Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

stockscalper

macrumors 6502a
Aug 1, 2003
917
235
Area 51
The 970FX does draw only 12.3 watts of power, but what hasn't been mentioned is that it has automatic speed cycle reduction processing which means the cpu automatically lowers the mhz if nothing heavy is being run thereby reducing the wattage being drawn even further. This chip is ideal for use in a Powerbook. I've been saying all along that the .90 micron chip would be the first in a PB. BTW, Intel's new .90 chip draws 100 watts! :confused: You just thought the G5 ran hot; any computer running Prescott will take 200 fans to cool it and be a big as the QEII :D
 

johnnyjibbs

macrumors 68030
Sep 18, 2003
2,964
122
London, UK
What quantities are we looking at here?

Do they have enough of these chips to supply 2.4GHz ones in the PowerMacs and 1.4 or 1.6GHz ones in the PowerBooks?

I'd have thought the PowerMacs would be updated before the PowerBooks. Apple may have to pick and choose due to supply issues.
 

Snowy_River

macrumors 68030
Jul 17, 2002
2,520
0
Corvallis, OR
Originally posted by stockscalper
I've been saying all along that the .90 micron chip would be the first in a PB.

Not to be too nit-picky, but I don't think you mean .90 micron. The 970 is on a .13 micron process. In other words, a 130 nanometer (nm) process. A .90 micron process would be a 900 nm process, a factor of 10 larger than the 90 nm process that the 970FX is on. ;)
 

PPCTech

macrumors newbie
Jan 22, 2004
11
0
If you read the PPC Product Brochure, there is no indication that 2.4GHz will be available anytime soon, that is pure speculation. What is not speculation is the fact that this brochure confirms 2.2GHz PPC 970FX being available NOW. Read page nine:

http://www-306.ibm.com/chips/techli...B287256BF3006FBE54/$file/PPC_QRG_01-21-04.pdf

You will notice that the BUS speed listed is 1.1GHz, and since the PPC 970 line works on a 2:1 ratio, that signifies that the 2.2GHz is ready to go now.

Yes, the 2.4GHz+ are probably sampling, but don't be holding your breath for them until March/April.

-PPCTech
 

Henriok

macrumors regular
Feb 19, 2002
226
14
Gothenburg, Sweden
Originally posted by PPCTech
If you read the PPC Product Brochure, there is no indication that 2.4GHz will be available anytime soon, that is pure speculation.
Notice also that IBM didn't recognize the 2 GHz part of regular 970 until waaay after Apple introduced them.
 

stockscalper

macrumors 6502a
Aug 1, 2003
917
235
Area 51
The 970FX is currently being used in the G5 XServe is it not? If so then they are already in the manufacturing process. I still believe it will be some time before they make themselves into a Powerbook - maybe 4th QTR or later (MacBidouille says Mar o5). Look for dual 1.25 G4's in the meantime.
 

PPCTech

macrumors newbie
Jan 22, 2004
11
0
True to a point,

But since the PPC 970 was a brand new CPU line from IBM, very little was known in general, and most of the information coming out was saying 1.4-1.8 was max you would see, which probably totally depended on IBM bin rate of the new CPU itself.

Remember, we were also supposed to see the .13um versions of the PPC 970 all the way up to 2.6 before the 90nm process CPU's even became available, suggesting that there could have been trouble ramping up the CPU higher than expected, and the quick fix of transitioning to 90nm was their fix to get the PPC 970 line moving up again, but then again it could have just all been about economics.

Now that the PPC 970 isn't a "new" product, and more OEM's are picking up the CPU for various applications, you would think if there were significantly higher clocked PPC 970FX's floating around, you would have heard something by now about them. I am guessing that it will be a few more months until IBM will start pushing 2.4+ out the door, simply because the move to 90nm process is obviously a difficult one.

Intel for example having problem pushing the Prescott out on time, along with massive leakage resulting in a massive 103 watts of heat.

AMD has yet to move to 90nm, as TheInquirer has reported that they have been having trouble transitioning to 90nm as well (and is the same process that IBM is using because they are in a Cross Licensing agreement.)

I firmly believe 2.2GHz PPC 970FX are out now and limited supply of 2.4+ are even in Apple's hands.

And even if Apple did announce a 2.4+ PowerMac G5 revision in the next few days, you still won't see them ship until March/April... much as they did with the initial batch of G5's which were supposed to ship in August.

I ordered my Dual G5 shortly after their announcement, and didn't receive it until 2 1/2 months later.

-PPCTech
 

PPCTech

macrumors newbie
Jan 22, 2004
11
0
Yes stockscalper,

The PPC 970FX (90nm version of the PPC 970) is currently in the Xserve at 2.0GHz.

My question is, if Steve wants to hit the 3GHz marker by mid-summer, they need to put a revision out fast. If Apple was ready now to put out a new revision, why would they totally miss the MacWorld SF to announce such a huge product? Is waiting for the 20 year of the Mac really that big of a deal?

I don't believe so.... I believe that IBM being the first producer of 90nm parts, made a successful change to the new process, but has not totally refined it to super high clock speeds yet, and it will be a few more months before they make a new batch that truly shine and hit the 2.4GHz+ mark.

-PPCTech
 

Henriok

macrumors regular
Feb 19, 2002
226
14
Gothenburg, Sweden
Originally posted by PPCTech
Remember, we were also supposed to see the .13um versions of the PPC 970 all the way up to 2.6 before the 90nm process CPU's even became available
No, I have seen no documents stating that. All I have seen it that IBM have stated that the 970 processors would reach 2.5+ GHz, with no mentioning of any fabrication process. I have always interpreted that they'd reach the higher clocks at a later point on a new fab.. ie 90 nm.

Originally posted by PPCTech
Now that the PPC 970 isn't a "new" product, and more OEM's are picking up the CPU for various applications, you would think if there were significantly higher clocked PPC 970FX's floating around, you would have heard something by now about them.
I don't know about that. Lot of information of this type is covered by heavy NDAs, which probably have been lifted now. If you look at Apple's relationship with Motorola, Apple have always been able to get procesosrs clocked way higher than what Motorola have had available to ordinairy customers. Apple getting 1.42 GHz parts when Moto them selves only says 1 GHz.

Originally posted by PPCTech
the move to 90nm process is obviously a difficult one.

Intel for example having problem pushing the Prescott out on time, along with massive leakage resulting in a massive 103 watts of heat.

AMD has yet to move to 90nm, as TheInquirer has reported that they have been having trouble transitioning to 90nm as well (and is the same process that IBM is using because they are in a Cross Licensing agreement.)
The troubles Intel are facing are completely different from what IBM and Moto are facing. IBM and AMD are using mature SOI technology to counteract the very thing Intel is struggling with, and the numbers that we've now seen with 970 concerning wattage suggest's that IBM's 90 nm fab is in VERY good health.

IBM's new facility in East Fishkill is very good, and built for making 90 nm parts and beyond. AMD's 130 nm fab was co-developed with Motorola and.. well.. it was developed with Motorola, nuff said. Even if they now have a licenced stuff from IBM they most certainly will lag IBM in deployment. And.. pushing for 90 nm in the summer isn't bad at all. AMD is in great shape AFAIK. Intel on the other hand must get their act together quickly. +100 W for a Prescott ain't cool at all.

Originally posted by PPCTech
I ordered my Dual G5 shortly after their announcement, and didn't receive it until 2 1/2 months later.
I ordered my 2x2 the morning after the keynote and i received mine 3.5 months later.. you guys in the US are prioritized, don't we know it? :)
 

Henriok

macrumors regular
Feb 19, 2002
226
14
Gothenburg, Sweden
Originally posted by PPCTech
My question is, if Steve wants to hit the 3GHz marker by mid-summer, they need to put a revision out fast.
IBM won't reach 3 GHz with 970FX. Steve talked about "the next generation processor" and FX isn't next generation, its almost exactly the same, fabbed on a new process.

The "next generation" will be 975, and it is targeted at +3 GHz.
 

PPCTech

macrumors newbie
Jan 22, 2004
11
0
Henriok,
No, I have seen no documents stating that. All I have seen it that IBM have stated that the 970 processors would reach 2.5+ GHz, with no mentioning of any fabrication process. I have always interpreted that they'd reach the higher clocks at a later point on a new fab.. ie 90 nm.

In a press release from IBM themselves, which ArsTechnica posted, it said this:

http://www.arstechnica.com/news/posts/1046380113.html

The new IBM PowerPC 970 is the heart of the PowerPC Blade. It is based on the 64-Bit Power 4 architecture which is also used in the processors of the IBM eServer pSeries. The 64-bit microprozessor

Offers full symmetrical multi-processing

Has a high reliability (with parity L1, ECC L2 and parity checked system bus)

Is manufactured in the latest 0,13 micrometer Copper/SOI CMOS technology

Runs at frequences ranging from 1.8 GHz - 2.5 Ghz


By IBM claiming that the PPC on a .13um process hitting 2.5GHz, I mistyped and put 2.6GHz in my other post, I took it that they are infact transitioning quicker, or problems with the .13um process.


I don't know about that. Lot of information of this type is covered by heavy NDAs, which probably have been lifted now. If you look at Apple's relationship with Motorola, Apple have always been able to get procesosrs clocked way higher than what Motorola have had available to ordinairy customers. Apple getting 1.42 GHz parts when Moto them selves only says 1 GHz.

Moto themselves never targeted their CPU's beyond smaller embedded applications, and high end networking equipment OEM's. IBM on the other hand, are pushing their processors for various high end applications in computing. In their own workstations, servers, blades. As well as targeting outside smaller technology companies actively pushing Linux. Navigate through the IBM Microelectronics website to the December 2003 Journal, and see the long list of companies actively using PPC CPU's. With IBM, Apple, and lots of other small companies trying to push PPC, someone is gonna leak something somewhere, or new products were start surfacing right away if the were on the market.


The troubles Intel are facing are completely different from what IBM and Moto are facing. IBM and AMD are using mature SOI technology to counteract the very thing Intel is struggling with, and the numbers that we've now seen with 970 concerning wattage suggest's that IBM's 90 nm fab is in VERY good health.

IBM's new facility in East Fishkill is very good, and built for making 90 nm parts and beyond. AMD's 130 nm fab was co-developed with Motorola and.. well.. it was developed with Motorola, nuff said. Even if they now have a licenced stuff from IBM they most certainly will lag IBM in deployment. And.. pushing for 90 nm in the summer isn't bad at all. AMD is in great shape AFAIK. Intel on the other hand must get their act together quickly. +100 W for a Prescott ain't cool at all.

Yes, SOI is a proven concept that both AMD and IBM have used in their latest processor designs, but it doesn't matter how mature the process is, once you transition to a new smaller process, it takes time to perfect the design. Case and point, AMD released "Hammer" samples at .13um 8 months or so before it launched, and the highest speed they could demo was 800MHz, later they launched at 1.4, 1.6, and few weeks later 1.8 rolled out, at that time .13um was still new to AMD, and they had to pay IBM $45M to bail them out to fix their process problems with it. By the way, the .13um was not co-developed with Moto as you said, their .18 copper interconnect was co-developed with Moto.

With IBM moving to 90nm and being so successful with .13um, does not mean instantly they will be able to crank up the MHz, it will take time before the process is refined enough for them to push the clock rate much higher.

I guess I am just worried as to the reason IBM made such a quick transition to 90nm, and that worried me that the .13um process for the 970 was leading to major problems.


The "next generation" will be 975, and it is targeted at +3 GHz.

Again, the naming scheme is still unconfirmed, and speculation at this part.

-PPCTech
 

wms121

macrumors regular
Mar 1, 2003
104
0
ppc990/POWER 8 rumors and the POWER 12

....macosrumors had a list ..that went up to the PowerPC 990 and the POWER 8.

This still validates my earlier rumor that DoD has a POWER 12 and it's a optical chip.

The POWER 8 was said to be at 32nm and 15 GHz.

POWER 12 optical with SSOI substrate was in the 20-25 GHz range, but not clear on the wafer scale.

12.5 nm?

Maybe that's why my "a51 buddy" called it a ..ahem.."P12".

<---meet me at http://www.treknology.org where I can answer those "warp drive questions" you guys have..and no one will laugh.
 

Sun Baked

macrumors G5
May 19, 2002
14,938
157
Re: ppc990/POWER 8 rumors and the POWER 12

Originally posted by wms121
....macosrumors had a list ..that went up to the PowerPC 990 and the POWER 8.

This still validates my earlier rumor that DoD has a POWER 12 and it's a optical chip.
One more time MacOSrumors is a slightly more reliable site for you Mac Rumor dollar than Mad Magazine.

So how does it validate the POWER 12?
 

Ge4-ce

macrumors regular
Mar 25, 2003
190
0
Belgium
If Apple would announce G5's at 2.2 Ghz, it would be a disaster! That's only a 10% speed increase!! If you save the money to buy a middle system and spend it to extra RAM you gain more speed!!!

No, they will have to release a 2.4 GHz at LEAST! This device also has to last until June or July before they announce the 3 Ghz! Nobody will buy a 2.2 Ghz when you can buy a 3 Ghz only half a year later.. (ok, they will sell it, but...) the jump is too big!

and Apple has never updated PowerMacs that soon... always at least 3 or 4 months, and mostly... half a year between updates!

I'm still hoping for a 2.6 though... (I've decided to buy the next powermac.. whatever it is...) Unless it's a 2.2 Ghz.. Then I'll wait!

And I would be angry.. but hell.. I'll live..
 

Rocketman

macrumors 603
Originally posted by PPCTech
If you read the PPC Product Brochure, there is no indication that 2.4GHz will be available anytime soon, that is pure speculation. What is not speculation is the fact that this brochure confirms 2.2GHz PPC 970FX being available NOW. Read page nine:

http://www-306.ibm.com/chips/techli...B287256BF3006FBE54/$file/PPC_QRG_01-21-04.pdf

You will notice that the BUS speed listed is 1.1GHz, and since the PPC 970 line works on a 2:1 ratio, that signifies that the 2.2GHz is ready to go now.

Yes, the 2.4GHz+ are probably sampling, but don't be holding your breath for them until March/April.

-PPCTech

Agreed. Supportedby theimagepostedtothisthread of PowerPC 970 & 970FX Processors specpage.

Chip name 970-1.8 970FX-2.0 970FX-2.2
SPECint2000 828 890 979
Dhrystone 2.1 7584 5800 6380
Typ Power 51W 24.5W 27W

Rocketman

Deliver my Powerbook G5 please.
 

yamabushi

macrumors 65816
Oct 6, 2003
1,009
1
2.4 won't have nearly the same impact in March or April as it would have today. If they have to wait that long for 2.4 then the 2.2 better start shipping soon. The 2.0 is old news.
 

wms121

macrumors regular
Mar 1, 2003
104
0
POWER 12 's and alien spaceships...

..this was from a "old associate of mine"..who worked for a "EMP/TEMPEST" vendor "'in the desert...'"..ahem..

They needed power convertors that might see megajoule pulse behavior over a interconnected grid. Their "EMP signature" was..
(ahem)..very nonstandard..and fragile.

He mentioned that the IBM rep he (sort of) knew..had carried "POWER6 opti-prototypes" at very high clock speeds. That
IBM was considering using the "architecture" for the "POWER 12" and beyond..in that the binding process had unique thermal qualities (for low-temp chips that is).

I checked to see what type of "optical chip" IBM might be using and found references to a Japanese plant doing Si-Nitride and
fabric opti-switch network processors. I could post the URL's again..but the macosrumors article was the first I had heard of working 32nm geometries..and companion devices.

It's the architecture that is significant...these "desert people" needed "quantum-computing" capable interfaces...256-bit codes and possible 1024 bit emulators in the support ware.

They must have had "something" they needed to talk to real bad.

<<<-hiding under the bed with a flashlight
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.