the board, but I stil don't understand why if they have "had the chips for somet time" and if they're already in Xserve G5's, Aren't we smart enough consumers to absolutely demand that it be put in PB's soon, and instead of a modest revision to the no-longer-worth-it G4 line.

P.S. I know thats a controversial comment I jsut made about no longer being worth it, but anyone who would look at this news and the probable next few months for apple would be kinda crazy to buy a G4 now or even to get one of the modest updates rumored to be released soon.

Of course, I still respect that the present G4 PB's are incredible.

First, the bus on the PB will be slower then what the PM and Xserve have. Those high speeds will create heat. The processor was just one component that created a lot of heat, the chipsets created a lot of heat as well. Apple could use the current chipsets in Xserve with the new processor, one less thing to design. Apple either has working samples of a new chipset for the PB and iMac with a slower bus, a third to a quarter of the processor speed.

Consumers make their demands on sales. When sales fall, a company is faced with two options, update or not. If they update, sales will probably rise, if they don't, sales will continue to fall. If itís not feasible to put the G5 in a PB, it doesn't matter what the consumers demands are. As mentioned above, the CPU is just one of many components. I would expect new PM's to have a bus no faster than 1GHz, even if it's a 2.5 or 3GHz CPU.

Third, the chips were samples that they have had. IBM released the chip this month, which means that they are available in volume. As time goes forward, the volume will increase as they refine the process and can get higher yields.

utilizer
Feb 16, 2004, 10:14 AM
I think we're all in for a bigger surprise than the one at last year's WWDC with the "annoucement" of G5s shipping in September.
PowerBook G5s will be annouced for sure this time around; seriously, the iBook was expected to keep the G3 for quite sometime, but look at what happened late in 2003. And when that gets an update, it'll be a 1.42 Ghz G4. The iBook and PB will virtually be the "same" machine. Apple cannot allow for Motorola to continue their stream of broken promises to trickle in. I believe a 1.6 Ghz G5 would suffice alot of users needs, especially with the arrival of Adobe's CS. That thing is sllllooowwww on anything less than that!!!

Anyways though, I'm calling it:
-PowerBooks with G5s at 1.6 Ghz, annouced very soon but shipping in late April; duals by the end of the year. -- 12 in. form factor retains the G4.
-G5 Tower: Dual 1.8 Ghz at low-end and Dual 2.733 Ghz at high end.
-iMac: 1.8 Ghz, maybe 2.0 Ghz in special edition 20 incher (brand it "20 for 20th" (anniversary!)
-iBook: Stuck with the "adequate" G4 7457 supplies from Moto. 'Nuff said!

I underestimated Apple last year -- I think this year, we can anticipate greater surprises!!:D

a17inchFuture
Feb 16, 2004, 10:23 AM
Originally posted by Lanbrown
First, the bus on the PB will be slower then what the PM and Xserve have. Those high speeds will create heat. The processor was just one component that created a lot of heat, the chipsets created a lot of heat as well. Apple could use the current chipsets in Xserve with the new processor, one less thing to design. Apple either has working samples of a new chipset for the PB and iMac with a slower bus, a third to a quarter of the processor speed.

Consumers make their demands on sales. When sales fall, a company is faced with two options, update or not. If they update, sales will probably rise, if they don't, sales will continue to fall. If itís not feasible to put the G5 in a PB, it doesn't matter what the consumers demands are. As mentioned above, the CPU is just one of many components. I would expect new PM's to have a bus no faster than 1GHz, even if it's a 2.5 or 3GHz CPU.

Third, the chips were samples that they have had. IBM released the chip this month, which means that they are available in volume. As time goes forward, the volume will increase as they refine the process and can get higher yields.

great, thats all well and good, but it seems like you were reitterating some of my points rather than making your own. The chips were samples, yes, WORKING samples that serve a purpose, otherwise there would be no point in giving them to anyone early. And they are now avalable in volume, meaning they could be purchased and put into PB g5's in volume, now....

So I guess my question is, I understand most of your comments, but some of them left me wondering if you were trying to contradict me or agree with me? And since you seem to know more about them than me, when would you guess they will be available or announced?

MongoTheGeek
Feb 16, 2004, 10:25 AM
Imagine some other things that could be done with this. One of the banes of overclocking is heat dissipation. Link the speed to the thermometer. When the CPU is cool boost the clock the speed over the rated 3Ghz. Then when you are waiting on the bus cut it down to 1Ghz and let the beast cool down.

a17inchFuture
Feb 16, 2004, 10:29 AM
Good point mongo, that is a huge evolution in the way we will think about speed and processing. What will also be interesting is if this continues to be a trend among future processors, or if they will all take on forms of this in the future.

P-Worm
Feb 16, 2004, 10:59 AM
Can you do that? I never thought of processors being instantly scalable like that.

P-Worm

strider42
Feb 16, 2004, 11:02 AM
Originally posted by xtekdiver
I wonder what else Apple has in mind for this chip? How about my new iTv? If M$ puts it in their Xbox what do you think the chances are that someone will hack it and be able to run OS X on it? Oh, that would be rich! A $200 Mac from Microsoft.

couple of points here, if indeed the next xBox will use a powerPC chip, it may nto be a G5 exactly. it may be slightly changed or merely related, such as the chip in the nintendo game cube.

As for hacking it to run OS X, you'd have to hack OS X itself to work without a boot rom, which seems very unlikely to ever happen. There's more to a mac than just the basic chip design of the CPU.

Whats more, if you did this, you'd be breaking the law probably in hacking OS X, and taking money directly away from apple as you wouldn't be buying their hardware, and you'd probably be pirating Os X (I bet you wouldn't buy a 130 dollar copy just to put on your 200 dollar mac)

Fukui
Feb 16, 2004, 11:04 AM
Originally posted by GetSome681
I think people are missing the point of this. Sure, powerbook g5. Think harder. Imagine if you're doing simple stuff (browsing web, word processor) and using like 15-25% cpu...imagine the battery life you're going to end up with!
Yea, this is something that apple should have had long ago. I mean, if you just typing in word you'll only need, what 25% ~ 35% of your CPU? I would imagine a G5 could be clocked down to 300Mhz and depending on if there's flash content or movies or whatever, that would be fine...

What I'm wondering about if there are any GPUs that throttle back clock if its not being used so much... those things suck power just as much as the CPU.

manu chao
Feb 16, 2004, 11:18 AM
Originally posted by P-Worm
Can you do that? I never thought of processors being instantly scalable like that.

P-Worm

I was under the impression that the Pentium M (Centrino) does exactly that, which is one of the main reasons why laptops using it have such a long battery life.
ftp://download.intel.com/design/mob...ts/25261202.pdf

Lanbrown
Feb 16, 2004, 11:25 AM
Originally posted by a17inchFuture
great, thats all well and good, but it seems like you were reitterating some of my points rather than making your own. The chips were samples, yes, WORKING samples that serve a purpose, otherwise there would be no point in giving them to anyone early. And they are now avalable in volume, meaning they could be purchased and put into PB g5's in volume, now....

So I guess my question is, I understand most of your comments, but some of them left me wondering if you were trying to contradict me or agree with me? And since you seem to know more about them than me, when would you guess they will be available or announced?

First, the chips are available in volume, who knows what that volume is. Second, know one knows when a PB will come out with a G5. Read the following line seven times. The CPU is just one of many components. With that said, there are other components that need their power consumption reduced as well as the heat generated reduced.

MrMacMan
Feb 16, 2004, 11:30 AM
Any day now...

Come on...

Release it...

Come on!

;)

jamilecrire
Feb 16, 2004, 11:53 AM
The "unannounced" laptop containing the ATI Radeon 9700 is the new Dell laptop. It's the funniest thing I've ever seen. 2" thick, 55 lbs, runs at 180 degrees F, and is over priced as well.

I wish apple would hurry up and announce the G5 Powerbooks so I can get the 3 month wait until they are in stores over with.

mobility3
Feb 16, 2004, 12:00 PM
Imagine the power, bandwidth and server space saved from the 35-45% reduction in posts when the PB G5's come out.

That is, until the PB G6 in 2006 rumors start coming in ;)

a17inchFuture
Feb 16, 2004, 12:01 PM
Originally posted by Lanbrown
Read the following line seven times.

That's not nice. Can't you engage in friendly conversation without resorting to condescension?

In the words of the Strokes, "Looking down sometimes felt nice".

ccuilla
Feb 16, 2004, 12:03 PM
Originally posted by a17inchFuture
Aren't we smart enough consumers to absolutely demand that it be put in PB's soon

You can demand all you want (smart or not). That isn't likely to change the pace of engineering, design, etc. Sorry.

Rincewind42
Feb 16, 2004, 12:22 PM
Originally posted by Lanbrown
Based upon what? Just because the processor supports it, doesn't mean that Apple is using it. I could not find anything on that mention PowerTune on the new Xserve. Got a link?

How about reading the linked article?

"Apple will be the first customer for the new chip. IBM and Apple confirmed that the XServe G5, announced in January at MacWorld, will use the 970FX."

I would think that if Apple says that they are using the 970FX in Xserves, then they are using them in Xserves.

daveL
Feb 16, 2004, 12:25 PM
Originally posted by Lanbrown
As mentioned above, the CPU is just one of many components. I would expect new PM's to have a bus no faster than 1GHz, even if it's a 2.5 or 3GHz CPU.
So, while IBM plays to their 97x roadmap, which Apple has prior knowledge of, you suspect that Apple has been sitting on their hands with respect to the two ASICS they designed for the G5 MB? Yeah, right. There's been every indication that Apple has been moving the ASICS to the 90nm process in parallel with IBM's efforts on the 970FX. So, I completely disagree that the FSB will stall at 1 GHz while the 970FX clocks at 2.5 GHz and above.

I also think you overstate yourself, sometimes.

daveL
Feb 16, 2004, 12:30 PM
Originally posted by Rincewind42
How about reading the linked article?

"Apple will be the first customer for the new chip. IBM and Apple confirmed that the XServe G5, announced in January at MacWorld, will use the 970FX."

I would think that if Apple says that they are using the 970FX in Xserves, then they are using them in Xserves.
In his defense (I had to read his post a couple times), I believe he is saying that, although the feature is present in the 970FX, Apple is not turning it on in the OS. His belief being that power consumption doesn't matter in the Xserve, while performance is paramount. Obviously, PowerTune is present on every 970FX, but you still have to turn it on in software and manage the feature to meet your power/performance profile for your specific application.

Lanbrown
Feb 16, 2004, 12:36 PM
Originally posted by a17inchFuture
That's not nice. Can't you engage in friendly conversation without resorting to condescension?

In the words of the Strokes, "Looking down sometimes felt nice".

You apparently think the CPU is what was holding a G5 PB up. It was one of the factors. There are other components that are power hungry and put out quite a bit of heat. So one of many hurdles is taken care of. Until the other hurdles are resolved as well, no PB G5. When will they be solved, ask Apple as no one here knows the answer.

Rincewind42
Feb 16, 2004, 12:38 PM
Originally posted by daveL
In his defense (I had to read his post a couple times), I believe he is saying that, although the feature is present in the 970FX, Apple is not turning it on in the OS. His belief being that power consumption doesn't matter in the Xserve, while performance is paramount. Obviously, PowerTune is present on every 970FX, but you still have to turn it on in software and manage the feature to meet your power/performance profile for your specific application.

It's probably true that you have to turn it on, but Apple usually ships machines in a more or less neutral state wrt power management. So I would expect it to be on by default (just like with the PMG5). But my first impression was that PowerTune was more automatic than that - i.e. it turns off specific parts of the chip that are not currently in use and in a way that doesn't necessarily affect performance when needed. In that case, you would never want to turn PowerTune off because there would be no downside.

Lanbrown
Feb 16, 2004, 12:49 PM
Originally posted by Rincewind42
How about reading the linked article?

"Apple will be the first customer for the new chip. IBM and Apple confirmed that the XServe G5, announced in January at MacWorld, will use the 970FX."

I would think that if Apple says that they are using the 970FX in Xserves, then they are using them in Xserves.

The thread deals with the FX and PowerTunes that is incorporated into it. The FX is already known that it is sued in the Xserve. Mr. Anderson wondered if PowerTunes was incorporated into the Xserve, which is what my post addressed.

Just because the processor supports it, doesn't mean the Xserve will use it. Nowhere in the tech specs of the Xserve does it mention that PowerTune will be used, the Xserve does use that processor. So, show me where the Xserve will use take advantage of PowerTune.

Lanbrown
Feb 16, 2004, 12:55 PM
Originally posted by daveL
So, while IBM plays to their 97x roadmap, which Apple has prior knowledge of, you suspect that Apple has been sitting on their hands with respect to the two ASICS they designed for the G5 MB? Yeah, right. There's been every indication that Apple has been moving the ASICS to the 90nm process in parallel with IBM's efforts on the 970FX. So, I completely disagree that the FSB will stall at 1 GHz while the 970FX clocks at 2.5 GHz and above.

I also think you overstate yourself, sometimes.

The bus is already faster then what they need. Why not make it 1:1 and really have a lot of headroom. Once the bus is faster then the chip, all it does it take more power and create more heat. There are servers that are designed to run at 100% CPU usage crunching large amount of data with a slower bus than that. Even at 90-nm, they chips would run cooler at a third or a quarter of the processor for systems like the iMac and PB. They could bump the bus speed up on the PB when it's plugged into the wall though. I don't see the need (at least for now) to have a 1.5GHz bus.

daveL, that was exactly what I meant about PowerTune and Xserve.

stingerman
Feb 16, 2004, 01:05 PM
Originally posted by Lanbrown
I would say that the Xserve is not using it. It can impact performance and heat is not an issue in servers. A properly designed server will be able to handle the chip running at full power indefinitely. The fans would increase their speed to match the heat output. Some tasks could push the CPU to 100% for a brief amount of time and then that process would be complete. A lot of processes that complete in a short amount of time could hurt performance as the chip is increasing and decreasing it's power.

Yes, Xserve is using the 970FX as Apple's own documentation says it is the 90NM 970, which is only the 970FX. And, yes, there is only one version of OS X Panther Server and Power is set to automatic and even at "High" it will slew down the processors if they start overheating. Why do you think there are so many temperature sensors in the Xserve? Just so it can tell you that your processor is overheating and can watch it burn? PowerTune is also about Voltage Islands and eFuses which are an intrinsic part of the 970FX. So, yes, Powertune is being used in the Xserve as well.

Lanbrown
Feb 16, 2004, 01:08 PM
Originally posted by Rincewind42
It's probably true that you have to turn it on, but Apple usually ships machines in a more or less neutral state wrt power management. So I would expect it to be on by default (just like with the PMG5). But my first impression was that PowerTune was more automatic than that - i.e. it turns off specific parts of the chip that are not currently in use and in a way that doesn't necessarily affect performance when needed. In that case, you would never want to turn PowerTune off because there would be no downside.

There is always a downside. The requirements on a server constantly change. It might be asked to do very little now and then asked to do a large amount of work the next second and then back to little again. Having the processor ready to go to 100% at all times is what is needed. The PM G5 is a totally different beast; it is a desktop machine. If someone is going to encode a movie, they brief delay in getting the processor up to 100% won't be noticeable. If you take that delay in the VA cluster, that could add up. Enterprise servers and communication equipment do not use power saving features because they must always be available. The processor and all of the other components are ready for 100% at all times. That is why they have a lot of fans, which are strategically placed for optimum cooling efficiency.

Rincewind42
Feb 16, 2004, 01:08 PM
Originally posted by Lanbrown
The thread deals with the FX and PowerTunes that is incorporated into it. The FX is already known that it is sued in the Xserve. Mr. Anderson wondered if PowerTunes was incorporated into the Xserve, which is what my post addressed.

Just because the processor supports it, doesn't mean the Xserve will use it. Nowhere in the tech specs of the Xserve does it mention that PowerTune will be used, the Xserve does use that processor. So, show me where the Xserve will use take advantage of PowerTune.

Well, as I said, Apple is usually aggressive with power saving technologies, so I would think it would be much more likely yes than no. I also got the impression that it was not an optional technology (i.e. 24.5W at 2Ghz required PowerTune) which would mean by extension that it was being used. If that is not the case, maybe the next OS X seed is what is needed to enable it.

And the Hardware Developer technote for the Xserve G5 hasn't come out yet - until it does I wouldn't expect to get any confirmation one way or another. You'll note that PowerMac G5 tech overview doesn't mention that it cycles the cpus down to ~1.3Ghz during idle either, but it does.

Lanbrown
Feb 16, 2004, 01:10 PM
Originally posted by stingerman
Yes, Xserve is using the 970FX as Apple's own documentation says it is the 90NM 970, which is only the 970FX. And, yes, there is only one version of OS X Panther Server and Power is set to automatic and even at "High" it will slew down the processors if they start overheating. Why do you think there are so many temperature sensors in the Xserve? Just so it can tell you that your processor is overheating and can watch it burn? PowerTune is also about Voltage Islands and eFuses which are an intrinsic part of the 970FX. So, yes, Powertune is being used in the Xserve as well.

This thread is NOT about what processor the Xserve has, it is about PowerTune. Slowing down the processor because of it overheating is different then PowerTune.

Show the Apple doc where it mentions PowerTune and the Xserve.

Rincewind42
Feb 16, 2004, 01:13 PM
Originally posted by Lanbrown
There is always a downside. The requirements on a server constantly change. It might be asked to do very little now and then asked to do a large amount of work the next second and then back to little again. Having the processor ready to go to 100% at all times is what is needed. The PM G5 is a totally different beast; it is a desktop machine. If someone is going to encode a movie, they brief delay in getting the processor up to 100% won't be noticeable. If you take that delay in the VA cluster, that could add up. Enterprise servers and communication equipment do not use power saving features because they must always be available. The processor and all of the other components are ready for 100% at all times. That is why they have a lot of fans, which are strategically placed for optimum cooling efficiency.

Most server work requires fast IO systems more so than fast CPUs. In those cases the CPU just becomes a dynamic routing device for all that IO. That doesn't require a huge amount of CPU power. And there is nothing to say that PowerTune wouldn't work well in a situation where requirements were constantly changing - a thousandth of a second isn't all that much time, typically on the order of the amount of time a thread is allocated before it is swapped out for another one. If the OS can hint the CPU with say "This thread wasn't using floating point last time, so turn off the FP units" it can be a huge power win. If you guess wrong, big deal - but I bet the technology would also be smarter than that, as you do get a long look ahead on what instructions are coming up so you can easily turn off and on a unit that isn't doing anything for a while.

So really, I can't see too many downsides, and if they do in fact exist and are significant, then they can always be turned off.

Lanbrown
Feb 16, 2004, 01:22 PM
Originally posted by Rincewind42
Well, as I said, Apple is usually aggressive with power saving technologies, so I would think it would be much more likely yes than no. I also got the impression that it was not an optional technology (i.e. 24.5W at 2Ghz required PowerTune) which would mean by extension that it was being used. If that is not the case, maybe the next OS X seed is what is needed to enable it.

PowerTune functionality is not required for the chip to work. The functionality is there if a company wanted to use it. PowerTune is there to reduce power consumption when the CPU load is low. That is fine for a notebook and a desktop. People leave them on all the time, so most of the time it sits there and does nothing. A server on the hand is different. During the day, users put a load on it. At night an administrator might have it perform a backup and other daily/weekly routines. So it might actually see more of a load at night then during the day. Batch processing could be done as well. One some billing systems, at 8 AM is when they start getting nailed and it doesn't ease up until last at night and it still see spikes and such in the wee hours of the morning. PowerTune would do absolutely nothing in that scenario. In a scenario where the CPU would spike briefly and then settle back down. PowerTune would cause a slight degradation in performance as the CPU would be allowed to be fully utilized and then have it throttle back. The majority of your servers today do not have a nice even CPU load.

arn
Feb 16, 2004, 01:23 PM
In case anyone was wondering...

IBM's presentation on the PowerPC 970 is at 4:45pm PST today at the ISSCC

There is a PowerTune presentation on Wednesday.

We have to rely on the regular tech-web-news to report on these presentations.

arn

MongoTheGeek
Feb 16, 2004, 01:31 PM
Originally posted by arn
In case anyone was wondering...

IBM's presentation on the PowerPC 970 is at 4:45pm PST today at the ISSCC

There is a PowerTune presentation on Wednesday.

We have to rely on the regular tech-web-news to report on these presentations.

arn

Does the timing suggest big announcements for tomorrow morning?

luggnutt
Feb 16, 2004, 01:37 PM
Originally posted by Lanbrown
The CPU is just one of many components. With that said, there are other components that need their power consumption reduced as well as the heat generated reduced.

If Apple engineering is at all like a typical engineering design system, they have a small group of engineers dedicated to portables, and maybe even PB. I would say this is very likely since teh PB has become a cash cow for Apple. They would keep in excellect contact with a "Futures Engineering" group that is looking a few years down the road, determining feasibility, and making crude prelim designs. I assume Apple has adopted the trend of concurrent design, so the PB engineering group would also keep close contact with suppliers, manufacturing, and everyone else that affects the product delivery timetable.

With all that in mind, what do you think the PB engineers have been doing for the last year? They have known that the G5 is coming for many years, and the future engineering group likely has had a few PB G5 concepts for a while.

The processor IS just one of many componants, but as far as I know it is the componant over which Apple has teh LEAST control. Apple has had time to develop all the rest of the PB, now the G5 is about ready.

Therefore, I see no reason why the PB G5 design would not be done within weeks. Please explain where I am wrong, as you seem quite adamant that PB G5 is much farther down the road.

Lanbrown
Feb 16, 2004, 01:41 PM
Originally posted by Rincewind42
Most server work requires fast IO systems more so than fast CPUs. In those cases the CPU just becomes a dynamic routing device for all that IO. That doesn't require a huge amount of CPU power. And there is nothing to say that PowerTune wouldn't work well in a situation where requirements were constantly changing - a thousandth of a second isn't all that much time, typically on the order of the amount of time a thread is allocated before it is swapped out for another one. If the OS can hint the CPU with say "This thread wasn't using floating point last time, so turn off the FP units" it can be a huge power win. If you guess wrong, big deal - but I bet the technology would also be smarter than that, as you do get a long look ahead on what instructions are coming up so you can easily turn off and on a unit that isn't doing anything for a while.

So really, I can't see too many downsides, and if they do in fact exist and are significant, then they can always be turned off.

Big deal? If you go with one thousandth of a second and 2200 processors, that is 2.2 seconds. In that 2.2 second look at how many calculation could be completed. Using the Apple benchmarks, that would be 19.8 gigaflops of double precision floating point operations or 6.82 million nucleotides. That is a lot of lost time as the CPU gets ramped up. What if the last thread only took a few second to complete?

utilizer
Feb 16, 2004, 01:47 PM
Originally posted by luggnutt


With all that in mind, what do you think the PB engineers have been doing for the last year? They have known that the G5 is coming for many years, and the future engineering group likely has had a few PB G5 concepts for a while.

The processor IS just one of many componants, but as far as I know it is the componant over which Apple has teh LEAST control. Apple has had time to develop all the rest of the PB, now the G5 is about ready.

Therefore, I see no reason why the PB G5 design would not be done within weeks. Please explain where I am wrong, as you seem quite adamant that PB G5 is much farther down the road.

You stated that FAR BETTER than I could! I'm glad you said it and posted it for the rest...it's true people. Just expect it very soon

Lanbrown
Feb 16, 2004, 01:57 PM
Originally posted by luggnutt
If Apple engineering is at all like a typical engineering design system, they have a small group of engineers dedicated to portables, and maybe even PB. I would say this is very likely since teh PB has become a cash cow for Apple. They would keep in excellect contact with a "Futures Engineering" group that is looking a few years down the road, determining feasibility, and making crude prelim designs. I assume Apple has adopted the trend of concurrent design, so the PB engineering group would also keep close contact with suppliers, manufacturing, and everyone else that affects the product delivery timetable.

With all that in mind, what do you think the PB engineers have been doing for the last year? They have known that the G5 is coming for many years, and the future engineering group likely has had a few PB G5 concepts for a while.

The processor IS just one of many componants, but as far as I know it is the componant over which Apple has teh LEAST control. Apple has had time to develop all the rest of the PB, now the G5 is about ready.

Therefore, I see no reason why the PB G5 design would not be done within weeks. Please explain where I am wrong, as you seem quite adamant that PB G5 is much farther down the road.

Where did I ever say that Apple was doing nothing in regards to a G5 in a PB? IBM could have gotten the processor done before Apple was ready for it in regards to it being used in a G5 PB. The chipsets required could be quite different then what the PM is using. Creating a new chipset from scratch at a new process (90-nm) is a huge undertaking. All of the 90-nm products available now were once fabbed at 130-nm. The 970FX was mainly just a die shrink from 130-nm with a few additions to it. Not the same as something built for 90-nm from the start. Most companies move an existing product to a new process as they know the ins and outs and know that it works. If they have problems with the end product, they know what it is, the process and not the design. 90-nm is very new to most companies, IBM included.

I don't know when a G5 PB will be ready, neither does anyone else here. If someone does know, they can't say because of an NDA. Look at some of the other threads. Like the Hypertransport 2 announcement. People said now that that is out, Apple will release new PM's. Didnít happen did it? Why do you think within weeks? The Xserve was announced last month. How do you know the holdup was IBM? When will G5 based PB's be available? When Apple is ready to release them that is when, not before and not after. Just because an announcement is made doesn't mean you will see Apple release the hardware that will use it the same day, same week or even the same month.

Consider what I'm saying as a dose of reality, instead of hanging on to a strand of hope at every announcement. How many times were the PM's and PB's supposed to be updated? At least three times in January, then again for HT2, etc.

hose this!
Feb 16, 2004, 02:18 PM
Since WHEN has Apple ever been speedy with updates, especially major ones?

What we can expect:

1) No news on PB updates for 6 months
2) An announcement, then a two month wait for actual shipping
3) A delay
4) Products ship, but something is wrong with it
5) Apple denies there's anything wrong
6) Consumer advocacy groups get called in
7) Apple tries to resolve the issue
8) Apple gets sued
9) Macolytes spend countless hours on Apple rumor sites using the MHz Myth as a crutch to defend plodding G4s

I'll give 2-1 if anyone cares to make a bet ;)

arn
Feb 16, 2004, 02:26 PM
Originally posted by MongoTheGeek
Does the timing suggest big announcements for tomorrow morning?

No.

arn

a17inchFuture
Feb 16, 2004, 02:30 PM
Lan brown: what do the pb's require that the present PB's don't have?

Everyone else: It seems like everyone else is right here, its not like the G5 powerbook idea is coming out of nowhere: the most significant reason that all of us have cited over the past 8 months as to why they wouldn't get G5s in powerbooks soon was the processors' heat production and energy consumption. With these two problems seemingly fixed, what (besides the bus) do you think needs to be implemented into the design of the PB? I mean, I know very little about what the differences between G5 and G4 will present as far as the other parts, but it would seem to me that they don't have to replace everything in the PB just becasue its a next gen processor. Am I wrong about this, or does Lanbrown actually know about as much as me?

Rincewind42
Feb 16, 2004, 02:44 PM
Originally posted by a17inchFuture
I mean, I know very little about what the differences between G5 and G4 will present as far as the other parts, but it would seem to me that they don't have to replace everything in the PB just becasue its a next gen processor. Am I wrong about this, or does Lanbrown actually know about as much as me?

Your right in that there is little that _has_ to change, but there is likely a lot that will change. Of course the good news here is that Apple has had all the components except the CPU under their control for the entire time. The CPU is the only thing that Apple simply had to wait for to make a PBG5. Yes, they would almost certainly need to do a process shrink for the system controller - but you can bet that Apple was already working on that design simultaneously with the 130nm part because they knew that IBM was working on a 90nm process. And the moment IBM started sampling on that process, you can bet that Apple was too. There is no reason to think that anything that Apple needed to process shrink for the PBG5 couldn't have been worked on simultaneously with the 970FX. So for all the reality checks that people seem to be claiming, here's one: Nobody is waiting for anybody to design the technologies they need for the future of their products. Apple knew that IBM was working on the 970FX, so they work on their system controller in step and so on we go. If Apple decides that it's time to do the PBG5, they could probably have it by May. If they decide they need to do more work, then they do more work. But you can bet that Apple hasn't been sitting on the side lines saying "The 970FX isn't done yet, so lets not bother working on the PBG5" - no, you can bet there are lots of PBG5 prototypes roaming around the Apple Campus right this moment.

0 and A ai
Feb 16, 2004, 02:56 PM
Originally posted by xtekdiver
I wonder what else Apple has in mind for this chip? How about my new iTv? If M$ puts it in their Xbox what do you think the chances are that someone will hack it and be able to run OS X on it? Oh, that would be rich! A $200 Mac from Microsoft.

why do people keep saying this.

IT hasn't happened with the gamecube it wno'thappen with the xbox 2. it just isn't.

Photorun
Feb 16, 2004, 03:07 PM
Very cool indeed! And with the demand for this chip and it's lower price Apple may actually be able to give us fast machines at [gasp] reasonable prices!

Bring on the iPod G5 mini!

Lanbrown
Feb 16, 2004, 03:14 PM
Originally posted by a17inchFuture
Lan brown: what do the pb's require that the present PB's don't have?

Everyone else: It seems like everyone else is right here, its not like the G5 powerbook idea is coming out of nowhere: the most significant reason that all of us have cited over the past 8 months as to why they wouldn't get G5s in powerbooks soon was the processors' heat production and energy consumption. With these two problems seemingly fixed, what (besides the bus) do you think needs to be implemented into the design of the PB? I mean, I know very little about what the differences between G5 and G4 will present as far as the other parts, but it would seem to me that they don't have to replace everything in the PB just becasue its a next gen processor. Am I wrong about this, or does Lanbrown actually know about as much as me?

Present PB's have a 167Mhz bus. The G5 would have a much faster bus. The bus alone would require a new chipset. The faster the bus, the more power will be required, that shortens the runtime down and since it requires more power, it will generate more heat. Second, I can't see Apple using the same case for the G5 as they did for the G4. So a new case will need to be designed. Maybe Apple wants to one up the competition with a 20" display. I highly doubt that, as it would be one very big notebook. Apple would probably want to put HT in it as well. That is something the current PB does not have. There might be some delays in other technologies that Apple wants to put in it that are not quite done yet. Maybe they wanted to out HT2 in it as the next clean slate design will be 18 or so months down the road. In April, SATA drives are expected to be available in volume. Maybe Apple has decided to be the first manufacturer or one of the first to offer SATA on their products. Right now only the PM G5 has it. They could use one SATA bus for both the optical drive and HD with no adverse effects. This would go for the iMac as well. If you think you will see a G5 based PB with the same features as the G4 version, you will be very wrong. If we knew what Apple was planning (feature wise), that would help with a timetable. If they were thinking of SATA drive, Fujitsu won't have them available until April, but samples are available now. If Apple wants to offer SATA, then they must wait. If not, then that can't be the holdup. The chipset at 90-nm could be the problem as well. Very few companies are offering 90-nm parts in volume. Maybe there is a chipset design problem that is being worked out right now. As I stated in a previous post, most companies have used an existing design and moved it to 90-nm. Ti has with their OMAP line, Sun with the SPARC (even though it's not available), IBM with the 970. Intel went with a new design and so far, it's late by almost two months. The chipset could be a new design and given that a new process always has some bugs to work out (ask Intel), that's why companies use an existing design first. It's a benchmark to when they run into a problem. They know it's not the design, but the process.

If IBM can supply Apple with enough 970FX processors right now, then there is absolutely no reason why a G5 based PB hasn't been announced except for a different problem. It's not like Apple isn't afraid to announce something that won't be available for a few months, G5 PM's and the 17" PB are two examples. So Apple obviously hasn't announced them for a reason. It's either that the volume production isn't high enough for Apple at this time or there are other problems. See first paragraph for examples.

I undoubtedly know more then you do. There are some who jump at every announcement to say that new updates are coming and since [insert announcement here] is out, updates will be coming. HT2 is a good example. HT2 is announced, new PM's will be announced shortly. So far, nothing. Why, because they probably won't use HT2 at this time.

Lanbrown
Feb 16, 2004, 03:28 PM
Originally posted by Rincewind42
Your right in that there is little that _has_ to change, but there is likely a lot that will change. Of course the good news here is that Apple has had all the components except the CPU under their control for the entire time. The CPU is the only thing that Apple simply had to wait for to make a PBG5. Yes, they would almost certainly need to do a process shrink for the system controller - but you can bet that Apple was already working on that design simultaneously with the 130nm part because they knew that IBM was working on a 90nm process. And the moment IBM started sampling on that process, you can bet that Apple was too. There is no reason to think that anything that Apple needed to process shrink for the PBG5 couldn't have been worked on simultaneously with the 970FX. So for all the reality checks that people seem to be claiming, here's one: Nobody is waiting for anybody to design the technologies they need for the future of their products. Apple knew that IBM was working on the 970FX, so they work on their system controller in step and so on we go. If Apple decides that it's time to do the PBG5, they could probably have it by May. If they decide they need to do more work, then they do more work. But you can bet that Apple hasn't been sitting on the side lines saying "The 970FX isn't done yet, so lets not bother working on the PBG5" - no, you can bet there are lots of PBG5 prototypes roaming around the Apple Campus right this moment.

You are basing your assumption that IBM and Apple were moving at the same pace. IBM could have completed theirs earlier than expected. Apple could have taken longer than expected. It could have been a combination of the two or just one. Very few products are on the 90-nm process. Quite a few are being sampled on it, but very little in volume.

I didn't see where anyone was claiming Apple hasn't done anything about a PB G5.

pbrennen
Feb 16, 2004, 03:33 PM
Originally posted by manu chao
I was under the impression that the Pentium M (Centrino) does exactly that, which is one of the main reasons why laptops using it have such a long battery life.
ftp://download.intel.com/design/mob...ts/25261202.pdf can anyone compare and contrast powertune, centrino, and transmeta processors?

rjwill246
Feb 16, 2004, 03:44 PM
IBM ( I'm Becoming Macintosh) and Apple look so good together these days, I wonder at what price Apple investors would 'bite' if IBM made an offer to 'buy' the fruit company?

daveL
Feb 16, 2004, 03:47 PM
Originally posted by Lanbrown
You are basing your assumption that IBM and Apple were moving at the same pace. IBM could have completed theirs earlier than expected. Apple could have taken longer than expected. It could have been a combination of the two or just one. Very few products are on the 90-nm process. Quite a few are being sampled on it, but very little in volume.

I didn't see where anyone was claiming Apple hasn't done anything about a PB G5.
Guess you didn't catch this thread:
http://www.macrumors.com/pages/2004/02/20040211102254.shtml

Unfortunately, the article, which was an interview with an IBM official, was pulled within hours of it's release. The main points were:

1) IBM has a 95% first mask success rate for ASICS they design, while their foundry clients' designs come in around 5%, and

2) IBM was about to delivery volume of their first SOC on the 90 nm process to Apple.

Here's a partial quote (Reeves is the IBM person):

"IBM is about to deliver its very first volume production of an SOC on 90nm, said Reeves. This will be a design for Apple."

MikeAtari
Feb 16, 2004, 03:58 PM
AMD is selling 64 bit notebooks NOW!

With a faster bus now!
I think if VIA can produce a 64 bit motherboard for a notebook, Apple should be able to as well.

skinEman23
Feb 16, 2004, 03:59 PM
Originally posted by Lanbrown
I undoubtedly know more then you do.

First, if you are going to be pompous do it correctly. You undoubtedly know more THAN we do. Second, I think that you are exactly right. If Apple had even the slightest hopes that it could release a PB G5 in spring, or even early summer, it would have said that. Instead, Jobs said he hopes to have one by the end of 2004. I have a hard time believing that he would say that and then release one a month later. I also have a hard time seeing a six-month old laptop continue selling at its initial price with absolutely no upgrades. Either a price drop or an upgraded G4 will come before a PB G5.

MikeAtari
Feb 16, 2004, 04:24 PM
http://www.e4me.com/products/products.html?prod=eMachines_M6805

Now, the story here is, that a 64bit laptop is available. But, it's a desktop replacement type laptop. i.e. it's hot with short battery life.

So, the question is, would Apple produce this type of machine? The current lineup of powerbooks says, no. I think this is a mistake.

Apple should, ( elect me to the board! ), co-produce G4 and G5 laptops. Until, the G5 laptop can be run as a "real laptop" and not just a Desktop Replacement.

Give us the choice.
We will decide if we want a cool running G4 or a hotter, faster G5.

Rincewind42
Feb 16, 2004, 04:28 PM
Originally posted by skinEman23
If Apple had even the slightest hopes that it could release a PB G5 in spring, or even early summer, it would have said that. Instead, Jobs said he hopes to have one by the end of 2004.

Why would you think that? If things go better than they expect, you can bet they will release sooner than later. But if something goes horribly wrong, then they will be releasing later rather than sooner. By saying he hopes to do it by the end of the year, he pretty much gives a huge leeway in when it happens without killing sales of the PBG4 altogether. If Jobs has said "sure, by spring 2004" PBG4 sales would be tanking right now. And then if in April Apple released a speed bumped G4, no one would buy it thinking that PBG5 must still be right around the corner. Whereas by saying end of 2004, they can release next week, or Dec 25 and neither affect current sales, nor have egg on their face if they fail to meet their "deadline".

digitalbiker
Feb 16, 2004, 04:38 PM
If he thinks that power consumption and heat aren't important in a clustered super computer configuration then he most certainly does not know more than us.

One of the major problems facing intel super computer clusters is power consumption and heat output. Once you get so many cpu's clustered maintaining steady power and keeping heat from crashing cpus becomes a major headache.

I could see where powertune technology could be very welcome by the Virginia tech group converting to xserve based clustering.

I think lanbrown is just trolling because his comments have been purely negative on a very positive news piece.

spankalee
Feb 16, 2004, 05:03 PM
So how do you know so much about PowerTune? I didn't see anywhere in the article where it said that using PowerTune adversly effected performance, or even that not using PowerTune was an option.

It sounds to me like from what I've read other places that PowerTune is a wholistic approach to reducing power consumption that has little to do with scaling core speeds and more to do with not powering parts of the chip that aren't being used.

It doesn't look like PowerTune is something that has to be enabled in the OS. Until IBM's presentation, my guess is that saying you can turn off PowerTune is like saying you can "turn off" SOI if the computer doesn't need it.



Originally posted by Lanbrown
PowerTune functionality is not required for the chip to work. The functionality is there if a company wanted to use it. PowerTune is there to reduce power consumption when the CPU load is low.

Lanbrown
Feb 16, 2004, 05:06 PM
Originally posted by digitalbiker
If he thinks that power consumption and heat aren't important in a clustered super computer configuration then he most certainly does not know more than us.

One of the major problems facing intel super computer clusters is power consumption and heat output. Once you get so many cpu's clustered maintaining steady power and keeping heat from crashing cpus becomes a major headache.

I could see where powertune technology could be very welcome by the Virginia tech group converting to xserve based clustering.

I think lanbrown is just trolling because his comments have been purely negative on a very positive news piece.

Try again; I've been here for sometime now. I am just adding a dose of reality. People have been talking about G5 PB's for months now. Whenever a scrap of news is released, guess what, G5 based PBís are on the way, or so they say.

Most server chips take much more power then the 970FX does. The Power4+ uses 35 watts at 1.2GHz. That's higher then the 970FX, which is at a higher clock speed.

Intel is a different story; they are new to the big iron game. Companies like Sun, IBM, Cray, Fujitsu, etc. have been in that game for a longtime and don't have cooling issues. Intel is making some very hot processors these days.

How would PowerTune help them? They are constantly crunching numbers. No rest for those processors.

Lanbrown
Feb 16, 2004, 05:11 PM
Originally posted by spankalee
So how do you know so much about PowerTune? I didn't see anywhere in the article where it said that using PowerTune adversly effected performance, or even that not using PowerTune was an option.

It sounds to me like from what I've read other places that PowerTune is a wholistic approach to reducing power consumption that has little to do with scaling core speeds and more to do with not powering parts of the chip that aren't being used.

It doesn't look like PowerTune is something that has to be enabled in the OS. Until IBM's presentation, my guess is that saying you can turn off PowerTune is like saying you can "turn off" SOI if the computer doesn't need it.

Because if something is off, you canít use it until itís ready. Data in the pipeline will have to momentarily wait and the CPU spends most of the time waiting for data.

Power management is not something new to chips. Just because it's built-in doesn't mean it must be used. Software can control PM features though. PM is a combination of hardware and software. If the fp is off and a request comes in, it now has to be turned on. That will create a delay.

Photorun
Feb 16, 2004, 05:12 PM
Originally posted by Lanbrown
Try again; I've been here for sometime now. I am just adding a dose of reality. People have been talking about G5 PB's for months now. Whenever a scrap of news is released, guess what, G5 based PBís are on the way, or so they say.

Most server chips take much more power then the 970FX does. The Power4+ uses 35 watts at 1.2GHz. That's higher then the 970FX, which is at a higher clock speed.

Intel is a different story; they are new to the big iron game. Companies like Sun, IBM, Cray, Fujitsu, etc. have been in that game for a longtime and don't have cooling issues. Intel is making some very hot processors these days.

How would PowerTune help them? They are constantly crunching numbers. No rest for those processors.

Here here and correct. Worse-so for Intel, their process, to use a non-technical term... sucks. Basically they're the Microsoft of chip makers, bloated and buggy, cheaply made with haphazard design. AMD is kinda like the Apple of x86 chips, really good company, not much marketing push, and of course people, especially peecee lusers, are an ignorant lot, and think Microshaft and Insmell are synonymous with "computer." But you're talking about server chips, yes, heat is an issue... but heat is something the server chips makers have addressed and know how to, better than Intel does... amen.

And LanBrown is most certainly not a troll.

Rincewind42
Feb 16, 2004, 05:13 PM
Originally posted by Lanbrown
How would PowerTune help them? They are constantly crunching numbers. No rest for those processors.

If PowerTune allows the 970FX to turn off the Altivec Units and one of the Integer units during a huge floating point operation, and thus reduce power usage by say 25%, then that translates directly into savings in electricity, and increases the reliability of the entire cluster by making it less likely that any particular system will fail. It will also allow the system to keep the CPU cycling at a higher frequency when the OS would normally reduce the power usage to avoid overheating.

I personally don't see any downside to PowerTune on any hardware.

Edit: forgot the system will cycle down if it is overheating regardless of settings.

stingerman
Feb 16, 2004, 05:14 PM
Originally posted by Lanbrown
This thread is NOT about what processor the Xserve has, it is about PowerTune. Slowing down the processor because of it overheating is different then PowerTune.

Show the Apple doc where it mentions PowerTune and the Xserve.

You don't even know what PowerTune is. You can't have a 970FX without it! PowerTune is a collection of technologies: Voltage Islands; Voltage Slewing inline with Frequency Slewing; eFuse; new protected and user instructions to allow the OS and Apps to work with PowerTune.

You assume that if Apple chooses to turn off automatic slewing (which is a user controllable feature), all these powertune futures are not available. Your wrong. 970FX is designed usign Voltage Islands which allows different parts of the processor to work at a lower voltage than the core or caches. So Powertune is part of the 970FX fabric, you can't have one without the other.

Software API wise, it allows developers to tell the processor that we need max performance now, and now we don't. IBM gives the example of DVD playback, the processor switches up in under a uS, decodes the frame, switches down, then back up in time to deliver 60FPS. In the process it uses 25% of the power.

So basically bring your knowledge up-to-date, PowerTune is not speed stepping, it is a completely new technology. And if you think super clusters are not drooling over this feature of the 970FX, you are out of your mind. Do you have any idea of the electrical costs associated with running a huge cluster. Rarely does an operation use 100% of each processor. The power savings would be a huge selling factor to small and large firms. I know, I run a couple of Rack full of Dell 2U's and power costs make up a big part of my monthly data center costs, even when the servers are hardly being hit, like after 10PM or so.

Snowy_River
Feb 16, 2004, 05:18 PM
...know one ...

If you know one, you know them all...


Just a little bit about typing what you mean. There is a difference between 'know' and 'no'.
This is one of those typos that's a real pet peeve for me.

Sorry for going off topic. Every now and then I need to vent about this kind of thing... :)



Corrected the typo 'peave' for 'peeve', as was pointed out by others.

stingerman
Feb 16, 2004, 05:19 PM
Originally posted by Rincewind42
If PowerTune allows the 970FX to turn off the Altivec Units and one of the Integer units during a huge floating point operation, and thus reduce power usage by say 25%, then that translates directly into savings in electricity, and increases the reliability of the entire cluster by making it less likely that any particular system will fail. It will also allow the system to keep the CPU cycling at a higher frequency when the OS would normally reduce the power usage to avoid overheating.

I personally don't see any downside to PowerTune on any hardware.

Edit: forgot the system will cycle down if it is overheating regardless of settings.

It's not like that with powertune. The system doesn't turn off, but actually "naps" not "sleeps" portions of the processor not being used. However there is a trickle of power going down those circuits that allows them to wake right up, not out of a deep sleep but out of a light nap, so to speak. It happens so quick that youwill not notice a delay as it takes only uSec to come fully awake.

Rincewind42
Feb 16, 2004, 05:24 PM
Originally posted by stingerman
It's not like that with powertune. The system doesn't turn off, but actually "naps" not "sleeps" portions of the processor not being used. However there is a trickle of power going down those circuits that allows them to wake right up, not out of a deep sleep but out of a light nap, so to speak. It happens so quick that youwill not notice a delay as it takes only uSec to come fully awake.

Cool, little more info in my head :). Do you have a link that tells more about PowerTune? I've been going off of intuition more than knowledge today :D.

PBGPowerbook
Feb 16, 2004, 05:27 PM
Originally posted by a17inchFuture
With these two problems seemingly fixed, what (besides the bus) do you think needs to be implemented into the design of the PB? I mean, I know very little about what the differences between G5 and G4 will present as far as the other parts, but it would seem to me that they don't have to replace everything in the PB just becasue its a next gen processor. [/B]

The powerbook g5 is

black anodized aluminum
red glowing apple in cover Š la current notebooks
grille/mesh on parts of side edges
backlit keyboard white letters on black keys

stingerman
Feb 16, 2004, 05:42 PM
Originally posted by Rincewind42
Cool, little more info in my head :). Do you have a link that tells more about PowerTune? I've been going off of intuition more than knowledge today :D.

If I posted the link, I'd have to kill everyone that saw it afterwards. Actually, it is all on IBM's research site. I'll do better than a link, I'll teach you how to fish on the Internet. In the Google search enter:

voltage islands research site:ibm.com

that should get you started. But the IBM research site has so much more.

Snowy_River
Feb 16, 2004, 05:45 PM
Originally posted by utilizer
Anyways though, I'm calling it:
-PowerBooks ... 12 in. form factor retains the G4...

God, I hope not! That would be terrible... Better to discontinue the 12" line and leave that size to the iBook than to have a G4 still in the PowerBook line after the G5 PowerBooks have been released...

stingerman
Feb 16, 2004, 06:13 PM
Originally posted by Snowy_River
God, I hope not! That would be terrible... Better to discontinue the 12" line and leave that size to the iBook than to have a G4 still in the PowerBook line after the G5 PowerBooks have been released...

I am hoping Apple has other plans for the 12" line. Not specifically the 970FX, which should work fine in that form factor at 1.4-1.6GHz. But, I would like to see them bring out a SoC design, that should let them reduce the thickness and weight. The weight under 3lbs. After all, you buy a 12" because its size and weght are more important than performance (though performance is important.) In the Wintel world small form factor notebooks 3 pounds and under seem to have a ceiling of 700MHz-1GHz, but Apple can beat that with a 90NM SoC design.

centauratlas
Feb 16, 2004, 06:29 PM
Originally posted by Lanbrown
I undoubtedly know more then you do.

I just have to say that if you are going to make a comment like that, you need to do more proof-reading because your writing certainly does NOT demonstrate it:
1. It isn't "then" - *thAn* you do. (above)
2. "known that it is sued in the Xserve". (1:49pm) It has an attorney?
3. " show me where the Xserve will use take advantage of PowerTune."(1:49pm) Huh?
4. " then the chip". Again, "than". (1;55pm)
5. " they brief delay in getting". The?

And that was a QUICK, incomplete look.

I agree with some of what you said, but a comment like that is SURE to turn people off.

Apple *wants* a PB G5, obviously they know sales are impacted. They won't *announce* it until they can say when it will be available (even if it slips a bit) and can sell it via pre-orders.

They likewise won't pre-announce more than 2 months because it would *kill* G4 sales and they can't afford a quarter of low(er) PB sales from any perspective. That is why Jobs says "by the end of 2004". It gives flexibility and it gives them the option of doing it sooner if at all possible.

MacEyeDoc
Feb 16, 2004, 06:39 PM
The article states that the chip could be used from PC's to laptops to servers. Apple appears to be using it in the Xserve, so what will they use it in next? I don't think I've heard any Apple pronouncements about when the next laptops or PowerMacs are coming. They could be working on one or the other, or both, but what would give them the biggest splash? They won't be selling PowerMacs for clusters anymore, so maybe that's why they came out with the Xserve first. (And Steve did say in his MacWorld SF keynote that he knew he was pissing off a thousand customers by diverting the first PowerMacs to Virginia Tech, but he didn't care). So if they have plans to sell several thousand Xserves to academic or government installations, and all the new PowerMacs and PowerBooks use the same chip, does that mean that we, the common consumers, will be waiting again? I'm sure you guys have an opinion . . . .

Messiatosh
Feb 16, 2004, 06:41 PM
This is what matters right now...check out the video...

iPod mini commercial (http://homepage.mac.com/chrisdieugenio/iMovieTheater28.html)

Dont Hurt Me
Feb 16, 2004, 06:47 PM
The common consumer is the last one Apple thinks of. the computer that saved all their hides was Imac crt. consumers bought it up. without it you could have said bye bye to Apple. Consumer lines are crippled by G4 then a little more- low clock G4s and then a little more- taking L3 out and then a little more- with bottom video cards.=== poor hardware for the consumer. They dont get it.:mad: Hey i know what! we can put a 52" screen on crippledG4 imac and maybe we will get a bunch of customers:rolleyes: If there is any one product screaming for 970fx or even current technology it is Imac.

daveL
Feb 16, 2004, 06:47 PM
Originally posted by MacEyeDoc
The article states that the chip could be used from PC's to laptops to servers. Apple appears to be using it in the Xserve, so what will they use it in next? I don't think I've heard any Apple pronouncements about when the next laptops or PowerMacs are coming. They could be working on one or the other, or both, but what would give them the biggest splash? They won't be selling PowerMacs for clusters anymore, so maybe that's why they came out with the Xserve first. (And Steve did say in his MacWorld SF keynote that he knew he was pissing off a thousand customers by diverting the first PowerMacs to Virginia Tech, but he didn't care). So if they have plans to sell several thousand Xserves to academic or government installations, and all the new PowerMacs and PowerBooks use the same chip, does that mean that we, the common consumers, will be waiting again? I'm sure you guys have an opinion . . . .
Well, IBM doesn't seem to having yield problems on the new process (see my previous post in the thread, I think). They're using 300mm wafers (big) and a 90nm process (small), which all means "lots of chips". Plus the Xservers are using lower clocked chips, at least for now. So, I don't think there is a shortage of parts coming from IBM.

Just my take on it.

savar
Feb 16, 2004, 06:52 PM
my friends wintel laptop can do something similar, although i dont know if it can vary clock speed "several thousand times per second"...anybody know how this compares to existing technology from other vendors?

Rincewind42
Feb 16, 2004, 06:56 PM
Originally posted by stingerman
If I posted the link, I'd have to kill everyone that saw it afterwards. Actually, it is all on IBM's research site. I'll do better than a link, I'll teach you how to fish on the Internet. In the Google search enter:

voltage islands research site:ibm.com

that should get you started. But the IBM research site has so much more.

Well, I tried "powertune site:ibm.com" and got nothing :). Thought the info might have still been under embargo somewhere. Thanks for the info tho.

stingerman
Feb 16, 2004, 06:56 PM
Originally posted by daveL
Well, IBM doesn't seem to having yield problems on the new process (see my previous post in the thread, I think). They're using 300mm wafers (big) and a 90nm process (small), which all means "lots of chips". Plus the Xservers are using lower clocked chips, at least for now. So, I don't think there is a shortage of parts coming from IBM.

Just my take on it.

No shortage of 970FX's at all, IBM can run huge quantities in just a matter of days.

They were smart to release the Xserve at 2GHz otherwise PM G5 owners would have put their orders on hold. Once the PM G5 revision is out, I suspect the Xserve's will very quickly jump up in speed.

spankalee
Feb 16, 2004, 06:59 PM
Originally posted by stingerman
You don't even know what PowerTune is. You can't have a 970FX without it! PowerTune is a collection of technologies: Voltage Islands; Voltage Slewing inline with Frequency Slewing; eFuse; new protected and user instructions to allow the OS and Apps to work with PowerTune.


Nice post stingerman. Thanks for coming in and shedding some light on the topic. So many uninformed people like to come into forums, act like they know more than everyone when they don't, and cry that the sky is falling about everything. I like it when the real knowledgeable people speak up.

MacEyeDoc
Feb 16, 2004, 07:17 PM
Originally posted by stingerman
No shortage of 970FX's at all, IBM can run huge quantities in just a matter of days.

They were smart to release the Xserve at 2GHz otherwise PM G5 owners would have put their orders on hold. Once the PM G5 revision is out, I suspect the Xserve's will very quickly jump up in speed.

Good point about the Xserve at 2.0 instead of 2.5 - people wold have smelled the PowerMac at 2.5 and sales would have nosedived. Any comments on the new PowerMac not being the typical Rev. B we are used to seeing, since it will be sporting a completely new chip technology? You usually wait for the Rev. B version to be a slightly faster and more solid version of the first, but now we have a new type of chip - will it be the Rev. A PowerMac G5FX?

jouster
Feb 16, 2004, 07:37 PM
Originally posted by Snowy_River
If you know one, you know them all...


This is one of those typos that's a real pet peave for me.


Heh. It's one of my pet peeves as well......

</pedant>

Multimedia
Feb 16, 2004, 07:46 PM
Originally posted by jamilecrire
The "unannounced" laptop containing the ATI Radeon 9700 is the new Dell laptop. It's the funniest thing I've ever seen. 2" thick, 55 lbs, runs at 180 degrees F, and is over priced as well.

I wish apple would hurry up and announce the G5 Powerbooks so I can get the 3 month wait until they are in stores over with. I don't think G5 PowerBooks will ship until October. I think anyone who thinks otherwise is fooling themselves. I think putting the 970FX into a PowerBook is a major engineering achievement that still needs another 6 months to accomplish.

I will be pleasantly surprised if it happens any sooner.

rdowns
Feb 16, 2004, 08:01 PM
Originally posted by Snowy_River
If you know one, you know them all...


Just a little bit about typing what you mean. There is a difference between 'know' and 'no'.
This is one of those typos that's a real pet peave for me.

Sorry for going off topic. Every now and then I need to vent about this kind of thing... :)



Posts pointing out other's typos shouldn't contain any of their own. A real pEEve of mine.



oops, someone else beat me to it.(edit)

Analog Kid
Feb 16, 2004, 09:44 PM
Originally posted by a17inchFuture
its not like the G5 powerbook idea is coming out of nowhere: the most significant reason that all of us have cited over the past 8 months as to why they wouldn't get G5s in powerbooks soon was the processors' heat production and energy consumption. With these two problems seemingly fixed, what (besides the bus) do you think needs to be implemented into the design of the PB? I mean, I know very little about what the differences between G5 and G4 will present as far as the other parts, but it would seem to me that they don't have to replace everything in the PB just becasue its a next gen processor. Am I wrong about this, or does Lanbrown actually know about as much as me?

The heat problem has not been "fixed"... Everyone is comparing the 90nm part with the 130nm part and then saying it's good enough of a laptop-- can't do that. You've gotta compare it to the current G4 systems.

The 970FX at 1.4GHz pulls about 50% more power than the current 7447s at 1GHz. Not a big difference, granted, and Apple is actually running them at 1.25GHz in the 15", but if they have a target run time and a heat budget then every watt has to be made up somewhere else...

As someone mentioned the G5 is running a bus that's both wider and faster than the G4. The G5 power specs don't account for this-- chips are spec'd with their outputs floating.

Then there's the memory which is another power hungry beast. The G4s have DDR SDRAM, but they don't really use it-- the G5 does. I wouldn't be surprised if Apple waits for DDR2 to be viable-- it cuts the memory power by 50-60%. That could go a long way towards balancing the power budget. The downside is that DDR2 is currently more expensive, but when did that ever get in the way?

The Xserve shell was modified for more efficient cooling. I'm sure they didn't throw out that extra hard drive on a whim. This in itself doesn't mean they can't close the case to make a laptop, but it's another sign.

I take the 90nm G5 as just another step down the path. We'll get there, but I don't think we're there yet. I'm still guessing second half of the year.

Analog Kid
Feb 16, 2004, 09:52 PM
Originally posted by PBGPowerbook
The powerbook g5 is:
black anodized aluminum
red glowing apple in cover Š la current notebooks
I think the hope is that they can get the power down far enough that the apple glows white again...

Bilba
Feb 16, 2004, 10:00 PM
Originally posted by Analog Kid
I think the hope is that they can get the power down far enough that the apple glows white again...

best post in this thread...
:D

lind0834
Feb 16, 2004, 11:03 PM
"If I had a pretty little peeve, I'd pamper it like it was a queen. You hearin' me, son?"

Lemming
Feb 16, 2004, 11:30 PM
Metals glow in this order, from coolest to hottest: Yellow, orange, red, blue, white.

aswitcher
Feb 16, 2004, 11:42 PM
Originally posted by PBGPowerbook
The powerbook g5 is

black anodized aluminum
red glowing apple in cover Š la current notebooks
grille/mesh on parts of side edges
backlit keyboard white letters on black keys

Comes with optional black anodized metal device known as a "paw" which has two buttons and a black scroll wheel, connecting with bluetooth and replacing the mouse (Its called a paw because Steve will never release a two button mouse!)

luggnutt
Feb 17, 2004, 12:22 AM
Originally posted by Lemming
Metals glow in this order, from coolest to hottest: Yellow, orange, red, blue, white.

This is incorrect. Any metal above absolute zero is always "glowing," radiating energy. The hotter the metal gets, the higher the energy of the radiation. Thus, the first color you see is red, the longest wavelength (lowest energy) of the visible spectrum.

As the metal heats up, higher energy colors are added to it: orange, yellow, green, blue, indigo, violet. The combination of the highest color with the lower ones yields the color we see. Thus, we never see green glowing metal because the metal would also be radiating red to yellow, which all combine and start to turn the visible color to white.

Therefore, the apparent color of heating metal proceeds red, orange, yellow, white. Metals cannot glow blue due to heating.

And, uh, powertune is rad.

Snowy_River
Feb 17, 2004, 12:31 AM
Originally posted by Lanbrown
PowerTune functionality is ...
PowerTune is ...
PowerTune would do ...
PowerTune would cause ...

LanBrown,

I suspect (though I don't know) that you don't know as much about PowerTune as you're claiming to. You may know a lot about power management processes on chips, but do you know that PowerTune is just another power management? Do you know exactly what it does and doesn't do? For the sake of keeping a more productive discussion, would you consider phrasing your statements in terms of what you do know and what you suppose, rather than stating your suppositions as facts?

Snowy_River
Feb 17, 2004, 12:36 AM
Originally posted by luggnutt
If Apple engineering is at all like a typical engineering design system...

You're assuming that Apple is a traditional top-down organization, and hasn't set things up in a more team/project oriented fashion. If they have, then there could be periods of time when there are litterally no engineers working on new PowerBooks.

deputy_doofy
Feb 17, 2004, 12:39 AM
Originally posted by utilizer
I think we're all in for a bigger surprise than the one at last year's WWDC with the "annoucement" of G5s shipping in September.
PowerBook G5s will be annouced for sure this time around; seriously, the iBook was expected to keep the G3 for quite sometime, but look at what happened late in 2003. And when that gets an update, it'll be a 1.42 Ghz G4. The iBook and PB will virtually be the "same" machine. Apple cannot allow for Motorola to continue their stream of broken promises to trickle in. I believe a 1.6 Ghz G5 would suffice alot of users needs, especially with the arrival of Adobe's CS. That thing is sllllooowwww on anything less than that!!!

Anyways though, I'm calling it:
-PowerBooks with G5s at 1.6 Ghz, annouced very soon but shipping in late April; duals by the end of the year. -- 12 in. form factor retains the G4.
-G5 Tower: Dual 1.8 Ghz at low-end and Dual 2.733 Ghz at high end.
-iMac: 1.8 Ghz, maybe 2.0 Ghz in special edition 20 incher (brand it "20 for 20th" (anniversary!)
-iBook: Stuck with the "adequate" G4 7457 supplies from Moto. 'Nuff said!

I underestimated Apple last year -- I think this year, we can anticipate greater surprises!!:D

I think using the iBook's processor shift is a bad example. If I'm reading some older IBM notes correctly (http://www-3.ibm.com/chips/products/powerpc/newsletter/jun2003/newproductfocus.html),
the G3 was running up to 1.1GHz, with a 1mb L2 cache, and a potential bus speed of 200MHz.
While this chip does not have Altivec on it, it would have given (and probably still would) Motorola's G4 a run for its money.
The answer, marketing-wise, is to give the iBook the older G4. That way, it LOOKS like Apple is being nicer to those new buyers, but in reality, it is escaping a marketing nightmare by having to explain the new G3s processing capability vs. the G4.
Just my opinion. :)

Snowy_River
Feb 17, 2004, 12:46 AM
Originally posted by Lanbrown
...People have been talking about G5 PB's for months now...

Do you suppose that could be because people are looking forward to them? I know I am. I'm not going to be in the market for a new PB for another 6 months to a year, but I still really want to see the PB G5 come out sooner rather than later. I love the rumors that point to the possibility that it could be announced tomorrow. Will I be grossly disappointed if it isn't released until Sept.? Not too much. Will I be grossly disappointed if it isn't released until Sept. '05? Yeah, that I'd find a bit harder to take. But, until it does come out, I enjoy reading the news about the technologies that are going to make it more likely that the PB G5 is on its way...

Rend It
Feb 17, 2004, 01:08 AM
Originally posted by luggnutt
This is incorrect. Any metal above absolute zero is always "glowing," radiating energy....

Therefore, the apparent color of heating metal proceeds red, orange, yellow, white. Metals cannot glow blue due to heating.


This is getting beat to death, and it's off topic, but what both of you are trying to refer to is the blackbody spectrum of electromagnetic radiation. At any nonzero temperature, any material will radiate energy in principle. The distribution of the electromagnetic energy flux (intensity of the light) as a function of wavelength (or frequency) is given by the so-called Planck law. For a given temperature, there will be a maximum, or peak, in the distribution. The temperature and peak wavelength are related by what's called Wein's displacement law:
(lambda_peak)*(Temp) = 0.002898[meter][Kelvin]. Square brackets indicate the units involved.

So, obviously the peak may occur at any wavelength, or color, including blue, infrared, ultraviolet, etc. But, because of the shape of the Planck distribution, what is observed is approximately the sum of wavelengths greater than or equal to the peak wavelength. Also, the blackbody law is not unique to metals. It's true for any material, including glasses, semiconductors, or whatever.

idkew
Feb 17, 2004, 01:32 AM
Originally posted by luggnutt
...
Therefore, the apparent color of heating metal proceeds red, orange, yellow, white. Metals cannot glow blue due to heating.

i seem to remember seeing some green glowing metal in the furnace while pouring metal for sculpture. maybe i'm seeing technicolor...

Sol
Feb 17, 2004, 01:37 AM
Would Apple use G5s in the iMacs before it uses them in PowerBooks? I know that such a move would not fit in with their consumer/proffessional product grid but iMac sales are not what they used to be and PowerBooks are selling well, even with G4 processors. Judging from the information posted on this forum it sounds like the G5 is not quite ready for portables yet but it is perfectly suited for desktops. If IBM really can use larger wafers with the smaller process and produce more good parts than the previous generation of 970 CPUs then all the desktops, including the eMac should be using these.

Of course Apple would not update any of their PCs to a new processor without making drastic changes to the case. Personally I miss the transparent plastics and patterns; you read it right, at least one person liked the Blue Dalmation iMac. Having said that, the colours of the iPod Mini's look promising and I would be happy with materials like that for the next generation of the iMac, eMac and full-size iPod.

u2mr2os2
Feb 17, 2004, 05:32 AM
I haven't read up yet on PowerTune, but it seems the original article was saying that what they have created is a single chip that can trade power efficiency for speed as needed in order to be used in a desktop, laptop, or embedded systems.

I take this to mean that you don't need to design a low-power version for a laptop - you just clock it low all the time, and it never uses much power. Certainly, you could utilize the PowerTune features to further optimize power consumption, but you'd not go faster than a certain lmit. The point is that a 970FX in a laptop clocked at 1.4GHz using little power could be the same chip as in a desktop running at 2.5GHz using a lot of power. Both could utilize power savings. It seems this way, you could just produce a bunch of the same chip rather than producing separate lower power versions.

Since the CPU could be rated at a higher clock that is normally used in say a laptop, it is possible that PowerTune could allow you to run the CPU in bursts of higher clock rates using more power, just keeping the average thermal output within the cooling capabilities. This doesn't sound too unusual, except that the CPU in the laptop rather than being limited to 1.4GHZ and using power management to slow down from there, it could be that it is rated to 2GHz or more, and can burst up to there, but to run at a sustained 100%, it would be limited to 1.4GHZ say.

phonic pol
Feb 17, 2004, 05:58 AM
Originally posted by aswitcher
Comes with optional black anodized metal device known as a "paw" which has two buttons and a black scroll wheel, connecting with bluetooth and replacing the mouse (Its called a paw because Steve will never release a two button mouse!) Stupid idea, it'll never catch on!

Rincewind42
Feb 17, 2004, 06:09 AM
Originally posted by u2mr2os2
I take this to mean that you don't need to design a low-power version for a laptop - you just clock it low all the time, and it never uses much power. Certainly, you could utilize the PowerTune features to further optimize power consumption, but you'd not go faster than a certain lmit. The point is that a 970FX in a laptop clocked at 1.4GHz using little power could be the same chip as in a desktop running at 2.5GHz using a lot of power. Both could utilize power savings. It seems this way, you could just produce a bunch of the same chip rather than producing separate lower power versions.

Since the CPU could be rated at a higher clock that is normally used in say a laptop, it is possible that PowerTune could allow you to run the CPU in bursts of higher clock rates using more power, just keeping the average thermal output within the cooling capabilities. This doesn't sound too unusual, except that the CPU in the laptop rather than being limited to 1.4GHZ and using power management to slow down from there, it could be that it is rated to 2GHz or more, and can burst up to there, but to run at a sustained 100%, it would be limited to 1.4GHZ say.

Well, the clock speed that a cpu runs at isn't just a matter of power savings or design, but also of the actual materials used - and unfortunately we still can't produce exact replicas every time yet :D . So regardless of what PowerTune brings to the table for power savings, the chip that goes into your laptop advertised as 1.4Ghz will almost certainly never go over that speed, and truthfully it could be a nightmare if it did. A small fault on the motherboard, or a crash in the OS and suddenly the chip is busy burning a hole in the bottom of your lap at 2.5 Ghz :( .So while your idea has merit (who doesn't want a faster computer!) there are practical issues that would likely prevent it.

Gorbag
Feb 17, 2004, 06:13 AM
Originally posted by idkew
i seem to remember seeing some green glowing metal in the furnace while pouring metal for sculpture. maybe i'm seeing technicolor...

Er,think you were seeing oxidation (burning) of a metal, here, which is a chemical process. For instance, if I recall my O level chemistry (Jeez! 30 years ago. Getting old!) Strontium will burn with a reddish/purple flame. Copper, i think, will give off a green light when burned. These are emission spectra that are telling you about the energy levels in the outer shell of the atoms' electronic configurations. Absolutely nothing to do with Black body radiation which has been much better explained in one of the previous posts than i could.

Could we BE more off-topic???

Rincewind42
Feb 17, 2004, 06:24 AM
Originally posted by Gorbag
Could we BE more off-topic???

Sure! What's the maximum airspeed velocity of an unladen swallow? To which you might ask African or European (http://www.style.org/unladenswallow/)?

And at some point we would have to wonder how it was we got here in the first place...

But to bring things back around... I don't think we'll see a 970FX actually in use until 10.3.3 finalizes. Something tells me that the Xserve and any other new 970FX Macs will require it.

ja0912
Feb 17, 2004, 07:00 AM
Originally posted by Rincewind42
But to bring things back around... I don't think we'll see a 970FX actually in use until 10.3.3 finalizes. Something tells me that the Xserve and any other new 970FX Macs will require it.

I agree. The only thing I can imagine responsible for holding back hardware (PMG5 rB) right now is the OS. I can't imagine Apple withholding the 970fx from Power Macs when they've already announced it for Xserves. Buyers are now holding off.

edit: mistake, spelling

Wendy_Rebecca
Feb 17, 2004, 07:29 AM
Originally posted by Snowy_River
You're assuming that Apple is a traditional top-down organization,

PHHHHHHHTTTTT!!!!!

That sound you hear is the coffee coming out of my nose. Apple IS THE MOST top-down organization ever, with King Jobs at the helm.

Honestly, did you think before you wrote that post?

Lanbrown
Feb 17, 2004, 07:36 AM
Originally posted by Rincewind42
If PowerTune allows the 970FX to turn off the Altivec Units and one of the Integer units during a huge floating point operation, and thus reduce power usage by say 25%, then that translates directly into savings in electricity, and increases the reliability of the entire cluster by making it less likely that any particular system will fail. It will also allow the system to keep the CPU cycling at a higher frequency when the OS would normally reduce the power usage to avoid overheating.

I personally don't see any downside to PowerTune on any hardware.

Edit: forgot the system will cycle down if it is overheating regardless of settings.

There are servers that have been running for years, some with a very high load. Do you know why processors in servers don't clock as high as their desktop counterparts? They are made differently. They are designed thicker to withstand the heat and run at 100% all day long without a complaint.

Turning features on and off allow them to cool and then when they turn on and they start being pushed to the limit; they heat up. This causes parts to contract and expand, which is actually bad for electronic and their electrical joints.

Overheating in a server? The fans are more then capable of cooling the system when all CPU's are running at 100%. A few of the fans can even fail and it will still keep it below the minimum temperature threshold. Typically ever PS will have it's own fan. Then the main cage area will have several fans and the board layout is designed with optimal cooling in mind. Like putting the memory in line with the airflow. This not only reduces noise, but also reduces heat zones within the chassis by the lack of airflow.

You keep mentioning overheating. ANY properly designed system will not overheat even when the system is being pushed to its limit. If it does, then the manufacturer took the cheap way out and didn't do proper engineering/testing. If a fan fails, then yes, system performance could be impacted, but not in a properly designed server. Not even a PS failure should impact system performance. That's why the good servers have N+1.

If you have a part of a CPU failing, it needs to be replaced. It was obviously not manufactured correctly and will only continue to fail.

Lanbrown
Feb 17, 2004, 07:47 AM
Originally posted by stingerman
You don't even know what PowerTune is. You can't have a 970FX without it! PowerTune is a collection of technologies: Voltage Islands; Voltage Slewing inline with Frequency Slewing; eFuse; new protected and user instructions to allow the OS and Apps to work with PowerTune.

You assume that if Apple chooses to turn off automatic slewing (which is a user controllable feature), all these powertune futures are not available. Your wrong. 970FX is designed usign Voltage Islands which allows different parts of the processor to work at a lower voltage than the core or caches. So Powertune is part of the 970FX fabric, you can't have one without the other.

Software API wise, it allows developers to tell the processor that we need max performance now, and now we don't. IBM gives the example of DVD playback, the processor switches up in under a uS, decodes the frame, switches down, then back up in time to deliver 60FPS. In the process it uses 25% of the power.

So basically bring your knowledge up-to-date, PowerTune is not speed stepping, it is a completely new technology. And if you think super clusters are not drooling over this feature of the 970FX, you are out of your mind. Do you have any idea of the electrical costs associated with running a huge cluster. Rarely does an operation use 100% of each processor. The power savings would be a huge selling factor to small and large firms. I know, I run a couple of Rack full of Dell 2U's and power costs make up a big part of my monthly data center costs, even when the servers are hardly being hit, like after 10PM or so.

PowerTune is a collection of other technologies. Ti showed a product last year, early last year that it could turn off components in a chip that wasn't in use. Hmm, some of the features in PowerTune. I wonder who thought of that first, it wasn't IBM, and it might not have even been Ti.

A cluster/server can never fully utilize 100% of the processors, but there are systems that can get over 98% processor efficiency, it involves hardware and software. Cray, IBM, Sun, Fujitsu and others all know how to make the most out of each processor. Some clusters use Ethernet as a cross-connect, which is very inefficient.

BTW, PowerTune can control the clock speed as well.

What you fail to realize is that just because something is there, doesn't mean it has to be used.

I feel sorry for you, having to use Dells.

Lanbrown
Feb 17, 2004, 07:49 AM
Originally posted by stingerman
It's not like that with powertune. The system doesn't turn off, but actually "naps" not "sleeps" portions of the processor not being used. However there is a trickle of power going down those circuits that allows them to wake right up, not out of a deep sleep but out of a light nap, so to speak. It happens so quick that youwill not notice a delay as it takes only uSec to come fully awake.

It will take at least one clock cycle to get that portion of the chip ready, now won't it?

formasfunction
Feb 17, 2004, 07:52 AM
Call me crazy, and its possible that someone else already pointed this out in the pages and pages of posts, but has anyone taken a look at the new Xserve? Did you notice the huge vents on the front of it? Small animals could be sucked inside this thing if they scurried too close. And this is using the same processor that everyone is proposing for the PB. Granted, these things are designed to run balls to the wall 24/7, and some come with dual processors, two things that a PB won't have to deal with, but so did the previous Xserves and there were no signs of need for extreme cooling. Even though IBM has said that this chip COULD be used for laptops, I don't think that it WILL be used by apple until less extreme measures are needed, even taking into account the power management of the chip. I'll eat my hat and sell my G4 PB if we see G5PBs any sooner than 7 months.

Lanbrown
Feb 17, 2004, 07:54 AM
Originally posted by MacEyeDoc
The article states that the chip could be used from PC's to laptops to servers. Apple appears to be using it in the Xserve, so what will they use it in next? I don't think I've heard any Apple pronouncements about when the next laptops or PowerMacs are coming. They could be working on one or the other, or both, but what would give them the biggest splash? They won't be selling PowerMacs for clusters anymore, so maybe that's why they came out with the Xserve first. (And Steve did say in his MacWorld SF keynote that he knew he was pissing off a thousand customers by diverting the first PowerMacs to Virginia Tech, but he didn't care). So if they have plans to sell several thousand Xserves to academic or government installations, and all the new PowerMacs and PowerBooks use the same chip, does that mean that we, the common consumers, will be waiting again? I'm sure you guys have an opinion . . . .

An update to the PM G5 would not be big news, just an update. A PB would be big news as it's a new product. A G5 iMac would be big news as it too would be a new product. Apple could release updated PM's without a big event. They would release an all new product at a big event. So maybe PB's or iMacs will be unveiled at the next event.

Lanbrown
Feb 17, 2004, 08:03 AM
Originally posted by Snowy_River
Do you suppose that could be because people are looking forward to them? I know I am. I'm not going to be in the market for a new PB for another 6 months to a year, but I still really want to see the PB G5 come out sooner rather than later. I love the rumors that point to the possibility that it could be announced tomorrow. Will I be grossly disappointed if it isn't released until Sept.? Not too much. Will I be grossly disappointed if it isn't released until Sept. '05? Yeah, that I'd find a bit harder to take. But, until it does come out, I enjoy reading the news about the technologies that are going to make it more likely that the PB G5 is on its way...

Yes, but every time a small piece of news is released, what is the first thing that is said; PB's anytime now. That has been said for MONTHS. To find the earliest date when we will see them is to find the next big event that Apple will be at. PM's updates can happen at anytime. Shipping times are a different story though.

billyboy
Feb 17, 2004, 08:06 AM
Before the G5 hits the Powerbook scene, Id like to see Apple push the G4 PB, which is already their cash cow, and use its undoubted appeal to attract new users to the Mac platform, as well as add to the buzz over the next generation of Apple's professional notebook.

Lanbrown
Feb 17, 2004, 08:11 AM
Originally posted by formasfunction
Call me crazy, and its possible that someone else already pointed this out in the pages and pages of posts, but has anyone taken a look at the new Xserve? Did you notice the huge vents on the front of it? Small animals could be sucked inside this thing if they scurried too close. And this is using the same processor that everyone is proposing for the PB. Granted, these things are designed to run balls to the wall 24/7, and some come with dual processors, two things that a PB won't have to deal with, but so did the previous Xserves and there were no signs of need for extreme cooling. Even though IBM has said that this chip COULD be used for laptops, I don't think that it WILL be used by apple until less extreme measures are needed, even taking into account the power management of the chip. I'll eat my hat and sell my G4 PB if we see G5PBs any sooner than 7 months.

Define seeing them before 7 months? I would expect to see them announced in June, shipping sometime later though. If you are talking of them shipping, then 7 months would be a good guess.

Others face it. Either Apple has buttoned down the leaks or there is nothing to leak. There were quite a few leaks about the G5 PM before it was released, even Apple let the cat out of the bag. There are some companies that nothing is leaked from and when there is a leak, it is usually done by the company to spoil an announcement from a competitor. I can't see Apple to going from quite a few leaks to none in a short amount of time.

u2mr2os2
Feb 17, 2004, 09:39 AM
Originally posted by Lanbrown
It will take at least one clock cycle to get that portion of the chip ready, now won't it?

But given the instruction pipeline, could it know it needs to wake up that portion before the instruction gets executed so there is no power-up delay?

Rincewind42
Feb 17, 2004, 09:54 AM
Originally posted by u2mr2os2
But given the instruction pipeline, could it know it needs to wake up that portion before the instruction gets executed so there is no power-up delay?

The CPU should be able to determine that it needs to wake up a CPU unit at least 6 cycles before it is used. The 6 cycles is how long it takes for an opcode to get through dispatch on the PPC970. It takes at least 3 cycles (likely 4 or more) for an instruction to come in from fetch to reach decode, so it can be working on discovering what units are needed even sooner. And the CPU can determine what parts of an instruction unit can be turned down/off by watching instructions as they flow through the unit.

I have no idea how much of this information is used in PowerTune, so take it for what it is.

Lanbrown
Feb 17, 2004, 10:09 AM
Originally posted by Rincewind42
The CPU should be able to determine that it needs to wake up a CPU unit at least 6 cycles before it is used. The 6 cycles is how long it takes for an opcode to get through dispatch on the PPC970. It takes at least 3 cycles (likely 4 or more) for an instruction to come in from fetch to reach decode, so it can be working on discovering what units are needed even sooner. And the CPU can determine what parts of an instruction unit can be turned down/off by watching instructions as they flow through the unit.

I have no idea how much of this information is used in PowerTune, so take it for what it is.

In theory yes, but processors can make bad decisions and that's where the longer pipelines start to lose their momentum. With processors mainly waiting for data, shutting down any part of the chip or reducing the clock speed is a bad idea. It's a great idea for portables and desktops though; the effects are minimal on them.

spankalee
Feb 17, 2004, 10:41 AM
Originally posted by formasfunction
Call me crazy, and its possible that someone else already pointed this out in the pages and pages of posts, but has anyone taken a look at the new Xserve? Did you notice the huge vents on the front of it? Small animals could be sucked inside this thing if they scurried too close. And this is using the same processor that everyone is proposing for the PB. Granted, these things are designed to run balls to the wall 24/7, and some come with dual processors, two things that a PB won't have to deal with, but so did the previous Xserves and there were no signs of need for extreme cooling. Even though IBM has said that this chip COULD be used for laptops, I don't think that it WILL be used by apple until less extreme measures are needed, even taking into account the power management of the chip. I'll eat my hat and sell my G4 PB if we see G5PBs any sooner than 7 months.

Well, not only will a PowerBook not be dual processor, but it's likely to run at speeds of 1.4 to 1.6 Ghz and have more aggressive power management settings. The 970fx uses a lot less power at 1.4Ghz than 2Ghz.

I expect to see the case be designed for better cooling as well. There might be some more obvious vents.

Another thing not mentioned yet is that the bus doesn't need to run an half the processor speed. There's options for 1/4, and maybe the 970fx has more multipliers. Having a lower bus speed is common in notebooks and would lower heat and save battery life.

Rincewind42
Feb 17, 2004, 10:53 AM
Originally posted by Lanbrown
In theory yes, but processors can make bad decisions and that's where the longer pipelines start to lose their momentum. With processors mainly waiting for data, shutting down any part of the chip or reducing the clock speed is a bad idea. It's a great idea for portables and desktops though; the effects are minimal on them.

Well, that depends. If the time it takes to wake up the unit is 6 cycles or less, then it will be completely hidden by the dispatch time. If the pipeline needs to be flushed, then it will take even more time and the CPU should by then know what will be executed when. And if it is manly waiting for instructions to come from main memory, then it really doesn't matter what gets shutdown when because the latency from main memory through dispatch to instruction unit is on the order of a couple dozen cycles. PC3200 ram takes at least 5ns (10 cycles @ 2Ghz) to send the critical word to the CPU, the 970 looks at instructions 8 at a time so you need 2 RAM reads on a PMG5 which requires 7.5ns (15 cycles @ 2Ghz) to send from RAM to controller. Then tack on 3 cycles for fetch, 6 cycles for decode and you have 24 cycles from RAM to execution. And this is when everything is optimal. There is also bank switching in the RAM, translation on the system controller, packet transmission overhead between the controller and the CPU, and any number of stages that can be introduced between fetch and decode due to resource allocation issues.

Snowy_River
Feb 17, 2004, 12:25 PM
Originally posted by Wendy_Rebecca
PHHHHHHHTTTTT!!!!!

That sound you hear is the coffee coming out of my nose. Apple IS THE MOST top-down organization ever, with King Jobs at the helm.

Honestly, did you think before you wrote that post?

Just because the organization as a whole is top-down doesn't mean that the project management end of things is. In fact, given that many studies have shown that rigidly top-down organizations are the slowest to innovate, the fact that Apple is regarded as a market leader in innovation would suggest that it's not all that rigidly top-down.

I have never worked for Apple's engineering department, nor have I ever spoken with anyone who has. I have no direct knowledge of what Apple's engineering structure is, and can only speak from external (distant) observation.

tiktokfx
Feb 17, 2004, 01:32 PM
Originally posted by Lanbrown
Big deal? If you go with one thousandth of a second and 2200 processors, that is 2.2 seconds.

Actually if you go with .001 of a second and 2200 processors, that's .001 seconds of real time. 2.2 seconds of net computational time.

Also, it seems rather odd to randomly assume times like .001, which is 7 orders of magnitude greater than the cycle of a 2.5 GHz processor. A much more likely estimate of the time it takes to modify the speed of a processor would be the depth of the pipeline times the cycle time, which would be about .0000000064 seconds.

This also assumes that the tuning component of the processor is incredibly stupid and tries to scale back every chance it gets, instead of monitoring itself and seeing when it's useful to scale back.

As soon as a machine hits a point where it needs high availability of speed, it's going to remain there for a reasonable period. Truly high-activity servers will simply never scale back, if they don't have to. And that fraction of a second is not going to amount to much.

aswitcher
Feb 17, 2004, 11:58 PM
Originally posted by spankalee
Well, not only will a PowerBook not be dual processor, but it's likely to run at speeds of 1.4 to 1.6 Ghz and have more aggressive power management settings. The 970fx uses a lot less power at 1.4Ghz than 2Ghz.

I expect to see the case be designed for better cooling as well. There might be some more obvious vents.

Another thing not mentioned yet is that the bus doesn't need to run an half the processor speed. There's options for 1/4, and maybe the 970fx has more multipliers. Having a lower bus speed is common in notebooks and would lower heat and save battery life.

As long as the bus is significantly better than the current one which I have read in several places is the real bottleneck of the current design. Would a half gig bus be feasible?
:confused:

Dave the Great
Feb 18, 2004, 01:01 AM
<QUOTE>Originally posted by formasfunction
Call me crazy, and its possible that someone else already pointed this out in the pages and pages of posts, but has anyone taken a look at the new Xserve? Did you notice the huge vents on the front of it? Small animals could be sucked inside this thing if they scurried too close. And this is using the same processor that everyone is proposing for the PB. Granted, these things are designed to run balls to the wall 24/7, and some come with dual processors, two things that a PB won't have to deal with, but so did the previous Xserves and there were no signs of need for extreme cooling. Even though IBM has said that this chip COULD be used for laptops, I don't think that it WILL be used by apple until less extreme measures are needed, even taking into account the power management of the chip. I'll eat my hat and sell my G4 PB if we see G5PBs any sooner than 7 months.</QUOTE>

I don't think that is exactly a fair comparison - comparing the xserve to a pb.
The xserve, in addition to being able to use dual processors, also has 3 available drive bays for up to 750 GB, 2 full size PCI slots, up to 8GB of mem, more I/O connectivity, and is made so you can stack them on top of each other. This is why they have the vents.

I hope the G5 PBs will soon be launched!

aswitcher
Feb 18, 2004, 02:37 AM
Originally posted by Dave the Great
<QUOTE>Originally posted by formasfunction
Call me crazy, and its possible that someone else already pointed this out in the pages and pages of posts, but has anyone taken a look at the new Xserve? Did you notice the huge vents on the front of it? Small animals could be sucked inside this thing if they scurried too close. And this is using the same processor that everyone is proposing for the PB. Granted, these things are designed to run balls to the wall 24/7, and some come with dual processors, two things that a PB won't have to deal with, but so did the previous Xserves and there were no signs of need for extreme cooling. Even though IBM has said that this chip COULD be used for laptops, I don't think that it WILL be used by apple until less extreme measures are needed, even taking into account the power management of the chip. I'll eat my hat and sell my G4 PB if we see G5PBs any sooner than 7 months.</QUOTE>

I don't think that is exactly a fair comparison - comparing the xserve to a pb.
The xserve, in addition to being able to use dual processors, also has 3 available drive bays for up to 750 GB, 2 full size PCI slots, up to 8GB of mem, more I/O connectivity, and is made so you can stack them on top of each other. This is why they have the vents.

I hope the G5 PBs will soon be launched!

In desperation you could always strap a 17" screen, keyboard and a decent battery to an Xserve to get your dream X5 now...as long as your strong and fit :p


Oh, just don't plan to use it naked, because then you'll have more to worry about than small furry creatures getting sucked into those huge vents :D

Analog Kid
Feb 18, 2004, 04:43 AM
Originally posted by spankalee
Another thing not mentioned yet is that the bus doesn't need to run an half the processor speed. There's options for 1/4, and maybe the 970fx has more multipliers. Having a lower bus speed is common in notebooks and would lower heat and save battery life.

Which further degrades it's performance advantage against the G4...

Some folks seem to want a PB G5 just to have a G5 in a PowerBook. I don't know what Apple's thinking on this is, but I'm guessing they want a real advance with the new design-- not just an incremented G...

I still argue that the 90nm part is one piece of the puzzle, and we have to wait for the rest to fall into place. System controller is probably ready-- but I'd guess it's higher power than the G4 equivalent as well because of bus speeds.

Something needs to have a net power reduction to balance the power increases. PowerTune might help with the wind sprints, but I'm still guessing (note that-- guessing) that RAM is the gating item.

Analog Kid
Feb 18, 2004, 04:56 AM
Originally posted by Rincewind42
Well, that depends. If the time it takes to wake up the unit is 6 cycles or less, then it will be completely hidden by the dispatch time. If the pipeline needs to be flushed, then it will take even more time and the CPU should by then know what will be executed when. And if it is manly waiting for instructions to come from main memory, then it really doesn't matter what gets shutdown when because the latency from main memory through dispatch to instruction unit is on the order of a couple dozen cycles. PC3200 ram takes at least 5ns (10 cycles @ 2Ghz) to send the critical word to the CPU, the 970 looks at instructions 8 at a time so you need 2 RAM reads on a PMG5 which requires 7.5ns (15 cycles @ 2Ghz) to send from RAM to controller. Then tack on 3 cycles for fetch, 6 cycles for decode and you have 24 cycles from RAM to execution. And this is when everything is optimal. There is also bank switching in the RAM, translation on the system controller, packet transmission overhead between the controller and the CPU, and any number of stages that can be introduced between fetch and decode due to resource allocation issues.

Not much you can do in a few ns other than stop the clock... That was an effective power saver with larger feature sizes, but if you stop the clock at 90nm you're still leaking a lot of current. Not to mention that clock gating has been around forever and hardly deserves a new name like PowerTune.

Granted Intel does that all the time, but I have to think IBM has more class... (Not to mention they have a research facility that really does innovate).

There's been talk of "voltage islands" which I suspect means voltage gating. Turn the power off to parts that aren't being used so they don't draw dynamic current but also don't leak.

Intel has been scaling voltage with frequency since the early XScales, which is clever. They moved the same idea into Centrino, I think.

Problem with voltage scaling is you can't do it in nanoseconds... This is a 25W part which is pulling something like 12-15 amps. I can't imagine they can ramp that kind of current in 6 cycles. Lead and trace inductance will slow it down. IR drop in the chip would be too high...

I might be wrong on this-- the Intel chips pull a whole lot more current (4x?), and I know they have insane slew rate requirements. Maybe it can be done...

I'm starting to agree with the camp that says this will be used in notebooks that typically run one task at the speed of the user, and not used in servers/workstations that are going full tilt for days.

Guess we'll find out soon enough! =)

Rincewind42
Feb 18, 2004, 06:33 AM
Originally posted by Analog Kid
Which further degrades it's performance advantage against the G4...

Yes, a G5 on a 500Mhz bus will not perform as well as a G5 on a 1Ghz bus. But it's a laptop - there will be compromises made. Unless Apple has discovered how to make dimensional pockets it's going to be really hard for them to put in more than 2 RAM slots (except maybe on the 17"). And if there are only 2 RAM slots, then you can't really require that RAM be installed in pairs, so your RAM (Even PC3200) wouldn't be able to keep up with a 500Mhz bus, let alone a 1Ghz one. At that point, the only advantage to a higher bus is bandwidth to other resources (which isn't a bad thing, but doesn't have as big a performance increase relative to RAM).

So my prediction remains a bottom end of 1.5Ghz/500Mhz bus. We'll see what we got when we get it :D.

Dave the Great
Feb 18, 2004, 12:05 PM
Originally posted by aswitcher
In desperation you could always strap a 17" screen, keyboard and a decent battery to an Xserve to get your dream X5 now...as long as your strong and fit :p


Why do you think I have been eating all that spinach and steak? :)


Oh, just don't plan to use it naked, because then you'll have more to worry about than small furry creatures getting sucked into those huge vents :D

Oh man! I didn't think of that. Looks like I will have to wait for an actual G5 PB! :)

spankalee
Feb 18, 2004, 02:14 PM
Originally posted by Analog Kid
Which further degrades it's performance advantage against the G4...


I don't get it...

Are you saying that a 1/4 multiplier is worse than the 1/8 multiplier used in the G4 powerbooks?


Some folks seem to want a PB G5 just to have a G5 in a PowerBook.


Well, because the G5 is a much better chip. I run my PowerBook to the max regularly, and I desperately need more horsepower. The G5 seems to be the only option for a drastic increase in performance, therefore I want a G5 PowerBook.

Simple, right?

aswitcher
Feb 18, 2004, 11:25 PM
Originally posted by Rincewind42
Yes, a G5 on a 500Mhz bus will not perform as well as a G5 on a 1Ghz bus. But it's a laptop - there will be compromises made. Unless Apple has discovered how to make dimensional pockets it's going to be really hard for them to put in more than 2 RAM slots (except maybe on the 17"). And if there are only 2 RAM slots, then you can't really require that RAM be installed in pairs, so your RAM (Even PC3200) wouldn't be able to keep up with a 500Mhz bus, let alone a 1Ghz one. At that point, the only advantage to a higher bus is bandwidth to other resources (which isn't a bad thing, but doesn't have as big a performance increase relative to RAM).

So my prediction remains a bottom end of 1.5Ghz/500Mhz bus. We'll see what we got when we get it :D.

4 Ram slots would be sweet... Could go for a full two gig without breaking the bank... Or just a gig with cheap 256 sticks...

500Mhz bus maybe overkill then, perhaps it will be 400mhz like the Ram to save on heat and power but not to hold things up when its needed

Rincewind42
Feb 19, 2004, 08:41 AM
Originally posted by aswitcher
4 Ram slots would be sweet... Could go for a full two gig without breaking the bank... Or just a gig with cheap 256 sticks...

500Mhz bus maybe overkill then, perhaps it will be 400mhz like the Ram to save on heat and power but not to hold things up when its needed

I don't know about 256 sticks being all that cheap really - from what I've seen you save only about $10 with a pair of 256s vs a pair of 512s, and that's only in PC2700 SO-DIMMs, standard PC3200 show no difference at all.

As for the bus, a 400Mhz would require 4x on 1.6 rather than 3x on 1.5. Also you would be slower than the RAM at that point (since the G5 bus has an overhead of about 10%). And if Apple does find a way to provide 4 slots, then slot pairing becomes more likely and we could see a 2x multiplier (although likely on slower RAM like PC2100 or PC2700).