[Eug]

Well, the G5 2.0 Xserve launch included one fact that some have overlooked. It uses 90 nm G5s, not the 130 nm G5s in the Power Macs.

That means IBM's 90 nm process is going pretty well. Thus, it's possible that G5 PowerBooks might get announced sooner rather than later. Mind you, if they get announced sooner, I may just wait for a revision B G5 PowerBook.

If they get announced later (like late summer), then I may be more tempted to get the first iteration of the G5 PowerBook.

I'm not counting on a February G5 PowerBook announcement, but I hope it happens. It'd sure be nice to get a new Rev. B G5 PowerBook for Xmas.

[tvfollower]

I was thinking the very same thing...

At first, I thought they keynote was pretty stupid this time, until I realized the importance of the G5 Xserve!

I mean, screw these iPod minis...all hail PowerBook G5!

I'm counting on a September release, though...February...oh man...I really do hope you're dream comes true...

[southtdi]

I thought they were going to squeeze the size down smaller than 90 nm before the G5 moved to the powerbook.

Also, remember he said the cooling in the 1U Xserve case was a real challenge to overcome. That is thicker than the PB case so that will make it even harder to put the G5 in the PB right now.

I would just say be patient. I beleive we will have it this year but most likely not until next fall.

[Eug]

Originally posted by southtdi:
I thought they were going to squeeze the size down smaller than 90 nm before the G5 moved to the powerbook.

Also, remember he said the cooling in the 1U Xserve case was a real challenge to overcome. That is thicker than the PB case so that will make it even harder to put the G5 in the PB right now.

I would just say be patient. I beleive we will have it this year but most likely not until next fall.

I disagree. Nobody said that it must be 65 nm or whatever before it makes it into a PowerBook. And it's not as if we're getting 65 nm this year anyway.

Remember, we aren't looking for desktop 2.0 GHz G5s (130 or 90 nm) here. We are looking for G5 embedded-oriented processors that will do 1.5 GHz, at a lower voltage. ie. They will be significantly lower power chips, and would be viable in a PowerBook design.

The only questions in my mind are how many PowerBook revisions there will be in 2004, and if there are two revisions, what that first 2004 revision will be: G4 or G5.

I do agree though, that it will likely be a G5 later in the year though. A G4 refresh for the PowerBook just doesn't seem likely to me.

[workerbee]

Originally posted by Eug:
Thus, it's possible that G5 PowerBooks might get announced sooner rather than later.

Oh do I wish it were so... but did you see the fans on the thing?!

[Eug]

Originally posted by workerbee:
Oh do I wish it were so... but did you see the fans on the thing?!

Compare the fans and heatsinks on the G4 1.42 Power Mac and the G4 1.33 PowerBook.

Anyways, one good thing is that by the time G5 PowerBook is released, VPC 7.0 will be out. I rarely use VPC, but it's always convenient to have as a backup for unusual situations.

[workerbee]

Originally posted by Eug:
Compare the fans and heatsinks on the G4 1.42 Power Mac and the G4 1.33 PowerBook.

Stop it, will you? I'm trying to keep a cool head here, and you mess with it by implanting wild ideas�

[OperaMan]

I think it's a step, but I don't see G5 PB's soon, unless you wanna use it as a laptop/leaf blower.

[sillydog701]

I think they will have a new cooling technology. Something totally "unheard of". Fans are noisy.

[maceye]

Wasn't there a rumor a few months ago about Apple investing in a new type of liquid cooling? I think it was something to do with ultra-small tubes embedded in the heatsink (or something like that)? I remember the cooling system was supposed to be super-efficent, especially compared with existing liquid cooling solutions. If those rumors were true then maybe Apple will be able to combine the new cooling tech with the new 90nm G5s... which would mean G5 Powerbooks soon-ish

[fizzlemynizzle]

my guess is a liquid cooling system based on convection and fans. Liquid will be circulated through tubing that passes over or through heat sinks. As the heat from the processor heats the tubing, the liquid evaporates (or at least circulates due to nonuniform temperature) causing it to be drawn through the tube, past the fans and possibly another heatsink that acts as a radiator. The expansion and contraction of the liquid allows it to circulate without the need for a pump. The liquid they use will have to be benign to the system components as well as to people if a leak develops. If a leak does develop hopefully the CPU will automatically restrict itself and the fan(s) will spin up at an increased RPM to provide emergency cooling.

Sega did something like this with the Dreamcast game console in 1999. Some versions were simply air cooled and others used liquid convection. It worked very well considering the Dreamcast was a very compact box.

[PBAddict]

Originally posted by OperaMan:
I think it's a step, but I don't see G5 PB's soon, unless you wanna use it as a laptop/leaf blower.

I agree... I Think it is going to be 2005. I would not buy a Rev 1 of a G5 powerpook. So realistically 2nd half 2005, for a model I would buy.

[SEkker]

I still think we're going to see dual G4 PBs (at least in PB17 form factors) before G5s. This buys Apple another 6-9 months for G5PB development, which is going to be needed due to the requirements for completely new motherboard design.

I would not WANT a revA G5 PB if it were sold in 2004 -- Apple has been spinning its wheels just dealing with the white spot issues with the PB15, and the motherboard build problems on the new iBooks. And these are based on motherboard and other designs Apple is more familar with.

A dual, 1.25 GHz G4 PB17 should be low power and give a nice speed boost, even over the single 1.3 GHz G4 model now out.

IBM may be ready, but Apple is not, to release viable G5 PBs at this time.

[David Hagan]

Did it ever occur to anyone that the aluminum PowerBooks released in the past year are stop-gap measures until the G5 is ready? As I have said so many times before, Apple was working with IBM long before the G5 was publicly announced for the Power Macs. What makes anyone think that the desktop version of this processor was all they were collaborating on during that time? It would seem to me -- Jobs the visionary that he is and the fact that he has pointed this trend out at his keynotes -- that laptops are increasingly becoming more and more popular. So I reiterate, what makes anyone think that Apple hasn't also been hard at work on a PowerBook G5? After all, it took them 7 months to come out with an xServe. So...there is hope.

[Eug]

Yeah, I think it's pretty naive to think that Apple/IBM weren't working right from the start with a low power G5 in mind.

Remember, Apple sells a LOT of laptops and it is a key part of their stategy. Ever since the beginning, laptop chip issues were high on their list of things to tackle. Furthermore, IBM wants to expand its embedded market beyond just G3s, and a low power G5 is the perfect chip for this.

Dual G4 PowerBooks? While technically feasible, it will never happen. It is expensive, takes a lot space (and thus weight), and thus overall it offers little benefit of going to the G5 directly. In fact, I suspect it'd be HARDER to do a dual G4 than it would be to do a single G5. And indeed, a single 1.6 GHz G5 90 nm would use LESS power than a dual 1.25 GHz G4.

Oh and don't forget that The Steve has already said G5 PowerBooks are coming by the end of 2004. He just didn't say exactly when in 2004.

[vsurfer]

Originally posted by Eug:
Yeah, I think it's pretty naive to think that Apple/IBM weren't working right from the start with a low power G5 in mind.

Remember, Apple sells a LOT of laptops and it is a key part of their stategy. Ever since the beginning, laptop chip issues were high on their list of things to tackle. Furthermore, IBM wants to expand its embedded market beyond just G3s, and a low power G5 is the perfect chip for this.

That twinkle in his eye, knowing look, and promise of great things to come this year had me immediately thinking of G5 powerbooks by December. Seeing a G5 in a 1U space only months after the release of the G5 Tower had me thinking they'd achieve size reduction and solve cooling by December too, if not sooner.

[cdhostage]

Personally, I think Apple's stressing thinness of laptops far too much in the recent PowerBooks. I'm on a 15Al right now, and the sheer area of the screen is wonderful. I've played with the big ones but they weren't for me. What I'm getting at is, the things are so big in height and width, it would be perfectly reasonable to increase depth and weight in order to accomodate a reasonable battery and cooling setup for the hot G5.

If Apple can't make a useful cooling setup fit into the current slim 12" and 15" forms, they can definitely make the first PowerBook G5 only the largest size for video production and desktop-replacement purposes. The battery should be as generous as possible instead of made to save space/weight, as my 15Al's seems to be.

[Eug]

I won't buy a 15" laptop over 6 lbs. I won't buy any laptop over 7 lbs.

[Pierre B.]

Now can someone explain me this simple thing? How is possible that Voodoo PC sells 64 bit notebooks featuring 3 real hours on battery, and Apple cannot do that with the G5, especially the new one? Perhaps the restrictions by the physical dimensions of the powerbooks? Or are the Athlons so much power effective?

[fizzlemynizzle]

Originally posted by Pierre B.:
Now can someone explain me this simple thing? How is possible that Voodoo PC sells 64 bit notebooks featuring 3 real hours on battery, and Apple cannot do that with the G5, especially the new one? Perhaps the restrictions by the physical dimensions of the powerbooks? Or are the Athlons so much power effective?

because voodoo PC is a very small volume assembler of other companies' hardware. The difference between a Voodoo and a Powerbook is the difference between a fairly well made kit car and a mass produced luxury car. Apple meets much higher quality standards.

[Eug]

I have assembed some info on power utilization and die sizes on the various G4 and G5 chips out there, in this post.

Some relevant points about an iMac G5 and a PowerBook G5 include the fact that even at 130 nm, the G5 only needs 19 Watts at 1.2 GHz, and 19 Watts typical (assuming the same measurement guidelines) is very similar to the G4 1.33 GHz found in the 17" PowerBook.

Now nobody wants to drop back to 1.2 GHz (even if it is a G5) for the top-of-the-line laptop, but with the die shrink and potential voltage changes alone I suspect that 1.6 GHz is very feasible.

If this low power 1.6 GHz number is true, then it would be perfect for both an iMac G5 and PowerBook G5 refresh, at around 1.2 GHz for the low end, and 1.6 GHz at the high end.

And that's not even considering any other potential power enhancements.

[jasong]

Oh do I wish it were so... but did you see the fans on the thing?!

Servers are supposed to have heavy duty fans, high power, fast fans that move a ton of air through the machine. They need that air flow to deal with 2 internal power supplies (the bulk of the PowerBooks power supply is external) 2 processors (to the PowerBooks one) that may run full tilt all the time (I know we like to think our PowerBooks run all the time, but that's just not the case, the X-Serve is up and running 24/7 pumping out data constantly to all your machines). The X-Serve has 4 high speed hard drives that are again running all the time (vs 1 in the PowerBook that is used as little as possible). X-Serves are designed to be grouped in tight quarters with other hot servers, your PowerBook usually has a fair amount of breathing room in a relatively cool environment.

So in short, don't take the X-Serves (or even the G5s) cooling system as a requirement to run a G5 processor. It won't happen next week, but we will see 64 G5/6 processor in a PowerBook sometime in the next 12 months.

-- Jason

[Pierre B.]

Originally posted by Eug:

If this low power 1.6 GHz number is true, then it would be perfect for both an iMac G5 and PowerBook G5 refresh, at around 1.2 GHz for the low end, and 1.6 GHz at the high end.

I still find hard to believe that Apple will use the current (90 nm) G5 in the Powerbooks. The reason is that even a 1.6 GHz @ 90 nm G5, can barely be used in a Powerbook, according to current evidence. And then, what next? How they will update the processor to higher frequencies?

No, I think that there will be a new family of 64-bit processors specifically designed for mobile usage. Remember, back in the introduction of the G5, IBM talked about new PPC processors, intended for high performance embedded applications and featuring multi-GHz frequencies and SIMD enhancements. But while, usually, IBM is rather chatty about new technologies, since then, the thing remained totally silent. Guess why...

[issa]

Originally posted by Pierre B.:
The reason is that even a 1.6 GHz @ 90 nm G5, can barely be used in a Powerbook, according to current evidence.

Would you please share this "evidence" with the rest of us? Links to power specifications, etc. would be most appreciated. (I can't seem to find any...)

Thanks!

[Eug]

Originally posted by Pierre B.:
I still find hard to believe that Apple will use the current (90 nm) G5 in the Powerbooks. The reason is that even a 1.6 GHz @ 90 nm G5, can barely be used in a Powerbook, according to current evidence. And then, what next? How they will update the processor to higher frequencies?

No, I think that there will be a new family of 64-bit processors specifically designed for mobile usage. Remember, back in the introduction of the G5, IBM talked about new PPC processors, intended for high performance embedded applications and featuring multi-GHz frequencies and SIMD enhancements. But while, usually, IBM is rather chatty about new technologies, since then, the thing remained totally silent. Guess why...

PowerTune may be part of what you're looking for, and the details are to be presented on Feb. 16.

"A 64b PowerPC microprocessor is introduced in 130nm and redesigned in 90nm SOI technology. PowerPC 970 implements a SIMD instruction set with 512kB L2 cache. It runs at 2.0GHz with a 1.0GHz bus in 130nm. The 90nm design features PowerTune for rapid frequency and power scaling and electronic fuses."

[Pierre B.]

Originally posted by issa:
Would you please share this "evidence" with the rest of us? Links to power specifications, etc. would be most appreciated. (I can't seem to find any...)

Thanks!

Eug has summarised most if not all of the publicly available specifications. With a G5 130 nm 1.8 GHz running at 47 W, and the 1.2 GHz running at 19 W, you cannot expect the 1.6 GHz version to run at less than 38-40W. Now, if we are optimists, we could assume that the transition 130 nm --> 90 nm will cut power consumption by a 30-40%. We still talk about 25 W typical power or more. That PowerTune thing may indeed be the answer. I would not hold my breath though.

[issa]

Pierre B.,

Thanks, anyway. The math and extrapolations are all well and understandable; but I was hoping to see actual manufacturer's specs to draw my own conclusions. Can't help but feel there may be more involved here.

Oh well, no rush. I'm sure specs will show up in time, we'll see what PowerTune might add to the mix; and maybe we will even be delighted before long with an announcement of a chip designed specifically for Portable use.

[Pierre B.]

Originally posted by issa:
Can't help but feel there may be more involved here.

Of course there are more parameters involved and we know nothing about these. I simply expressed my personal feelings on the situation, based on the publicly available information.

I'm sure specs will show up in time, we'll see what PowerTune might add to the mix; and maybe we will even be delighted before long with an announcement of a chip designed specifically for Portable use.

I am all eyes and ears to see and hear about PowerTune.

[Eug]

Originally posted by Pierre B.:
Eug has summarised most if not all of the publicly available specifications. With a G5 130 nm 1.8 GHz running at 47 W, and the 1.2 GHz running at 19 W, you cannot expect the 1.6 GHz version to run at less than 38-40W. Now, if we are optimists, we could assume that the transition 130 nm --> 90 nm will cut power consumption by a 30-40%. We still talk about 25 W typical power or more. That PowerTune thing may indeed be the answer. I would not hold my breath though.

The thing I find very interesting is the die size. The chips are now only 66 mm2. That's TINY by high-performance CPU standards these days. That in itself will go a long way in reducing power requirements.

I'm predicting it will be under 25 W for a 1.5-1.6 GHz part at 90 nm, and less for a 1.4 GHz part. However, it will need to be at a lower voltage than the usual desktop 90 nm G5s, and thus may be a more expensive part than the clock speed would suggest, but hopefully PowerTune can in lowering the power requirements so that the low power chips don't have to be too different from the other chips.

In some ways it would be analogous to Motorola sub 1 GHz low power G4 parts. Much lower power utilization requirements than the normal 1 GHz parts, but at higher cost. (The problem with the Motorola low power parts is that they - so far - max out at 1 GHz, too low for a PowerBook revision.)

[Eug]

The sad part for the state of PowerBooks is the fact that my 2002 TiBook still has 4/5ths the CPU speed of the 2004 AluBook. The G5 PowerBook needs to appear as soon as possible.

[mchang]

Getting in late in this discussion, so let's just jump in:

Since the introduction of the G5, I've been fantasizing about the laptop version. Yum! But Apple has lots of hurdles ahead.

Sure, the G5 consumes about 50W, which is far less than P4 HT processors, competing with Intel's other processor, the Pentium-M (Centrino, Banias, whatever), is the key to moving some great G5 laptops.

Even at 1.6GHz and 130nm, the P-M goes for 24.5W max power dissipation. That's nice. Combine that with a 48WH battery, and you get 3-4 hours runtime easily in the Dell 600m, say.

The P-M is very power frugal. They aimed for 1W average power dissipation. ONE WATT. That's through a combination of lots of on-chip technology as well as operating system support for very fine-grained control of power consumption. That's important.

Apple has got to get the G5 in a mobile version that can be power-scaled easily. They also need to integrate support into the operating system.

Both of these things take lots of time. I wouldn't expect one until Christmas.

[Eug]

Originally posted by mchang:
Getting in late in this discussion, so let's just jump in:

Since the introduction of the G5, I've been fantasizing about the laptop version. Yum! But Apple has lots of hurdles ahead.

Sure, the G5 consumes about 50W, which is far less than P4 HT processors, competing with Intel's other processor, the Pentium-M (Centrino, Banias, whatever), is the key to moving some great G5 laptops.

Even at 1.6GHz and 130nm, the P-M goes for 24.5W max power dissipation. That's nice. Combine that with a 48WH battery, and you get 3-4 hours runtime easily in the Dell 600m, say.

The P-M is very power frugal. They aimed for 1W average power dissipation. ONE WATT. That's through a combination of lots of on-chip technology as well as operating system support for very fine-grained control of power consumption. That's important.

Apple has got to get the G5 in a mobile version that can be power-scaled easily. They also need to integrate support into the operating system.

Both of these things take lots of time. I wouldn't expect one until Christmas.

I could be wrong, but I seem to remember the 1 W number applying to a different situation altogether - ie. not under reasonable usage with a fast Pentium M.

We're talking G5 1.6 here, which is more in the Pentium M's upper end speeds. The Pentium M 1.7 is a great chip, but it isn't an uber-low power chip either. It is low power, yes, but uber-low power now. It will be overall a better design than a G5 1.6, but a G5 1.6 is still quite feasible for a relatively long life laptop IMO.

IBM's PowerTune (of which we'll hear more in February) won't be the complete answer, but it's a good start. I fully suspect a sub-25 Watt max part from IBM, at 1.5-1.6 GHz.

[mchang]

The 1W number is their target average consumption. That is definitely a marketing number. Even at 2x or 3x that number, though, 3W is pretty miserly.

Intel has designed a mobile, power-friendly processor from basically the ground up, so IBM and Apple will have to compete with that, which will be tough.

[Pierre B.]

Originally posted by mchang:
Intel has designed a mobile, power-friendly processor from basically the ground up, so IBM and Apple will have to compete with that, which will be tough.

It will be more than tough if Apple and IBM have not already worked in the same direction, as Intel did.

Originally posted by Eug:

The sad part for the state of PowerBooks is the fact that my 2002 TiBook still has 4/5ths the CPU speed of the 2004 AluBook.

33% up in processor power in 14 months, is poor update perfomance CPU-wise, to say the least. Furthermore, Pentium-Ms really trounce the G4s of the Powerbooks in FPU operations. I hope Apple extends the use of IBM desktop chips to the other product lines as soon as possible. As for the Powerbook: if they don't do something quickly, I think their pro mobile line is going to lose much of the value it has. I know it is the whole package that counts, but this is not the best marketing argument today, especially when the competition has some remarkable achievements.

Sadly, Apple is for now haunted by Motorola's ghost; not only the IBM chip is found only in the Powermacs-Xserves, but the iBook took also a Motorola chip. I am not saying that this is bad in itself, the iBook is more powerful and close to the Powerbook than ever; but the unavailability of a powerful processor for the Powerbook impedes the iBook line too.

[Graymalkin]

With the introduction of the 90nm G5s the propect of G5 Powerbooks is a lot more realistic than it was just two months ago. Even without in-chip clock steppings the chipset can automagically underclock the processor and memory bus in order to keep cool. Also a G5 even clocked at 1.4GHz is going to seriously out perform an equivilently clocked G4. In most tests between the low end G5s and G4s the G4s are either DP 1.25GHz or DP 1.42GHz systems. They've got twice the processing muscle as the single G5s and are just able to keep up or surpass them. In a one on one shootout the G5 is going to bring dwon the hurt on the G4.

If any G5 Powerbook is in the works I would bet it won't be out until about the summer. I fully expect to see 3GHz PowerMacs released in the next few months. I think it would make sense to release updated Powerbooks at about the same time as they would be due for an upgrade by then. It would also give the PowerMacs a visible advantage over the mobile product. The PowerMacs might be say 2, 2.5 and 3GHz while the Powerbooks might be 1.3, 1.4, and 1.5GHz each. That's just me guessing though. The G4 pretty much hits the wall at 1.42GHz as far as MOT is concerned so sticking the Powerbooks with the G4 for another whole year would just put them in a crappy position in the market.

[mchang]

I thought I read that IBM wouldn't be delivering the 90nm G5 until the fall at earliest. Probably won't be seeing 3GHz in the 130nm.

[Pierre B.]

Originally posted by mchang:
I thought I read that IBM wouldn't be delivering the 90nm G5 until the fall at earliest. Probably won't be seeing 3GHz in the 130nm.

The G5 in the new Xserve is a 90 nm processor.

[mchang]

my bad, read it wrong. they were sampling LAST autumn, not this coming autumn .

w00t.

[Eug]

Originally posted by mchang:
The 1W number is their target average consumption. That is definitely a marketing number. Even at 2x or 3x that number, though, 3W is pretty miserly.

Intel has designed a mobile, power-friendly processor from basically the ground up, so IBM and Apple will have to compete with that, which will be tough.

I still don't know to what you're referring exactly.

Maybe Intel has a 1 W idle with their slowest chip, but that's a 600 MHz Pentium M, running at 0.84 Volts. The TDP of that CPU is only 4 Watts, which is very impressive, but who wants a 600 MHz Pentium M these days?

OTOH, the Pentium M 1.5 GHz has a TDP of 25 Watts (with a voltage of about 1.5, almost double that of the 600 MHz CPU), and idles at over 7 W.

Yes, the Pentium M is an excellent chip, and I believe that the G5 90 nm laptop CPUs won't be quite as power efficient. However, I do not believe the Pentium M is magic. The G5 in a laptop with a few tweaks will be competitive in terms of power, and I fully believe that IBM has a 1.5+ GHz chip in the works that has a maximum wattage in the ballpark of 25 Watts.

[Pierre B.]

Originally posted by Eug:
Yes, the Pentium M is an excellent chip, and I believe that the G5 90 nm laptop CPUs won't be quite as power efficient. However, I do not believe the Pentium M is magic. The G5 in a laptop with a few tweaks will be competitive in terms of power, and I fully believe that IBM has a 1.5+ GHz chip in the works that has a maximum wattage in the ballpark of 25 Watts.

How do you think such a [email protected] GHz would compare against a Pentium M at, say, 1.6 GHz? I mean calculation power, not power consumption.

[Eug]

Originally posted by Pierre B.:
How do you think such a [email protected] GHz would compare against a Pentium M at, say, 1.6 GHz? I mean calculation power, not power consumption.

I dunno. I'd just be guessing, but I'd say it'd be in the same ballpark, with the edge going to the Pentium M 1.6 on average.

[Pierre B.]

Originally posted by Eug:
I dunno. I'd just be guessing, but I'd say it'd be in the same ballpark, with the edge going to the Pentium M 1.6 on average.

Of course there is the test software issue, so anyway things would not be clear. However, do you know how the Pentium M compares to the mobile Athlon64?

[Eug]

Originally posted by Pierre B.:
Of course there is the test software issue, so anyway things would not be clear. However, do you know how the Pentium M compares to the mobile Athlon64?

Not really comparable.

The Mobile Athlon64 isn't even in the same league for power specs. IIRC, it uses much more power than even a G5 2.0 90 nm.

[cowerd]

I fully believe that IBM has a 1.5+ GHz chip in the works that has a maximum wattage in the ballpark of 25 Watts.

Yes, but I'm not so sure its the .09u 970 thats in the Xserve.

The footprint seems a little small for a CPU--like it was missing some components.

[Eug Wanker]

Originally posted by cowerd:
Yes, but I'm not so sure its the .09u 970 thats in the Xserve.

The footprint seems a little small for a CPU--like it was missing some components.

The G5 0.13 and 0.09 have pretty much the same number of transistors - ie. no missing parts. It's just very small because of the die shrink and because they didn't add extra cache. Both version have 512 KB L2 cache. Often with a die shrink companies like Intel add extra cache, but IBM did not. I suspect it's partially because power characteristics and time-to-ship were deemed more important than raw power.

[mchang]

Originally posted by Eug:
I still don't know to what you're referring exactly.

Maybe Intel has a 1 W idle with their slowest chip, but that's a 600 MHz Pentium M, running at 0.84 Volts. The TDP of that CPU is only 4 Watts, which is very impressive, but who wants a 600 MHz Pentium M these days?

OTOH, the Pentium M 1.5 GHz has a TDP of 25 Watts (with a voltage of about 1.5, almost double that of the 600 MHz CPU), and idles at over 7 W.

Yes, the Pentium M is an excellent chip, and I believe that the G5 90 nm laptop CPUs won't be quite as power efficient. However, I do not believe the Pentium M is magic. The G5 in a laptop with a few tweaks will be competitive in terms of power, and I fully believe that IBM has a 1.5+ GHz chip in the works that has a maximum wattage in the ballpark of 25 Watts.

I was referring to the Intel specs on the P-M. they quote target or 1W for "typical operation" for the regular model (1.6GHz-1.3GHz), and 0.5W with their ultra-low-voltage parts (900Mhz-600Mhz).

The 7W figure is usually quoted from somewhere in here:
Intel

But that's the operating envelope for a passively-cooled device.

Check here:
dynamic power

And you'll see power drawn as measured on the processor to be only 3W on a 1.4GHz P-M at (arguably) full load, and 1.35W at idle.

I'm not trying to trumpet ChipZilla. I'm just stating some facts, and they add up to that it's an entire system that needs to be optimized. Let's just hope Apple can do it.

Mark

[Pierre B.]

Originally posted by mchang:
I'm just stating some facts, and they add up to that it's an entire system that needs to be optimized. Let's just hope Apple can do it.

Mark

I have doubts that Apple can introduce something similar in power efficiency in a reasonable time frame (before year's end). Intel's investment on this technology was huge and it took time.

The only possibility is something in the rumor realm today: MacRumors wrote sometime ago that IBM and Apple are preparing a new chip (if I remember correctly, 64-bit), specially for notebooks and embedded applications. It would not come in the light of day though before 2005.

[Pierre B.]

Got it!

[Eug Wanker]

Originally posted by mchang:
I was referring to the Intel specs on the P-M. they quote target or 1W for "typical operation" for the regular model (1.6GHz-1.3GHz), and 0.5W with their ultra-low-voltage parts (900Mhz-600Mhz).

The 7W figure is usually quoted from somewhere in here:
Intel

But that's the operating envelope for a passively-cooled device.

Check here:
dynamic power

And you'll see power drawn as measured on the processor to be only 3W on a 1.4GHz P-M at (arguably) full load, and 1.35W at idle.

I'm not trying to trumpet ChipZilla. I'm just stating some facts, and they add up to that it's an entire system that needs to be optimized. Let's just hope Apple can do it.

Mark

If you're talking <1.5 W at idle, you're talking low-power mode in battery-optimized configs, but full load will not give you 3 W for a 1.4 GHz Pentium M. It's over 20 Watts.

Sure it helps, but the numbers are not comparable at all to the G5 numbers at all. Remember that the G5 already also has bus slewing and PowerTune which I have not factored into the equation. The G5's design is likely not as robust for power conservation as Banias' design, but it should help a lot still.

By the way, Centrino 2 is coming, likely debuting at 1.8 GHz, but they won't be out until Q2 at the earliest. But if Apple takes too long, they'll be falling way behind yet again.

Originally posted by Pierre B.:
Got it!

Page 2 rumour. I'd just ignore it in this case to be quite honest. Even if it does exist, it's probably unrelated.

[cowerd]

Often with a die shrink companies like Intel add extra cache, but IBM did not. I suspect it's partially because power characteristics and time-to-ship were deemed more important than raw power.

It seems that companies like IBM, Intel, AMD and Moto add extra cache during a die-shrink . There is also the issue of ideal chip footprint, which the 130nm 970 seems to approach at ~120mm^2. So this seems like an on-board controller or some other systems integration seems to be missing from the 970 variant. After all power dissipation from the CPU is not the only problem as the present northbridge in the desktop G5 series seems to dissipate a significant amount of heat as well.