IBM's 90nm PowerPC 970FX

[macmunch]

Hooo you forgett something !

Why should Apple bring a PowerBook G5 that only cotains a G5 Chip and a new Architecture ?

Enclosure ! Same ?
Graphic Chip ! same ? ---> There is no new one from ATI or Nvidia
Battery ? ---> G5 chip and the same stupid battery ? 2 to 4 hours ?

Yup, right they dont will do !!! We will see PB G5 later this year and for me I think it will be a real Generation 5 Powerbook !

Like a new Graphic Chip
G5 processor
This new Power Technology (fuelcell or ?) Thosiba showed prototyp Laptops with it in CeBit 2003 I saw them for real !

So also Apple will do when the get the Technologies !

 

[stingerman]

Originally posted by 123
WTF are you talking about?

Simple integer instructions have 0.5 cycles latency and 0.5 cycles throughput (P4), which is fast. Double-pumping is a good thing (tm) and has nothing to do with the RISC/CISC debate or whatever you are talking about, end of discussion. "out of necesssity"... tsss.

Sounds great on paper, now go look at the comparative SpecINT scores and hit reality. Do your own due diligence before fly off the handle. The P4 is a great attempt to extend an aging ISA, but she's old and needs to be retired. Look at the W difference between IBM's unencumbered design compared against even the 90NM Prescot! The 90NM 970 has a lot more room to grow than the 90NM P4 design. This is more a reflection of the age of the respective products. IBM is working off of a clean slate using current best practices. Intel engineers are under the weight of a huge legacy, so of course IBM's engineers are free to innovate. Why do you think Intel came out with such a radical approach to their Itanium design? They know this and it is no reflection on them, more a consequence of history. IBM has the momentum right now and probably for at least the next two to three years.

 

[ionas]

battery power of the PB G5

the 12 and 15" versions must AT LEAST hold 4.5 hours while doing office and internet activly.

else i dont see a comparison possible to the upcoming centrino technology which will feature cpus cores with 2MB memory on die.

this said they would need to build in a very very large battery.

also the system has to be fanless during normal day use.

both requirements are hard to met with a 90nm g5 at speeds above 1.5 ghz.

so - suprise me apple - please - else you will lose your pretty notebook market slowly but steady.

 

[stingerman]

Originally posted by Kermit
Now, I do not want to ruin everybodys day here, but I still have to point it out. That 24+ W at 2.0 GHz is typical heat dissipation, meaning that the the number for maximum h.d. might be as much as 5-10 W more.

Keep in mind that Apple has a really cool dynamic bus slewing feature. So it can instantly scale the frequency of the processor. I also suspect that the PowerTune feature in the FX allows for Voltage scaling, which has an even greater impact on heat (a quadratic reduction, in addition to the linear reduction currently achieved with the frequency alone.)

If Apple applies the same technology to the GPU and the HD, you've got yourself one very low power and quiet Notebook.

 

[stingerman]

Originally posted by amir12

Although using the P-M wattage as a comparison, things seem optimistic for apple!

I also have a question for the more technical: Does anyone know how the current g4/g5's (and a prediction for the FX) scale processor wise. I mean in comparison to Intel's SpeedStep technology. Also how great are the power settings in OSX, or other third party battery management apps.

The P-M is actually a modified old P-III. As you mentioned the typical W for the P-M is greater than that of the 970FX, so this is very exciting for Apple - making a PB @2GHz a real possibility and looking like Apple will beat Intel to it!

The Apple architecture uses bus slewing and not the older stepping. The G5 dynamically changes its frequency based not only on power source but also usage. The more demand placed on it, the greater the frequency. Of course, Processor temperature is constantly monitored and the frequency will scale down if the temp gets too hot. The dynamic nature of Apple's bus slewing technology is a lot more efficient than the power source bound stepping in legacy products.

An added note is the new dynamic Voltage slewing that IBM has been writing about in their research papers and evidently is part of the FX "PowerTune" design. As the frequency decreases, the FX may be able to dynamically reduce the Voltage as well. Thus, the spec shows a range of 1-1.3V for the 970FX, where as the 970 is fixed at 1.3V. This has a much greater effect on heat than frequency alone.

One more IBM innovation that may have found its way into the FX is what IBM terms "Voltage Islands". The Core can be divided up into different regions, where certain regions or Islands can always be running at the lower voltage not withstanding the voltage of the Cache or the core processing logic.

 

[gekko513]

From The Register

The 970FX, meanwhile, consumes a mere 12.3W at 1.4GHz, paving the way for PowerBook G5s. That figure is comparable to the 7.5W a 1GHz consumption of the G4-class Motorola MPC7447 that drives the current PowerBook G4s.

 

[stingerman]

Originally posted by tortoise
Unfortunately, not even close. Latency matters a LOT for memory performance, and L3 cache will run rings around main memory for latency. To add more injury, the latency to main memory for the PPC970 is actually pretty mediocre as such things go, fully TWICE the latency of the AMD64 chips to main memory, and possibly worse than the P4EE (I don't know the latency numbers for this off the top of my head).

So no, the higher bandwidth is really no substitute. If the latency was comparable your comparison would be valid, but in practice they aren't even close.

In practice they matter a lot less than you would have us believe. Look ahead algorithms minimize and in many cases circumvent latency factors altogether. And, yes a faster bus does minimize latency issues. Too, the G5 bus is Switched with out-of-band management, which is much faster than the shared busses and contention slowdowns of the XEON systems.

Another factor affecting performance is bandwidth and the G5 architecture has greater bandwidth. So when you are dealing with larger data sets, latency is a much smaller component of the performance equation. Apple has designed the G5 to perform very well on multimedia applications, including photos, videos and music. Transferring large textures for 3-D realtime graphics greatly benefits as well.

So to simply speak of latency as the driver behind performance is misleading. I believe that the G5 has the best balance of Latency, Bandwidth and bus frequency for an overall faster system. Couple that with IBM's advanced look ahead algorithms and it isn't an issue at all.

 

[rog]

Well this is disappointing news. 7 months later and only 10% faster. Uh oh. Also, the 2GHz 970fx is still too much of a power hog to be used in a powerbook. While the 1.4 GHz is better, who wants such a slow, single processor machine in 2004? Remeber the G5 is only about 20% faster than a same clock speed G4. In altivec, it's no faster at all. Looks like they'll have to move to a 65nm process before fast G5s get into powerbooks. Very sad for pro portable users.

 

[stingerman]

Re: battery power of the PB G5

Originally posted by ionas

so - suprise me apple - please - else you will lose your pretty notebook market slowly but steady.

Why the Battle Front trolling? You expect a 2MB core P-M to use less W than the 970FX? That's madness, heat is a function of capacitance and the more transistors you put in there the more energy will be used and heat dissipated. The greater the number of transistors the greater the required voltage, the greater the heat! It's a vicious cycle for Intel. Evidently, the 970FX design isn't anywhere near its limits yet and cache tricks are not needed yet to get meaningful increases in performance.

The fact is that the P-M needs the larger Cache to scale performance wise. Simply increasing the frequency will not give the P-M the performance scaling the clock would have you believe. A higher clock requires deeper pipelines as the various stages need to be simplified, forcing additional stages to be added and thus the P-M will start looking more and more like the P4. Intel needs a new design altogether to keep up with the 970. No BF intended, the Pentium is a good design for its generation, but really we are in the year 2004 now!

 

[stingerman]

Originally posted by rog
Well this is disappointing news. 7 months later and only 10% faster. Uh oh. Also, the 2GHz 970fx is still too much of a power hog to be used in a powerbook. While the 1.4 GHz is better, who wants such a slow, single processor machine in 2004? Remeber the G5 is only about 20% faster than a same clock speed G4. In altivec, it's no faster at all. Looks like they'll have to move to a 65nm process before fast G5s get into powerbooks. Very sad for pro portable users.

Yet another troll joins in. Jokers.

 

[rog]

Originally posted by stingerman
Yet another troll joins in. Jokers.

Oh yes, I'm such a troll for not jumping for joy and drooling over every extra MHz that IBM/Apple can give us. Not everybody on this forum blindly praises Apple for everything they do. Sorry, but I'm telling it like it is. While the G5 is a drastic improvement over the G4, it hasn't gotten any faster in 7 months, it's not in the portable or consumer lines, and it isn't contributing to growth of the platform. It also hasn't resulted in price cuts needed to grow market share. The news posted at the top of this thread is simply underwhelming. In the real world (please join it) Apple is competing with PCs, not all of which run on celerons and windoze, that are drastically cheaper and faster, and the vast majority of consumers barely know what an OS is or what the implications of one are, or why OSX is vastly superior to anything else. Get over yourself.

 

[wizard]

Instead of looking at IBM's average power numbers look at Apples maximum power numbers for the XServe. 25 whats is bad enough in a laptop but doubling that is a big issue.

The problem with the numbers is this; the people who would most like to use the 970 in a laptop are going to be the ones least likely to be able to sustain the desired peformance. As soon as demand on the processor goes up, and along with it wattage, you will see some sort of bus throttling. In the end you would get better results from an IBook.

Dave

Originally posted by stingerman
Though I don't know what Apple will do, technically the G5 is more than capable to go into a 1" PB, even the PB 12". Re-look at those W comparisons between the G4 and the G5. These numbers tell us that Apple has a lot to work with. They can easily put a 2GHz 970FX into a 15" and 17". They can probably bump the 12" to 1.6.

More than that, the FX will easily go to 3GHz+ if you compare it to the W that the P4, even the Prescott is planning to use (120W+!). I would even venture that 4GHz is not that far away for the PowerMacs. Apple really has a loaded gun here. What a turn of events. Maybe we will see Apple surpass Intel in GHz in addition to raw performance by the end of the year.

 

[tace]

Number of transistors?

Originally posted by Samurai980
Well it is nice to see IBM try to improve things, but why still such a low amount of L1, L2, and L3 cache?

The Pentium 4 was soundily defeated by the AMD 64 FX-51 (L1:128k,L2:1MB) and then Intel released the Pentium 4 Extreme Edition which is just a P4 with L1: 8k, L2: 512K, and a L3: 2MB and then it soundily defeated the AMD 64.

Now the PowerPC 970 has a measily 32K L1 and a tiny 512K L2. Why doesn't IBM/Apple upgrade this tiny thing? I mean there would be a significant speed and performance boost as a result.

Another interesting difference is the transistor count of the three processors. P4EE is 178m, AMD64 is 106m, and Apple is 58m.

Aside from that, Apple needs to put out a true 64-bit OS.

You can't directly compare the number of transistors of the 3 chips considering one of them is a RISC and the other is CICS.

Also, 64 bit OS is all nice but it is a bit more involved than you might think to release such a thing out the door as a commercial product.

 

[isgoed]

Originally posted by jderman
Beef up the battery in the current unit and wallah,

Hi Mr. American,

This is pretty funny: It is NOT "wallah" . The correct spelling is "voilá". It is French.

(for post see page 1, one of the first posts)

signed,

Mr. European

 

[wizard]

You are not even warm here. Intels solution to their ALU design issues are a joke.

An ideal ALU would execute an instruction in a single cycle. Now granted not many processor do that anymore due to long pipe lines to get things done. In Intels case they have extremely long pipelines so double pumped ALU's give them nothing over the competition. In fact those long pipe lines could be considered a liability.

There is nothing cool about the P4, it was and is a processor designed to run at high clock rates for marketing purpose irregardless of the benefits of those high speeds. To think otherwise is to be a slave to the manipulation of the marketing droids.

Originally posted by tortoise
Errmm... Double pumped ALUs are bleeding edge technology, and increase the performance of the CPU at relatively little cost. Everyone could benefit from this technology. And it makes sense if you consider it; the ALU is simpler than some of the other pipelines, and if you can manage to drive it twice as hard as the other pipelines you have a nice performance improvement on your hands on the integer side with only marginally more chip complexity. It is a terrible stretch to call either the AMD64 seires or the 970 inefficient, especially the AMD64 which is a performance leader with respect to intel hardware running much faster.

Thanks
Dave


You make it sound like it is a bad thing, but it is actually a cool thing. Mind you, I don't think Intel's cores are all that efficient in many ways, but they have found a way to exploit a design inefficiency that currently exists in both the PPC970 and AMD64 CPUs. I seem to vaguely remember reading somewhere that AMD was working on double-pumped ALUs, and they have no difficulty performing as well or better than the PPC970 clock for clock in their current incarnations.

Expect to see double-pumped ALUs in future AMD and IBM chips. It makes too much sense for them not to do it too.

 

[isgoed]

Originally posted by stingerman
As far as W usage, all I can say is WOW. IBM has pulled off a technological miracle. According to Motorola's charts the 970FX dissipates as mush W at 2GHz as the G4 7447 at 1.3GHZ! Now we know that the G5 architecture is a lot more efficient heat and energy wise than the G4 architecture. And, the 970FX can evidently dynamically scale not only frequency but also voltage. The chart implies that the FX will dynamically scale from 1.4-2GHz and 1-1.3V.

This is really a huge advance. The only unknown is the ASIC Controller. But, we have to believe it has been refined to the 90NM process as well. This means that the 970FX will easily go into Apple's entire line of computers. This is what Steve Jobs probably meant when he said that the G5 was the MACs future at the last show. It is exactly what Apple needs to do and quickly!

Can i quote this post 20 times and just add "I agree" to it?

No, i will probably get banned.

Ok, then i just do it once.

I agree.

 

[isgoed]

my 2.7Ghz G5

As soon as a cheap 90nm G5 (iMac/eMac/Cube) is out, I buy it. I set the bus to 900mhz and overclock the processor to 2.7Ghz (3X). This machine will once and for all burn all those windows users of the face of the earth.

 

[Snowy_River]

Originally posted by singletrack
The CPU isn't the only thing that draws a lot of power. A big system chip and fast RAM draws considerable power also. I recall Apple said the system chip in the G5 desktop was a big power drain also.

So it's nice the CPU size and power requirements have come down but that's only a part of the equation.

Yes, but you can bet that Apple hasn't been taking a sequential approach to this. That is to say, it's not a matter of 'okay, the chip is scaled down to a reasonable wattage, now let's work on the system controller.' I'd bet dollars to doughnuts that Apple has been working on a PB compatible system controller for at least six months, if not more.

G5 PBs may not be quite so far away after all...

"There's somethin' movin' out here and it ain't us!"

"Hudson may be right..."

 

[Snowy_River]

Re: New PowerMacs, sure; iMacs, maybe; PowerBooks, no way.

Originally posted by Sol
...While I would not go so far as to predict G5 powerbooks in the next revision, I see no reason now why iMacs could not use these new processors...

I've heard others make the argument, and I think that I agree, that the iMac won't exceed the power of the PB. This would be clearly placing a consumer level machine at a more powerful level than a Pro level machine. Since the introduction of the iMac that's never been done. The PB's already on somewhat shaky ground with the iBook barking up its tail-pipe. I'm sure Apple is doing everything they can to get the PB updated as soon as possible, and the iMac would follow shortly there after.

Of course, this is all just a guess...

 

[BenRoethig]

Re: Re: another reason this is good for apple...

Originally posted by stingerman
I totally agree that this will give Apple a lot of flexibility in design. What I think we will see happen to the PowerMac G5 is that the main CPU compartment will shrink. Maybe only one Fan instead of two. This will allow the G5 to have two optical drives, 4 SATA HDs and an additional PCI-X slot.

It will also allow for the G5 to enter the consumer line in a big way with smaller and radical designs. That really is what the 20th Anniversary Mac needs to be about, the consumer line which needs the most help right now.

It would be nice. I like the speed and design of the G5, but I liked the functionality of the last generation G4s better. 2 Optical drives is a must.

 

[Mac-Xpert]

I do think that a g5 Powerbook will need to run at least at 1.6 ghz to make it a impressive update from the current 1.3 Ghz G4. As others have mentioned the G5 is not that much faster at the same clock speed as the G4. So a 1.4 Ghz G5 Powerbook would for sure disappoint people in it's performance.

But looking at Intels numbers on the centrino they might be able to come up with a 1.6 anyway. Maybe that rumored liquid cooling is ready to debut in the next Powerbook.

 

[manu chao]

About power consumption

Funny how everybody is quoting different numbers for apparently the same processor...

I don't claim to understand all this, but what about stating
- what model of the G4 (7455 or 7447/57)
- what process (180 or 130nm)
- what core voltage (1.1 or 1.3 or 1.6v) (not I/O voltage)
- whether it's typical or max consumption
- and maybe a source
when quoting any numbers?

Here are some numbers I found:

- 7447/7457, 130nm, 1.1v, typical (my interpretation), 1Ghz
--> 7.7 Watt
e-www.motorola.com/files/32bit/doc/app_note/AN2436.pdf

- dito for 1.3v, typical/max
--> 15.8/22 Watt
e-www.motorola.com/files/32bit/ doc/data_sheet/MPC7457EC.pdf

- 970FX, 90nm, 1.0v, typical, 1.4 Ghz
--> 12.3 Watt
(from the pdf which started this thread)

Whether Apple is using the 1.1v or 1.3v parts in their laptops, I don't know.

So far for the facts, now to the speculation. Voltage seems to matter a lot, as it was shown earlier in this thread, the Pentium M does not have a significantly lower power consumption at its top frequencies, but drops a lot when it scales down the frequency, which it does by reducing the voltage (at least partly). The 970 can also reduce the frequency but I have found no indication that it uses a reduced voltage for that. Therefore, it might not be as effective as the Pentium M in saving power when the processor idles.
Note that Intel uses the term 'Thermal design', and I have no idea whether that corresponds to typical or max, or something in between.

 

[Dont Hurt Me]

Re: Re: New PowerMacs, sure; iMacs, maybe; PowerBooks, no way.

Originally posted by Snowy_River
I've heard others make the argument, and I think that I agree, that the iMac won't exceed the power of the PB. This would be clearly placing a consumer level machine at a more powerful level than a Pro level machine. Since the introduction of the iMac that's never been done. The PB's already on somewhat shaky ground with the iBook barking up its tail-pipe. I'm sure Apple is doing everything they can to get the PB updated as soon as possible, and the iMac would follow shortly there after.

Of course, this is all just a guess...

this is exactly why apple isnt growing marketshare, games being played between product lines. they need to stop this silly thing and market best product period. all they do when they play these games is hurt themself. put a top speed G5 into Imac and they wouldnt be able to keep up with sales. continue the game of cant do this because of that product and watch that market get smaller and smaller. this is what has happened the last 2 years.

 

[Hattig]

Originally posted by nighthawk
Originally posted by csubear

Also, the data cache is only 32kb, the instruction cache uses a separate 64kb. The spec sheet for the P4 Extreme does not list a separate amount for instruction cache. I believe that this is one of the weaknesses of the x86 architecture -- it does not include a separate cache for instruction data. (Correct me if I am wrong, but isn't this the source of the data buffer overrun error that plagues PCs?)

x86 processors also have instruction caches. Athlon/Athlon64 has 64KB Instruction cache and 64KB data cache. P4 stores decoded instructions in a cache as well, around 8000 instructions worth on the P4, 12000 on later revisions. This saves having to do the decode stages again, which is a neat optimisation.

In addition, the P4 Extreme's L3 cache is running at 800mhz or 1/4 the speed of the processor. I could not locate the speed of the L1 and L2 for the P4e, but IBM explicitly states that both the L1 and L2 cache runs at the full processor speed. The data-sheet for the P4e does describe the two ALU (integer calcs) as running at "twice the core clock" which is 1600mhz for the P4e. This is also unlike the 970xx which runs at the full clock speed.

When the core clock of the P4 is 3.2GHz, that means that the ALU is running at 6.4GHz.

The P4 and the Athlon/Athlon64 both run their L1 and L2 caches at core speed. The P4 also has very low latency for its caches. Intel is renowned for its cache design.

 

[Snowy_River]

Originally posted by jderman
Guys,

It's not gonna happen tomorrow, or even day after tomorrow, but it will happen soon. They got over the big heat/power hurdel....
It's coming

Has it occurred to you that they got over that hurdle a while ago? Long enough ago that they've had time to redesign both the mother board and the case of the Xserve? What else have they had time to work on in that time frame?