[Eugene Fields]

From an Article on the Web...

The 130 nm G5 970 is dead


I wrote in a previous article that the newest rev. B G5 2.0 GHz Power Mac was no longer using the older 130 nm G5 970 chip.

This has been confirmed. Typing "ioreg -l | grep cpu-version" in the Terminal yields an identical CPU ID in the new G5 2.0 as the G5 2.5, and both are different from the rev. A and apparently also different from the earliest rev. B G5 2.0 machines. Thus, all new rev. B machines (including likely the 1.8 GHz machines) seem to be utilizing the 90 nm 970FX chip.

Given that the new iMac G5 will also use this chip, there remains no machines in Apple's lineup that use the old 130 nm G5 970 chip. In other words, we have already moved completely to a new chip generation now, a year after the original G5 was introduced.

G5 970 RIP.


posted by Eug @ 11:03 PM

[Eug Wanker]

Heheh.

P.S. Not all Macs are 90 nm. All G5 Macs seem to be 90 nm though.

[Eugene Fields]

Originally posted by Eug Wanker:
Heheh.

P.S. Not all Macs are 90 nm. All G5 Macs seem to be 90 nm though.

He Point well taken buddy !! Thanx for the confirmation too.

This G5 is unnerving.. How Whisper Quiet it really is. Im in HEAVEN.

[tooki]

Wait... you mean... I mean... Apple moved to new technology?!? You're kidding, right?



tooki

[Eug Wanker]

Originally posted by tooki:
Wait... you mean... I mean... Apple moved to new technology?!? You're kidding, right?



tooki

Of course, you've missed the point that it's good news that Apple can have 3-5 day shipping at the online store for the G5 970FX 90 nm, while at the same time ramp up iMac G5 970FX 90 nm production for their September release.

While there might be some potential delays in G5 iMac shipments with the spike in orders after Apple Expo in Paris, it seems that the main problems with 90 nm G5s IBM has faced are now behind them.

On to 3 GHz we march!

[Big Mac]

What is the 90nm 970 identification code that is found by typing ioreg -l | grep cpu-version? I get <00390202>. Is it possible that the 90 refers to 90nm?

[Eug Wanker]

Originally posted by Big Mac:
What is the 90nm 970 identification code that is found by typing ioreg -l | grep cpu-version? I get <00390202>. Is it possible that the 90 refers to 90nm?

When did you buy your dual G5? AFAIK, if you get 00390202 when you type "ioreg -l | grep cpu-version, then you have the old 130 nm 970. The new 90 nm 970FX is 003c0300.

BTW, if you type "ioreg | egrep Power" you should get PowerMac7,3 in there somewhere for a rev. B.

ie.

Rev. A - PowerMac7,2 00390202
Old Rev. B - PowerMac7,3 00390202
New Rev. B - PowerMac7,3 003c0300

P.S. Don't ask me what any of this means.

[Big Mac]

Yeah, it does identify itself as a PowerMac7,3 on the first line, but it also says PowerMac7_2PlatformPlugin. And it does indeed report 390202. I ordered my G5 in mid-July and received it about ten days later. I guess it is the "old" Rev B. I purchased from the Apple Developer Store. At least we got to the bottom of the issue, and it does not bother me that I don't have a 90nm part, since it's not like it's faster anyway. I apologize to everyone I unintentionally misled when I reported that all Rev B G5s had the new part.

[Eugene Fields]

Originally posted by Eug Wanker:
Heheh.

P.S. Not all Macs are 90 nm. All G5 Macs seem to be 90 nm though.

Well Here I am championing the New Chips cause my New RevB is So Perfect Only to find I have the Old Chips!! The machines right after mine have the new ones.... What a Joke!!

But what am I to make of this, My Mac is the Best Mac I ever Owned Quietest except for my Cube. Heck this Tower is Quieter than my Veraxed 1.42 which I considered Silent. So I am left with the fact that is is not the New FX Chips that have so remakably transformed the RevB, it's the New MotherBoard? Perhaps in the RE-Engineer they changed the Threshold Points on all noise producing components to bring them Out a critical Resonance, to produce a way more refined experience, Or Im just very luck to have a Silent Mac? Well Im pleased as could be. The Tower, seems a Silent sculpture on my desk at ear level I hear Nothing. WOW Im very Impressed, FX or no... And the New 9800XT has a Verax Style GO3 Fan thats silent as well. Masterfull!

[Eriamjh]

From what I've been reading the 90nm G5s are having major supply problems from IBM. Betcha they wish they had some 130nm processors now.

[d.fine]

There's another thread where this is discussed

[Catfish_Man]

Originally posted by Eriamjh:
From what I've been reading the 90nm G5s are having major supply problems from IBM. Betcha they wish they had some 130nm processors now.

??? They do have 130 nm ones, lots of them. The 90 nm ones are the ones they wish they had (more of).

[Eug Wanker]

Apple is still quoting 3-5 business days for shipping of the dual 2.0 G5s (which are now 90 nm 970FX based). That's pretty good if you ask me.

[Landos Mustache]

Great, now we got two Eugene's obsessed with nm's

[unregistered]

When a chip is manufactured, it is tested for what speed at which it will reliably operate. It is possible that as IBM tries to produce 2.5Ghz chips, they are ending up with procs that will only test reliably at 2Ghz.

Why could IBM not still be producing 130nm processors to meet the 2Ghz demand and dumping the 90nm chips that won't run reliably at 2.5Ghz (but may at 2Ghz) into the same bin? If this were true, then Apple could be shipping a mix of 90nm and 130nm 2Ghz machines.

[Eugene Fields]

Originally posted by Landos Mustache:
Great, now we got two Eugene's obsessed with nm's

LOL

Certifiable....

[Eug Wanker]

Originally posted by unregistered:
When a chip is manufactured, it is tested for what speed at which it will reliably operate. It is possible that as IBM tries to produce 2.5Ghz chips, they are ending up with procs that will only test reliably at 2Ghz.

Why could IBM not still be producing 130nm processors to meet the 2Ghz demand and dumping the 90nm chips that won't run reliably at 2.5Ghz (but may at 2Ghz) into the same bin? If this were true, then Apple could be shipping a mix of 90nm and 130nm 2Ghz machines.

It'd make more sense to stick with the 90 nm process. It's potentially cheaper per chip, unless IBM is having bad yield issues with the 2.0. However, if they can produce 2.5 GHz in volume, they can most definitely produce the 2.0 in greater volume.

Plus it seems from looking around the net that lately everyone is getting the 003c0300 970FX 90 nm chip. We'll have to see more posts, but it does truly sound that they've scrapped the 130 nm process now completely for the G5.

BTW, the iMac G5 undoubtedly uses the 970FX. If it's an enclosure that has space restrictions, then a 1.8 GHz 970 130 nm would likely be too hot for it.

[Big Mac]

I was under the impression that the 90nm process didn't reduce power consumption and therefore heat by that much, but it did change the way the heat was dissipated (over a smaller area). Isn't that one of the reasons why IBM announced scaling is dead?

[Eug Wanker]

Originally posted by Big Mac:
I was under the impression that the 90nm process didn't reduce power consumption and therefore heat by that much, but it did change the way the heat was dissipated (over a smaller area). Isn't that one of the reasons why IBM announced scaling is dead?

IBM reduced heat dramatically with its 90 nm shrink. For instance, the G5 2.0 970FX uses roughly HALF the power of the G5 1.8 970. You're thinking Intel, who has 90 nm chips that are hotter than its 130 nm chips.

However, IBM's 90 nm chips still suffer from increased power density. They use less power per MHz, but more power per unit area in some parts of the chip. Hence, the need for liquid cooling in the 2.5, despite the fact it uses the same amount of power as the older G5 1.8 970.

BTW, Freescale is set to announce a dual-core G4-class chip on Oct. 5. The interesting part is that some claim it might be as low as 25 Watts at 1.5 GHz. That would make a G4 dual 1.5-1.8 GHz PowerBook possible. (I wouldn't count on it though. )

[Maflynn]

I'm not trying to sound like a jerk but I think this 130nm vs 90nm is much ado about nothing. My 2.0 (rev b ) runs fine whether its fabricated using 90nm technology or not. I saw a few people work themselves up in a lather in the apple forum a while back.

So far I've heard two explanations both exact opposites. The 90nm can consume less power thus produce less heat. vs. The 90nm produces the same amount of heat but since the area is smaller the heat is more acute.

All I know is the G5 needs 9 fans, or liquid cooling - that tells me this puppy is HOT. My G5 is awesome, it does everything I want it to.

Regards
Mike

[Catfish_Man]

Originally posted by Maflynn:
I'm not trying to sound like a jerk but I think this 130nm vs 90nm is much ado about nothing. My 2.0 (rev b ) runs fine whether its fabricated using 90nm technology or not. I saw a few people work themselves up in a lather in the apple forum a while back.

So far I've heard two explanations both exact opposites. The 90nm can consume less power thus produce less heat. vs. The 90nm produces the same amount of heat but since the area is smaller the heat is more acute.

All I know is the G5 needs 9 fans, or liquid cooling - that tells me this puppy is HOT. My G5 is awesome, it does everything I want it to.

Regards
Mike

PowerPC 970 typical power @ 2GHz: roughly 50 watts spread over about 120 mm^2
PowerPC 970fx typical power @ 2GHz: roughly 25 watts spread over about 66 mm^2

At 2.5Ghz I'd guess that the total draw of both processors and the northbridge is around 160 watts, which is quite hot (roughly equivalent to the power draw of the original dual 2GHz, but with smaller chips). If the northbridge is being made on .09 micron then it might be somewhat closer to 130 (assuming northbridge typical power scales similarly to the 970's).

The 9 fans is primarily to allow all the fans to run at low speed to keep the noise down. A 2.5GHz 970fx probably uses about as much power per unit area as Intel's Prescott chip (50ish watts over 66 mm^2 vs. 100ish watts over 110 mm^2).

[Eug Wanker]

Originally posted by Catfish_Man:
PowerPC 970 typical power @ 2GHz: roughly 50 watts spread over about 120 mm^2
PowerPC 970fx typical power @ 2GHz: roughly 25 watts spread over about 66 mm^2

At 2.5Ghz I'd guess that the total draw of both processors and the northbridge is around 160 watts, which is quite hot (roughly equivalent to the power draw of the original dual 2GHz, but with smaller chips). If the northbridge is being made on .09 micron then it might be somewhat closer to 130 (assuming northbridge typical power scales similarly to the 970's).

The 9 fans is primarily to allow all the fans to run at low speed to keep the noise down. A 2.5GHz 970fx probably uses about as much power per unit area as Intel's Prescott chip (50ish watts over 66 mm^2 vs. 100ish watts over 110 mm^2).

Actually:

Code:
Speed     CPU   Process   Typical power  Die size  Typ. power/area
------------------------------------------------------------------
1.8 GHz   970   130 nm     51   Watts    118 mm2      0.43 W/mm2  
2.0 GHz   970   130 nm     66   Watts    118 mm2      0.56 W/mm2
2.0 GHz  970FX   90 nm     24.5 Watts     66 mm2      0.37 W/mm2
2.5 GHz  970FX   90 nm     50   Watts     66 mm2      0.76 W/mm2

Note that the max power of these chips is MUCH higher than the typical power ratings, but you get the idea.

Originally posted by maflynn:
I'm not trying to sound like a jerk but I think this 130nm vs 90nm is much ado about nothing. My 2.0 (rev b ) runs fine whether its fabricated using 90nm technology or not. I saw a few people work themselves up in a lather in the apple forum a while back.

So far I've heard two explanations both exact opposites. The 90nm can consume less power thus produce less heat. vs. The 90nm produces the same amount of heat but since the area is smaller the heat is more acute.

All I know is the G5 needs 9 fans, or liquid cooling - that tells me this puppy is HOT. My G5 is awesome, it does everything I want it to.

As you can see, the 90 nm chips are much cooler than the 130 nm chips, at the same GHz. However, the 90 nm chips scale much higher. So while the new chips are cooler per GHz, the heat density is potentially significantly higher for high clocked chips.

So both of the explanations you've heard are essentially correct, depending on which chips you are talking about.

IOW, the 2.5 GHz chips are about as hot as the old 1.8 GHz G5 chips, while the new 2.0 GHz chips consume less than half the power of the old 2.0 GHz chips. However, the 2.5 GHz chips are only 56% of the die size of the old 1.8 GHz chips, so while the heat output is nearly the same, the heat density is much higher.

However, Catfish_Man is right. Remember, a lot of the fans in the case aren't even used to cool the CPU. And the multiple fans that are used to cool the CPU are there because Apple wants to keep the machine quiet. More fans = less air movement required per fan. Less air movement = less noise.

[Catfish_Man]

Originally posted by Eug Wanker:
Actually:

Code:
Speed     CPU   Process   Typical power  Die size  Typ. power/area
------------------------------------------------------------------
1.8 GHz   970   130 nm     51   Watts    118 mm2      0.43 W/mm2  
2.0 GHz   970   130 nm     66   Watts    118 mm2      0.56 W/mm2
2.0 GHz  970FX   90 nm     24.5 Watts     66 mm2      0.37 W/mm2
2.5 GHz  970FX   90 nm     50   Watts     66 mm2      0.76 W/mm2

Note that the max power of these chips is MUCH higher than the typical power ratings, but you get the idea.

close enough. Thanks for the exact numbers though. I had forgotten/hadn't heard that the 2GHz 970 was that high. IBM did some pretty impressive work on the 970fx getting it down to 25

[Spliffdaddy]

big difference between the thermal output of 2.0 & the 2.5

Did the 2.5 require a biggish voltage (Vcore) increase? If so, then that speaks volumes about the frequency limit of the current core. Obviously, it's about 2.5GHz - before adding more voltage yields heat, not stability. I might even suggest that the cores wouldn't attain 2.5GHz in 'normal' (air-cooled) conditions - hence the liquid cooling system. You hafta keep in mind that a P4 is even hotter than the G5 - and they use air-cooling.

If you assume that there's a also a big difference between "typical" and "maximum" heat dissipation - akin to the substantially different ratings of the G4 - then the 2.5 is generating something near 90watts under full load. Which translates to well over 1W per square millimeter of surface area. hello P4.

What if all P4s shipped with watercooling? They'd certainly be running at higher clockspeeds than they currently are.

[Big Mac]

Eug brings up some great points, and has numbers to back them up. On the other hand, the real world performance of the 90nm process have been less than stellar across the industry. The heat density of the chips has changed, and at such a small process the transistors apparently begin to leak the excess heat. AppleInsider has had terrific discussions on the subject - just search for "970fx or 90nm and heat." Since IBM is still having supply problems with the 970 (which should have theoretically been resolved weeks prior to the Rev B announcement), I think I'm glad I have an original 970. <Rampant Speculation>Eug states that many have recently received 970fx G5s. At the same time, we've seen an increase in G5 complaints. I don't know how strong the correlation between the two variables is, but it's something to consider</Rampant Speculation>

[FlatLyna]

Originally posted by Spliffdaddy:
You hafta keep in mind that a P4 is even hotter than the G5 - and they use air-cooling.

If you assume that there's a also a big difference between "typical" and "maximum" heat dissipation - akin to the substantially different ratings of the G4 - then the 2.5 is generating something near 90watts under full load. Which translates to well over 1W per square millimeter of surface area. hello P4.

What if all P4s shipped with watercooling? They'd certainly be running at higher clockspeeds than they currently are. [/B]

..but how many pc's ship with DUAL P4's ? Xeon's maybe in workstations, so the G5 has twice the heat output to cool. The noise level would be a major diffrentiator here as moving ~180 wattswith aircooling from your average pc would a noisy job ;-)
Also if you look into the commercial compact water cooling setups that many companys are rushing to produce, few offer drastically increased cooling capability, typically just a drop in noise.

[Eug Wanker]

Originally posted by Spliffdaddy:
big difference between the thermal output of 2.0 & the 2.5

Did the 2.5 require a biggish voltage (Vcore) increase?

The numbers are not published, but as you said, judging by the numbers it's safe to say it requires a significant voltage boost.

If so, then that speaks volumes about the frequency limit of the current core. Obviously, it's about 2.5GHz - before adding more voltage yields heat, not stability.

Perhaps, but that's with the current chip revision.

I might even suggest that the cores wouldn't attain 2.5GHz in 'normal' (air-cooled) conditions - hence the liquid cooling system.

Apple has already said that air-cooling of the 2.5 was problematic, not because of the overall heat, but because of the heat density.

You hafta keep in mind that a P4 is even hotter than the G5 - and they use air-cooling.

Yeah, but Intel's P4 90 nm is 122 mm, which is 85% larger than the G5's die size. Thus the P4's heat density is lower, even though total power output is significantly higher. Also, one might suspect that Intel has spent more time hand-massaging the design to minimize hot spots, which IBM might not have been able to do because time-to-market concerns. Automated chip design is fast, but ain't perfect obviously.

If you assume that there's a also a big difference between "typical" and "maximum" heat dissipation - akin to the substantially different ratings of the G4 - then the 2.5 is generating something near 90watts under full load. Which translates to well over 1W per square millimeter of surface area. hello P4.

The 90ish Watts under full load seems about right for the G5 2.5. However, IBM doesn't give us any official number.

What if all P4s shipped with watercooling? They'd certainly be running at higher clockspeeds than they currently are.

Perhaps, but the point is moot, since they aren't.

Originally posted by Big Mac:
<Rampant Speculation>Eug states that many have recently received 970fx G5s. At the same time, we've seen an increase in G5 complaints. I don't know how strong the correlation between the two variables is, but it's something to consider</Rampant Speculation>

What complaints?

Originally posted by FlatLyna:
..but how many pc's ship with DUAL P4's ? Xeon's maybe in workstations, so the G5 has twice the heat output to cool. The noise level would be a major diffrentiator here as moving ~180 wattswith aircooling from your average pc would a noisy job ;-)
Also if you look into the commercial compact water cooling setups that many companys are rushing to produce, few offer drastically increased cooling capability, typically just a drop in noise.

Good point. One cannot direct compare the cooling of one P4 vs two G5s. That said, even with one G5, it's likely Apple would have wanted liquid cooling, for the heat density issue.

[Big Mac]

Concerning the increase in G5 complaints, I have seen a disturbing number of complaints from new G5 owners in the last two to three weeks on this very forum. Freezes of every sort - including fan-blast-freezes to freezes in applications. I did not previously see such a large number of issues. If the people complaining are using 970fx G5s, then I have to imagine there is some positive correlation.

[Eug Wanker]

Originally posted by Big Mac:
Concerning the increase in G5 complaints, I have seen a disturbing number of complaints from new G5 owners in the last two to three weeks on this very forum. Freezes of every sort - including fan-blast-freezes to freezes in applications. I did not previously see such a large number of issues. If the people complaining are using 970fx G5s, then I have to imagine there is some positive correlation.

Maybe, but it could always be due to the new PowerMac7,3 motherboard too (which runs both the 970FX and the 970). You could always start a poll.

[Eugene Fields]

Originally posted by Eug Wanker:
Maybe, but it could always be due to the new PowerMac7,3 motherboard too (which runs both the 970FX and the 970). You could always start a poll.

Well in a sense I feel fortunate that beacause I have the Old Chips. I really have a RevB, as thi is the second Go round that the Engineers have had with my Chip. The FX Variety is more like a revA, in that Those chips, had never seen production, so its possible that the new chipsets may be Noisier than the Old as we have the second attempt at Silence while the FX are on their first outing - in production. The noises generating from the recources allocated to Cool the chip and the Threshold Points that may need adjustment for the new Chips, that now work better, with the former gneration Chips. Is this possible? As I never hear any Fan activity at all. Alot of the new users are asking about Fan Ramp Up when using Safari Etc., I never get this...

[Eugene Fields]

Well in a sense I feel fortunate that beacause I have the Old Chips. I really have a RevB, as this is the second Go round that the Engineers have had with the Chip. The FX Variety is more like a revA, in that Those chips, had never seen production, so its possible that the new chipsets may be Noisier than the Old as we have the second attempt at Silence while the FX are on their first outing - in production. The noises generating from the recources allocated to Cool the chip and the Threshold Points that may need adjustment for the new Chips, that now work better, with the former gneration Chips. Is this possible? As I never hear any Fan activity at all. Alot of the new users are asking about Fan Ramp Up when using Safari Etc., I never get this...

My consolation prize, If I have to have the old stuff at least its totally silent. ALso now by being in the middle of the range of Mhz ,the 2, may be quieter than the 2.5, as the stress levels are lower on the PS etc..

[Big Mac]

I like your reasoning, Eugene. I love my Rev B 970 G5. Strangely, the only time I hear a small increase in fan speed for a minute or so is, ironically, during non-resource intense productivity work. I can play UT2004 or WC3 for an hour and not hear any change in fan speed.

[Graymalkin]

I want a dual G5, I'm not particular about the processor stepping. I'm also taking donations. *pathetic face*

[Ratm]

Originally posted by Eug Wanker:
[BOn to 3 GHz we march! [/B]

Yes!!! saving the pennies as we speak. My current maching is the Blue&White G3 350mhz PowerMac with the zip drive

[Riemann Zeta]

So I guess the only thing that is holding up a 15/17" Powerbook G5 (90nm chip availability notwithstanding) is the super high-power northbridge/southbridge/DDR controller that IBM uses with the G5. It is still 130 nm, and uses a shitload of power (like 60W) right? If IBM can get that down to 90nm, I see no reason why a 1.8 GHz G5 Powerbook could not be easily produced.

Imagine how impressive a 15" PB with a 1.8GHz G5 and a Radeon Mobility 9800 would be...I'd rush to empty my bank account.

[ArizonaJoe]

How is Apple going to keep the new iMac cool, given all the fans and spaces needed on the G5 desktop? The new iMac is supposed to be a pizza box, everything in the same case as the LCD.

Joe

[Catfish_Man]

Originally posted by ArizonaJoe:
How is Apple going to keep the new iMac cool, given all the fans and spaces needed on the G5 desktop? The new iMac is supposed to be a pizza box, everything in the same case as the LCD.

Joe

Well, given the power numbers above a 1.8GHz 970fx should be <20 watts. The northbridge on the original G5 was about 60 watts iirc, and the procs were 66 each, so about 180 total. If the bus speed was cut to single 600MHz instead of dual 1000MHz that'd lower its power some. Going to single channel DDR333 and 90 nanometer manufacturing would help some as well. I'd guess the total output would be 40ish watts, and I think that's a fairly conservative guess, it could be lower.

Getting a little bit more fanciful, we could think about a hypothetical low power 970fx variant with an on chip memory controller (single channel DDR333 for lower power, if needed). Getting rid of that super high speed bus between the proc and the northbridge entirely could definitely help cut power, although it'd have to be replaced with something else to hook up to the rest of the system (probably hypertransport, anyone know how much power that takes?). I've heard that one reason the Pentium-M has a large cache is to lower the power draw of the bus by using it less, so perhaps we could hope for a 1MB cache as well. No idea if that would actually help though, it could be a myth.

[macaddict0001]

ding ding ding it sort of is a myth because actually the bus hardly uses any power compared to lcd hard drive processer even the cache uses power a bit. A much better thing to do would be to invent a solid state hard drive like a cartridge it would be a much bigger power improvement and it wouldn't be that hard. The only problem no os really boots from anything like, that but it could be adapted.

[Eug Wanker]

Originally posted by Eug Wanker:

Code:
Speed     CPU   Process   Typical power  Die size  Typ. power/area
------------------------------------------------------------------
1.8 GHz   970   130 nm     51   Watts    118 mm2      0.43 W/mm2  
2.0 GHz   970   130 nm     66   Watts    118 mm2      0.56 W/mm2
2.0 GHz  970FX   90 nm     24.5 Watts     66 mm2      0.37 W/mm2
2.5 GHz  970FX   90 nm     50   Watts     66 mm2      0.76 W/mm2

Note that the max power of these chips is MUCH higher than the typical power ratings, but you get the idea.

IBM has published more info on the power characteristics of the 970FX:



Max power is listed at 100 Watts for the 2.5, which is exactly twice the 50 Watt "typical" spec.

[Spliffdaddy]

ouch.

[Al G]

I just got a new rev. B DP 2.0 CTO. According to information from this (and other similar) threads, it is a late October 2004 build and has 970 processors, not 970FX.

I am bummed because I was hoping to save a bit on my electric bill. This machine also has a fan rev problem. I have the worst G5 karma.

Machine Model: Power Mac G5
CPU Type: PowerPC G5 (2.2)
Number Of CPUs: 2
CPU Speed: 2 GHz
L2 Cache (per CPU): 512 KB
Memory: 1.5 GB
Bus Speed: 1 GHz
Boot ROM Version: 5.1.8f5


% ioreg -l | grep cpu-version
| | | "cpu-version" = <00390202>
| | | "cpu-version" = <00390202>


Name: Power Macintosh G5 (June 2004)
Model: M9454 PowerMac G5 2.0GHz-DP
Bus speed: 1.0GHz
Factory: G8 (USA (where?))
URL: Technical specifications by apple-history.com
Production year: 2004
Production week: 43 (October)
Production number: 6872 (within this week)

[Eug Wanker]

Originally posted by Al G:
I just got a new rev. B DP 2.0 CTO. According to information from this (and other similar) threads, it is a late October 2004 build and has 970 processors, not 970FX.

I am bummed because I was hoping to save a bit on my electric bill. This machine also has a fan rev problem. I have the worst G5 karma.

Machine Model: Power Mac G5
CPU Type: PowerPC G5 (2.2)
Number Of CPUs: 2
CPU Speed: 2 GHz
L2 Cache (per CPU): 512 KB
Memory: 1.5 GB
Bus Speed: 1 GHz
Boot ROM Version: 5.1.8f5


% ioreg -l | grep cpu-version
| | | "cpu-version" = <00390202>
| | | "cpu-version" = <00390202>


Name: Power Macintosh G5 (June 2004)
Model: M9454 PowerMac G5 2.0GHz-DP
Bus speed: 1.0GHz
Factory: G8 (USA (where?))
URL: Technical specifications by apple-history.com
Production year: 2004
Production week: 43 (October)
Production number: 6872 (within this week)

Hmmm... This is very interesting. If this grep cpu-version info for 970 vs 970FX is correct, then IBM is still having severe problems with 970FX yields.