PowerPC 975 or Not?

[AidenShaw]

Announced in Feb 2003

Are you sure about that? I seem to remember reading that announcement a few months after WWDC - on MacRumors nonetheless.

I held one in my hands before WWDC - it was NDA and before official product announcement, but real.

Note the date on this article:

http://www.macobserver.com/article/2003/02/27.11.shtml


IBM Introduces PowerPC 970 Servers, Touts 970 As "Fastest PowerPC So Far"
by Bryan Chaffin,
12:00 PM CST, February 27th, 2003

Many have hoped that Apple will move the PowerMac line to the PowerPC 970, a new 64-bit PowerPC processor being developed by IBM.

With Motorola lagging behind in G4 development, and the G5 being nowhere in sight, many have hoped that Apple could turn to IBM's 970 line to erase the performance gap between Intel/AMD and Apple's G4-based products. This suggestion has been given more power by the fact that IBM has officially included AltiVec support in the 970, a technology crucial to many of Apple's developers.

Today, IBM announced the first products to use the 970 -- the PowerPC Blade -- at Europe's CeBit conference.

The PowerPC Blade will be added to IBM's BladeCenter product line, a line of servers powered by Intel processors. The company is touting the PowerPC Blade as "superior to Intel Blades for certain applications in the High Performance Computing Sector," and goes on to offer some details on the PowerPC 970 itself.

Included in those details is a speed range for the processor of 1.8 GHz to 2.5 GHz, higher than previous announcements from the company. ...


Hmmm.... Now why was there surprise at 2.0GHz at WWDC?

 

[thatwendigo]

First of all, you can fit seven 1U servers in a 7U space - that should be pretty obvious.

Yeeeees... So what? My point was that 7 1U servers with two processors each takes up far more space per processor than a 7U blade with 14 dual-processor blades in it (2 processors per 1U, as opposed to 4 processors per 1U).

Saying that a PPC 750fx can't run in a notebook based on a typical rating of 24.5 watts at 2GHz ignores many facts..
* Who says it has to be 2 GHz?
* What would the average wattage be if it ran at the bus slewed 1300 MHz?
* PC manufacturers already make notebooks with Pentium 4 desktop processors and Athlon 64s which use MUCH more than 24.5 watts. Sure their battery life sucks unless they de-clock themselves, but thermally, it has been done.

I'll assume that you mean the PowerPC 970FX and go from there. In the same order as your questions:

1. Nobody says that the 970 has to be run at 2.0ghz, but if it's going to be competitive with what Intel is offering right now, it's going to need to be higher clocked than 1.4-1.6ghz. Also, I find it unlikely that the 1.3ghz slewed clock of the 970FX is as efficient as the 2.0ghz Dothan's minimum power of some 5-8 watts (as I recall). In other words, you're trying to put a server-derived chip into a laptop format to compete against a chip designed to go in laptops. It sounds like a losing proposition.
2. I can't tell you what the average wattage is, but since they're on the same process and at the same core clock, but the 970FX is hotter, that doesn't bode well for the slewed wattage in my mind.
3. Correction: PC manufacturers make 10-pound monsters with less than an hour of battery life, but call it a laptop. Were Apple to get into the laughable "desktop replacement" market and give up on at all being sleek or lightweight, then they could certainly make a laptop that would be speed competitive. The problem would be that the Centrino systems would then clean the floor with them by being cheaper and just about as fast.

 

[Sun Baked]

Saying that a PPC 750fx can't run in a notebook based on a typical rating of 24.5 watts at 2GHz ignores many facts..
* Who says it has to be 2 GHz?
* What would the average wattage be if it ran at the bus slewed 1300 MHz?
* PC manufacturers already make notebooks with Pentium 4 desktop processors and Athlon 64s which use MUCH more than 24.5 watts. Sure their battery life sucks unless they de-clock themselves, but thermally, it has been done.

jmho,
ffakr

The MAJOR fact people are ignoring is that you don't design around the "Typical" number, but the "Maximum" number.

People see the 25W number and say it's low enough for the PowerBook and forget that the Maximum Power Dissipation for the 2GHz PPC970FX is 55W (Apple's marketing docs).

IBM Presentation on PPC Power Dissipation thread

 

[ClimbingTheLog]

So what's preventing you from making corporate orders of Apple products then? There's obviously something new "always" on the horizon (sometimes the horizon is further away).

The situation has been somewhat misrepresented. Yes, you buy what you need when you need it - as long as it's in your budget. It's very hard to buget for and justify mystery machines. This isn't so much an issue for a generic desktop but it's a showstopper for new product category machines. Noone budgeted for a G5 supercomputer last year because they didn't know it was coming. Same for the XServe RAID. You'll probably see some this year, because now people know about it, but you could have seen more last year if Apple wasn't so wrapped up in its own secret chic.

 

[AidenShaw]

sorry

Yeeeees... So what? My point was that 7 1U servers with two processors each takes up far more space per processor than a 7U blade with 14 dual-processor blades in it (2 processors per 1U, as opposed to 4 processors per 1U).

OK, I see my confusion.

You said "48", when you meant "28 processors per 7U" for the BladeCenter.

Then you said 14 in a 7U, and I read that as 14 1U, not 7 dualCPU 1U.

You weren't actually wrong, but had you used a few more words it would have been easier to pick up at a quick first reading.

Sorry 'bout that!

 

[NNO-Stephen]

Yeeeees... So what? My point was that 7 1U servers with two processors each takes up far more space per processor than a 7U blade with 14 dual-processor blades in it (2 processors per 1U, as opposed to 4 processors per 1U).



I'll assume that you mean the PowerPC 970FX and go from there. In the same order as your questions:

1. Nobody says that the 970 has to be run at 2.0ghz, but if it's going to be competitive with what Intel is offering right now, it's going to need to be higher clocked than 1.4-1.6ghz. Also, I find it unlikely that the 1.3ghz slewed clock of the 970FX is as efficient as the 2.0ghz Dothan's minimum power of some 5-8 watts (as I recall). In other words, you're trying to put a server-derived chip into a laptop format to compete against a chip designed to go in laptops. It sounds like a losing proposition.
2. I can't tell you what the average wattage is, but since they're on the same process and at the same core clock, but the 970FX is hotter, that doesn't bode well for the slewed wattage in my mind.
3. Correction: PC manufacturers make 10-pound monsters with less than an hour of battery life, but call it a laptop. Were Apple to get into the laughable "desktop replacement" market and give up on at all being sleek or lightweight, then they could certainly make a laptop that would be speed competitive. The problem would be that the Centrino systems would then clean the floor with them by being cheaper and just about as fast.

1. 
   
   for starters, I am only taking issue with one part of your statement. you say that when the G5 goes into PowerBooks, it's got to be higher clocked than 1.4-1.6Ghz. why? PowerBooks are currently topping out at 1.5Ghz. why the hell couldn't Apple start off with 1.4Ghz in the low end, and possibly 1.8Ghz in the high end, with 1.6 being stuck right in the middle?
   
   they could and I think you need to reconsider that point in your statement. The current PowerBooks are far from the most powerful laptops on the market. The primary goal right now is to move them to an architecture that has room to grow, the G4 is definitely out of steam for the time being. They need to move it to the G5, and with a different bus multiplier and some other tricks, this could be done. for instance, they could make it a 1:4 bus multiplyer. for instance, on a 1.8Ghz PowerBook, it would have a 450Mhz bus. that still beats the hell out of the current system bus which is the main issue of the G4. well, one of them. and furthermore, the PowerBooks don't have to be 1 inch thin. although that is ideal, think back to the PowerBook G3s. they most certainly were NOT 1 inch thin.
   
   either way, consider this point.

 

[ffakr]

The MAJOR fact people are ignoring is that you don't design around the "Typical" number, but the "Maximum" number.

People see the 25W number and say it's low enough for the PowerBook and forget that the Maximum Power Dissipation for the 2GHz PPC970FX is 55W (Apple's marketing docs).

IBM Presentation on PPC Power Dissipation thread

Your link leads to another link on PPC 750fx thermal output.. NOT 970fx thermal output.

I've yet to see any indication that the 970fx Max is 55 W. Could you provide a proper link to this info?

thanks,
ffakr.

 

[~Shard~]

for starters, I am only taking issue with one part of your statement. you say that when the G5 goes into PowerBooks, it's got to be higher clocked than 1.4-1.6Ghz. why? PowerBooks are currently topping out at 1.5Ghz. why the hell couldn't Apple start off with 1.4Ghz in the low end, and possibly 1.8Ghz in the high end, with 1.6 being stuck right in the middle?

Although I would be quite happy with a 1.6 GHz G5 PowerBook, I think one aspect of what thatwendigo is getting at is that if you clocked the chips at 1.4 GHz or 1.6 GHz, they would probably get taken down by Centrinos when one does a cost/performance comparison. Of course the G4 PowerBooks are lagging right now, and any G5 would be welcome, but unfortunately, a Centrino system would probably match up with a lower clocked G5 PowerBook quite closely, yet be quite less expensive. This is just my impression though, please correct me if I’m misinformed.

The current PowerBooks are far from the most powerful laptops on the market. The primary goal right now is to move them to an architecture that has room to grow, the G4 is definitely out of steam for the time being.

Agreed – the G5 is the chip of the future (until the G6 comes out!) and it is obvious that Apple is going to work on putting them into the PowerBooks. However, the 1.5 GHz 17” PowerBooks are pretty fast, all things considered, and I wouldn’t classify them as slow by any means. Slow compared to some Intel systems, perhaps, but there a lot of other factors at play that would still make me select a PowerBook over an Intel laptop. Plus, my G4 does everything I need it to, and more - everyone has their different requirements of course, but for many people, a G4 is more than adequate (supplemented with a lot of RAM of course!)

They need to move it to the G5, and with a different bus multiplier and some other tricks, this could be done. for instance, they could make it a 1:4 bus multiplyer. for instance, on a 1.8Ghz PowerBook, it would have a 450Mhz bus. that still beats the hell out of the current system bus which is the main issue of the G4. well, one of them.

Easier said than done - take it from an Electronics Systems Engineer who has dabbled in microprocessor and hardware design. There are numerous other factors involved with cramming one of those G5 beasts into a laptop, which have been addressed numerous times on these forums – motherboard redesign, heating/power issues, etc. and a lot of other technical issues I won't re-hash here. It will obviously be done at some point, but I don’t see G5 PowerBooks being announced until 2005 (and who knows when they’ll ship!) Just being realistic.

The fact that some people overlook (not saying you’re one of them, by the way), is that the G5 is a server-class chip - putting the G5 in a PowerBook is the equivalent of putting an Itanium into an Intel laptop – not that easy to accomplish! The other fact, like it or not, is that the G5 was not designed to be a portable chip. And although Apple will of course turn it into one, and re-engineer the PowerBooks to incorporate and handle them, it will still take some time.

and furthermore, the PowerBooks don't have to be 1 inch thin. although that is ideal, think back to the PowerBook G3s. they most certainly were NOT 1 inch thin.

Yes, but technology has come a long way since then – think back 15 years ago to the monsters that they used to call laptops! I don’t think that’s a fair excuse for making the G5 PowerBooks thicker, and many people won’t be satisfied with the reasoning and necessity of a larger enclosure. As a result, Apple will no doubt be in a bit of an engineering quandary, as larger enclosures will no doubt be required!

I’ll definitely be interested to see the G5 PowerBooks when they come out, (especially if they’re 1” thick!) and I have no doubts I will drool over the engineering marvel that Apple will accomplish as I have many times before.

 

[Sun Baked]

Your link leads to another link on PPC 750fx thermal output.. NOT 970fx thermal output.

I've yet to see any indication that the 970fx Max is 55 W. Could you provide a proper link to this info?

thanks,
ffakr.

Read between the lines... Typical Power Dissipation is NOT the design criteria, Maximum Power is.

All we have seen is IBMs Typical numbers, not their max. Apple posted the Max awhile ago, and everyone ignored it.

If you do a search you can probably find the original link.

 

[NNO-Stephen]

Although I would be quite happy with a 1.6 GHz G5 PowerBook, I think one aspect of what thatwendigo is getting at is that if you clocked the chips at 1.4 GHz or 1.6 GHz, they would probably get taken down by Centrinos when one does a cost/performance comparison. Of course the G4 PowerBooks are lagging right now, and any G5 would be welcome, but unfortunately, a Centrino system would probably match up with a lower clocked G5 PowerBook quite closely, yet be quite less expensive. This is just my impression though, please correct me if I’m misinformed.



Agreed – the G5 is the chip of the future (until the G6 comes out!) and it is obvious that Apple is going to work on putting them into the PowerBooks. However, the 1.5 GHz 17” PowerBooks are pretty fast, all things considered, and I wouldn’t classify them as slow by any means. Slow compared to some Intel systems, perhaps, but there a lot of other factors at play that would still make me select a PowerBook over an Intel laptop. Plus, my G4 does everything I need it to, and more - everyone has their different requirements of course, but for many people, a G4 is more than adequate (supplemented with a lot of RAM of course!)



Easier said than done - take it from an Electronics Systems Engineer who has dabbled in microprocessor and hardware design. There are numerous other factors involved with cramming one of those G5 beasts into a laptop, which have been addressed numerous times on these forums – motherboard redesign, heating/power issues, etc. and a lot of other technical issues I won't re-hash here. It will obviously be done at some point, but I don’t see G5 PowerBooks being announced until 2005 (and who knows when they’ll ship!) Just being realistic.

The fact that some people overlook (not saying you’re one of them, by the way), is that the G5 is a server-class chip - putting the G5 in a PowerBook is the equivalent of putting an Itanium into an Intel laptop – not that easy to accomplish! The other fact, like it or not, is that the G5 was not designed to be a portable chip. And although Apple will of course turn it into one, and re-engineer the PowerBooks to incorporate and handle them, it will still take some time.



Yes, but technology has come a long way since then – think back 15 years ago to the monsters that they used to call laptops! I don’t think that’s a fair excuse for making the G5 PowerBooks thicker, and many people won’t be satisfied with the reasoning and necessity of a larger enclosure. As a result, Apple will no doubt be in a bit of an engineering quandary, as larger enclosures will no doubt be required!

I’ll definitely be interested to see the G5 PowerBooks when they come out, (especially if they’re 1” thick!) and I have no doubts I will drool over the engineering marvel that Apple will accomplish as I have many times before.

please note that I agree with you that we more than likely will not see a PowerBook G5 until MWSF '05 at the earliest.

Only one issue with your statements. You say that the G5 at a low clock speed 1.6Ghz won't be able to compete with Intel Centrino notebooks, yet you say that a 1.5Ghz G4 is plenty of power. this is a bit confusing. Seems to me that a 1.6Ghz G5 is more powerful than a 1.5Ghz G4. The cost will be about the same most likely, and yet the G5 PowerBooks will be more powerful, so how the Centrino's will have an even greater lead in any price/performance comparison than they already do... please explain.

 

[ffakr]

The cost will be about the same most likely, and yet the G5 PowerBooks will be more powerful, so how the Centrino's will have an even greater lead in any price/performance comparison than they already do... please explain.

The cost of a 1.6 GHz G5 would likely be much less than a 1.5 GHz G4.

.09 micron 970fx'es are only 65^mm. That's pretty tiny. And they are made on a 300mm wafer. IBM should be able to make them much more efficiently than Motorola especially considering that Moto has historically had bad yields.

Also, a 1.5 G4 is the high end for Moto.. the top sort is always the most scarce.
A 1.6 GHz 970fx would be the absolute bottom of the speed bin barrel. They would be chips that might otherwise be considered rejects. Apple could probably get slow G5s for a song.

I, also, really don't expect we'll see Powerbook G5s any time soon. I spend last night with a 17" 1.33GHz Powerbook and I thought it was a plenty speedy machine for a laptop so I'm not freaking out because we still have a G4 in the powerbook line either. I've simply been trying to debunk the idea that it would be absolutely impossible to put a chip that hot into a notebook (Alienware would definately argue that)
All that said, there is one really compelling reason to move the powerbooks to a low clock G5 processor.. because you could make the transition to a modern architecture sooner rather than later. Why is it a bad idea to move to the G5 architecture early on even if the resultant machine is only slightly faster than the existing machine. It's a win-win situation in my eyes. Users get a faster machine, and Apple has an easier time ramping up to more agressively clocked G5s later on.

as always
jmho
ffakr

 

[~Shard~]

please note that I agree with you that we more than likely will not see a PowerBook G5 until MWSF '05 at the earliest.

Only one issue with your statements. You say that the G5 at a low clock speed 1.6Ghz won't be able to compete with Intel Centrino notebooks, yet you say that a 1.5Ghz G4 is plenty of power. this is a bit confusing. Seems to me that a 1.6Ghz G5 is more powerful than a 1.5Ghz G4. The cost will be about the same most likely, and yet the G5 PowerBooks will be more powerful, so how the Centrino's will have an even greater lead in any price/performance comparison than they already do... please explain.

My apologies, I should have clarified this better - I am kind of speaking of 2 different things, so perhaps I didn't make myself clear. All I meant was that independently, on its own, a 1.5 GHz G4 is quite powerful for the average user's needs. Do Centrino's and the like outperform it on certain benchmarks and price? Yes, of course. The G4 PowerBooks could technically be the slowest laptops out there (although obviously they aren't) but if it does the job for users, it does the job, right? And I was essentially saying the same for the G5, just to a lesser degree. They of course will be faster than the 1.5 GHz G4s, but not significantly so as to tip the scales, so to speak, when doing certain comparisons to the PC laptop world. So as for your inference that Centrinos will have an even greater lead than they already do, I never said this nor implied this, so my apologies. Obviously the G5 PowerBooks will fare better in benchmarks than their G4 counterparts. Also please take into account that by the time the G5 PowerBooks are released, there will no doubt have been advancements to the Centrino line as well, so although it's fair to compare current G4 PowerBooks with current PC offerings, we cant really compare current PC offerings with future G5 PowerBooks.

Fair enough?

 

[PPC970FX]

IBM have for a year had a PPC CPU running in a lab at 3Ghz. Thay did not say if it was 970, 970fx, 975, 980. But it was a 3Ghz PPC CPU. Dear Stev, I have been a god boy all year. Plz give me a 3Ghz at WWDC. Then I will not dobut you again.

If Stev/ IBM makes it ot 3Ghz they will blow AMD, Intel out of the water. Can`t wait ot see the x86 people to see the dust behaind the PM. IBM is not moto. IBM have a lot more money, and they need apple, to make god programs, so that the samme programs could be used on there Power 5 CPU. And IBM are pissed off at Intel after they tok over the PC world.

muhahah

 

[NNO-Stephen]

IBM have for a year had a PPC CPU running in a lab at 3Ghz. Thay did not say if it was 970, 970fx, 975, 980. But it was a 3Ghz PPC CPU. Dear Stev, I have been a god boy all year. Plz give me a 3Ghz at WWDC. Then I will not dobut you again.

If Stev/ IBM makes it ot 3Ghz they will blow AMD, Intel out of the water. Can`t wait ot see the x86 people to see the dust behaind the PM. IBM is not moto. IBM have a lot more money, and they need apple, to make god programs, so that the samme programs could be used on there Power 5 CPU. And IBM are pissed off at Intel after they tok over the PC world.

muhahah

that has to be the worst bastardization of the english language I've ever come across.

 

[maverick13]

Guys, guys let's be a little serious here. Of course the G5 Powerbooks will be faster than the Centrino ones when they are released. I give you that the Centrinos may get an update as well but they are not IBM POWER4-based 64 bit G5s
Think is just my opinion based to whatever I've read about this chip's architecture.

Maverick

PS: Do you know the consider the POWER4 to be the most powerfull CPU available? And the G5 has an excellent FP performance at the level of the Itanium 2

 

[NNO-Stephen]

Guys, guys let's be a little serious here. Of course the G5 Powerbooks will be faster than the Centrino ones when they are released. I give you that the Centrinos may get an update as well but they are not IBM POWER4-based 64 bit G5s
Think is just my opinion based to whatever I've read about this chip's architecture.

Maverick

PS: Do you know the consider the POWER4 to be the most powerfull CPU available? And the G5 has an excellent FP performance at the level of the Itanium 2

the POWER 5 craps all over the POWER 4 in terms of performance.

 

[thatwendigo]

for starters, I am only taking issue with one part of your statement. you say that when the G5 goes into PowerBooks, it's got to be higher clocked than 1.4-1.6Ghz. why? PowerBooks are currently topping out at 1.5Ghz. why the hell couldn't Apple start off with 1.4Ghz in the low end, and possibly 1.8Ghz in the high end, with 1.6 being stuck right in the middle?

A 970Fx 1.4ghz would be lower performing than the G4 at 1.5ghz, unless we see numbers that tell us otherwise. To take some of the rest of what you say and throw it in, reducing the bus multiplier is just asking for the performance to degrade even further, since a huge part of the strength of the G5 is its underlying architecture. Take that away (FSB, SATA, and all) and you neuter a powerful chip and it gets slaughtered on the portable market for being neither as battery-efficient nor as powerful as the Centrino. So, what you're advocating here is a crippled G5 in a thicker powerbook that would lose in performance and battery life tests to its competition, all so that there could be the magical "G5" in the name? I'll take a miss on that one, and I hope Apple does, too.

I'm not sure who's "primary goal" you're talking about here, but I want Apple's "primary goal" to be making the best laptop possible, not caving in to people who don't understand system design and just want a shinier, higher number in the specs.

Of course the G4 PowerBooks are lagging right now, and any G5 would be welcome, but unfortunately, a Centrino system would probably match up with a lower clocked G5 PowerBook quite closely, yet be quite less expensive.

There's also the minor issue of battery life to be dealt with, since current Centrinos can live for five hours (in actuality, not just marketing) if you're not pushing them all-out for rendering and similar tasks. When allowed to slew, the thing drops around two thirds of its clockspeed and dissipates a paltry 6-8 watts. The current MPC7447A in the PowerBook dissipates 11 watts in general.

Agreed – the G5 is the chip of the future (until the G6 comes out!) and it is obvious that Apple is going to work on putting them into the PowerBooks.

I disagree. The 970 is the future of the lowend, and a serious performance level chip needs to take its place as quickly as possible, because Intel and AMD are about to take some serious leaps. The Athlon 64 line is moving to 90nm and low-k dialectric processes, while having already achieved 2.4ghz clockspeeds and competiition with the fastest PC systems. Jonah, the Centrino-based 90nm dual-core processor, is due out next year if it doesn't slip its target, which I find less likely now that Intel has publically declared the Pentium 4 to be at least temporarily dead.

The fact that some people overlook (not saying you’re one of them, by the way), is that the G5 is a server-class chip - putting the G5 in a PowerBook is the equivalent of putting an Itanium into an Intel laptop – not that easy to accomplish! The other fact, like it or not, is that the G5 was not designed to be a portable chip. And although Apple will of course turn it into one, and re-engineer the PowerBooks to incorporate and handle them, it will still take some time.

No matter how often you point this out, people will call you and I the ones who are dreaming when we talk about other chips being better solutions. Go figure.

All we have seen is IBMs Typical numbers, not their max. Apple posted the Max awhile ago, and everyone ignored it.

The 1.33ghz xServe utilizes the MPC7455, a generation behind the current and much lower power MPC7447A. If you double the typical wattage for the MPC 7447A at 1.5ghz, you come up with 16-22 watts (depending on your source), as opposed to 33 watts for the older processor.

The cost of a 1.6 GHz G5 would likely be much less than a 1.5 GHz G4.

.09 micron 970fx'es are only 65^mm. That's pretty tiny. And they are made on a 300mm wafer. IBM should be able to make them much more efficiently than Motorola especially considering that Moto has historically had bad yields.

This I agree with, actually. The G5 should be significantly cheaper per chip than the G4 ever was, in part because of the size of the die, but also because IBM seems to have a firmer rein on their quality control than Motorola did. It's likely a significant part of the allowance for massive technology upgrades in the towers while keeping the same price points.

I've simply been trying to debunk the idea that it would be absolutely impossible to put a chip that hot into a notebook (Alienware would definately argue that)

I never said it was impossible and never would happen, no matter what/ I did, however, say that the current form factor and battery life would basically be eradicated by doing something like this. There is no PC laptop I've ever come across that's as quiet as an Apple PowerBook of the same generation, and that's largely because even the Centrino is hotter than the current offering.

Why is it a bad idea to move to the G5 architecture early on even if the resultant machine is only slightly faster than the existing machine. It's a win-win situation in my eyes. Users get a faster machine, and Apple has an easier time ramping up to more agressively clocked G5s later on.

I'd be remarkably surprised if the G5 at an even remotely similar clock was faster the the MPC7447A in a portable. If you cut back the bus multiplier and ramp the chip down to save on battery life, then you're killing the main benefits of the chip itself - which says nothing of the less efficient AltiVec implementation.

 

[~Shard~]

So, what you're advocating here is a crippled G5 in a thicker powerbook that would lose in performance and battery life tests to its competition, all so that there could be the magical "G5" in the name? I'll take a miss on that one, and I hope Apple does, too.

This is actually a good summary, emphasis on crippled - the G5 essentially would indeed have to be crippled to be re-engineered for the PowerBooks at this stage, and cutting corners here and there to accomplish this task (i.e. messing with the bus multipliers) will take away from what the G5 truly is, and, in essence, what makes it a G5 in the first place!

I disagree. The 970 is the future of the lowend, and a serious performance level chip needs to take its place as quickly as possible, because Intel and AMD are about to take some serious leaps. The Athlon 64 line is moving to 90nm and low-k dialectric processes, while having already achieved 2.4ghz clockspeeds and competiition with the fastest PC systems. Jonah, the Centrino-based 90nm dual-core processor, is due out next year if it doesn't slip its target, which I find less likely now that Intel has publically declared the Pentium 4 to be at least temporarily dead.

My apologies, you misunderstood me - by saying the G5 was the future for Apple, all I meant was the G5 or a future derrivative, whether that is 90nm, FX, G6, POWER5, POWER6 - whatever Apple wants to eventually call it - would be the future. I was simply referring to the G5 architecture in general, which I think will be around for some time in one form or another, and not the current 970 chipset itself. I agree with you - due to AMD and Intel's future roadmaps, Apple does need to make significant advancements in the near future to stay competitive. (And for some of you it may come as a surprise, but yes, Intel does have roadmaps, is NOT stalled out and still needs to be respected in this market.)

No matter how often you point this out, people will call you and I the ones who are dreaming when we talk about other chips being better solutions. Go figure.

I know what you mean.

I never said it was impossible and never would happen, no matter what/ I did, however, say that the current form factor and battery life would basically be eradicated by doing something like this. There is no PC laptop I've ever come across that's as quiet as an Apple PowerBook of the same generation, and that's largely because even the Centrino is hotter than the current offering.

This is somewhat where I am coming from as well. Although I still stand by prediction of G5 PowerBooks in 2005 (and definitely not sooner), it's more because I believe that right or wrong and regardless of the improbability, Apple will accomplish this engineering feat - whether it ends up being a good thing or not! Is it the best solution? Perhaps not - the G5 isn't suited for laptops, as I indicated above, just like the Itanium2s aren't suited for Intel laptops.

I'd be remarkably surprised if the G5 at an even remotely similar clock was faster the the MPC7447A in a portable. If you cut back the bus multiplier and ramp the chip down to save on battery life, then you're killing the main benefits of the chip itself - which says nothing of the less efficient AltiVec implementation.

Once again, to summarize your above point, you are effectively crippling the G5 if you do this. Will the G5 PowerBook perform better than the G4 PowerBooks? It depends, but probably. By how much? Not as much as many people would think, especially from getting their hopes up by seeing that glistening "G5" emblem on it.

Hmm, I can just see the posts now - all the people who have been whining and complaining for G5 PowerBooks, (and will continue to do so for the rest of the year), slamming Apple for not being able to release them for so long, and then by the time the G5 PowerBooks actually do come out, they'll be so disappointed with the specs and peformance that they'll just start complaining again. <sigh, shudder>

 

[NNO-Stephen]

1.4Ghz for the low end, to replace the current low end. the 1.8 would replace the 1.5Ghz G4. besides, even with the crippled bus, it would still be a huge advantage over the current ****ty busses in all G4 based macs.

 

[thatwendigo]

1.4Ghz for the low end, to replace the current low end. the 1.8 would replace the 1.5Ghz G4. besides, even with the crippled bus, it would still be a huge advantage over the current ****ty busses in all G4 based macs.

How is it an "advanage" if it drains the battery more but doesn't provide a performance advantage. Keep in mind that the single 7447A competes with the 1.6ghz and 1.8ghz 970 at this point, and that Motorola is developing the G4's successor as we speak - a 2.0ghz dual-core, 400mhz FSB, on-die DDR controller chip that's pin-compatible with the older 74xxs. All of that, and you get two processors at low latency and high clock. How much does this cost in heat? A little more than a singel 970FX 2.0ghz, at around 30 watts instead of 24.5 watts. That would be worth making a sacrifice for, but this wouldn't.

 

[NNO-Stephen]

go to Bare Feats. the 1.5 doesn't preform as well as the 1.6Ghz, much less the 1.8.

check your facts. the recently published FCP4 benchmarks might be useful.

 

[~Shard~]

go to Bare Feats. the 1.5 doesn't preform as well as the 1.6Ghz, much less the 1.8.

check your facts. the recently published FCP4 benchmarks might be useful.

I believe you when you say the 1.5 GHz G4 doesn't perform as well as the 1.6 GHz G5, however, unless I am looking at the wrong FCP test, it appears as though this test was conducted between the 1.5 GHz G4 PowerBook vs. the 1.6 GHz G5 PowerMac (among other systems of course). In this case, of course the 1.6 GHz G5 will outperform the 1.5 GHz G4. However, this is simply due to the architecture of the PowerMac vs. the PowerBook, where there is a lot more at play than simply the chip - FSB, RAM, the list goes on and on, coupled with the fact that you're comparing a Power desktop machine to a laptop. If the 1.6 GHz G5 was put in a PowerBook, you could not expect the same benchmarks out of it, due to the re-engineering and corner-cutting that would have to take place as previously mentioned.

Again, unless I'm looking at the wrong test (please correct me if I am), this appears to be a test between the systems as a whole, not simply the chips, which introduces a whole range of additional variables into the equation - as a result, you cannot accurately compare the processors on their own. Perhaps I am missing something though?

 

[thatwendigo]

I believe you when you say the 1.5 GHz G4 doesn't perform as well as the 1.6 GHz G5, however, unless I am looking at the wrong FCP test, it appears as though this test was conducted between the 1.5 GHz G4 PowerBook vs. the 1.6 GHz G5 PowerMac (among other systems of course). In this case, of course the 1.6 GHz G5 will outperform the 1.5 GHz G4.

MPC7455 1.25ghz scores 1412
MPC7447A 1.33ghz scores 1269
MPC7447A 1.5ghz scores 1236
IBM 970 1.6ghz scores 1123

7455 -> 7447A = 80mhz and 143 point shift
7447A -> 970 = 100mhz and 113 point shift
970 -> next 970 model = 200mhz and 157 point shift

143/80 = 1.79 per mhz
113/100 = 1.13 per mhz
157/200 = 0.78 per mhz

Despite the massive FSB, RAM, and I/O advantage, the G5 is slower clock-for-clock than the revised MPC7447A. This does not deny the performance shift, but it does show that the G5 is not necessarily as amazing in single processor systems as the G5 PowerBook zealots would have you believe.

If the 1.6 GHz G5 was put in a PowerBook, you could not expect the same benchmarks out of it, due to the re-engineering and corner-cutting that would have to take place as previously mentioned.

More interestingly, the G5 loses on efficiency even when it has its support systems in place. Also, the towers suck up an astounding 420+ watts at peak operation, and that's without add-on cards or anything other than the parts inside the case when it ships.

 

[Telomar]

MPC7455 1.25ghz scores 1412
MPC7447A 1.33ghz scores 1269
MPC7447A 1.5ghz scores 1236
IBM 970 1.6ghz scores 1123

7455 -> 7447A = 80mhz and 143 point shift
7447A -> 970 = 100mhz and 113 point shift
970 -> next 970 model = 200mhz and 157 point shift

143/80 = 1.79 per mhz
113/100 = 1.13 per mhz
157/200 = 0.78 per mhz

Despite the massive FSB, RAM, and I/O advantage, the G5 is slower clock-for-clock than the revised MPC7447A. This does not deny the performance shift, but it does show that the G5 is not necessarily as amazing in single processor systems as the G5 PowerBook zealots would have you believe.

There's actually a pretty fundamental flaw in how you're looking at those results and that is there is pretty much no meaning to the argument you are putting forward based on the numbers. Having numbers in an argument doesn't benefit it unless there is some meaning behind using them.

First things first, the 7455 to 7447 sees an increase in the level 2 cache along with a better graphics cards and other motherboard improvements. You'd find the level 2 cache is almost entirely responsible for the improvement hence the reason the jump is more substantial.

Just looking at the 1.5 GHz vs the 1.33 GHz, and these are both the same chip with similar motherboards, you can see for near identical G4 systems the scaling in performance is crap. In fact for a 167 MHz it gains 33 seconds and you want to talk about the 7447 being an efficient chip? Add a 78 seconds to the G5's times to simulate a 1.5 GHz chip and it still comes out ahead, which actually suggests clock for clock it is more efficient. Mind you there is cross over at about 1.42 GHz.

The G4 was a potentially good chip but it's currently crippled and scaling it further won't gain anything. They either need to improve the G4, and there are plenty of areas to improve, or swap it out for a G5 if they realistically want to continue the line otherwise performance will just go nowhere. As for the power debate I don't think you'd find either chip is a great deal better for power. A 1.6 GHz G5 will have about the same power consumption as a 1.5 GHz G4 and Apple is using a higher bus speed and higher frequency, for reference the 7447A has a max power consumption of 30W at 1.42 GHz or 5W higher than 1.33 GHz and I suspect 1.5 GHz would be very close to 40. Where the differences lie would be the supporting architecture and cost.

If IBM can sort out their manufacturing I actually think the 970 would make a much tidier package, then again I'm not working on the engineering team so am not aware of what other issues they are currently facing.

 

[~Shard~]

MPC7455 1.25ghz scores 1412
MPC7447A 1.33ghz scores 1269
MPC7447A 1.5ghz scores 1236
IBM 970 1.6ghz scores 1123

7455 -> 7447A = 80mhz and 143 point shift
7447A -> 970 = 100mhz and 113 point shift
970 -> next 970 model = 200mhz and 157 point shift

143/80 = 1.79 per mhz
113/100 = 1.13 per mhz
157/200 = 0.78 per mhz

Despite the massive FSB, RAM, and I/O advantage, the G5 is slower clock-for-clock than the revised MPC7447A. This does not deny the performance shift, but it does show that the G5 is not necessarily as amazing in single processor systems as the G5 PowerBook zealots would have you believe.



More interestingly, the G5 loses on efficiency even when it has its support systems in place. Also, the towers suck up an astounding 420+ watts at peak operation, and that's without add-on cards or anything other than the parts inside the case when it ships.

Thanks for sorting me out thatwendigo, I don't think I was looking at quite the correct info on BareFeats - interesting! It still essentially sums up what you, I, and others here have been saying at least.