970 Ready for Production?

[nighttraitor]

Year of the Laptop...

Hey all,

Long time reader, first time poster. With all this talk about the wattage used and heat disipation on the 970, and Uncle Steve's comment about this year being the year of the laptop, I can't help but agree with those out there who think that the 970 will be in the PB before the years end.

Just my 2 pennies.

 

[macrumors12345]

Originally posted by Rincewind42
For the G4 we have confirmed 15 watts at 1.067 Ghz (from Motorola's PDF) and a 'street' quote of 30 watts at 1Ghz.

Yes, I agree, we don't know for certain which version of the G4 is using. However, we can be sure that it is not dissipating 30 watts, because if it were the battery life of the PB G4 would be even worse than that of the Intel P4-M notebooks! In general it is safe to assume that Apple would use the lowest power version available at a given clock speed (for mobile applications), but as you said, if they screwed up when designing their mobos and didn't anticipate lower voltage chips being available, then you never know.

But the 1.2 Ghz part doesn't end up using that much more power than the G4.

Yes, I agree (though it still is more). Again, the 1.2 Ghz part is a lower voltage part than the 1.8 Ghz part. Apple could certainly use it in the PB, and the battery life would probably be somewhat worse than today's G4 but still better than a P4-M.

The thing is that the comparison is more likely to be between 7457 and 970, and the 7457 clearly dissipates significantly less heat than the 970. Nevertheless, the 970 is much more powerful, so there is a classic speed vs. power consumption tradeoff. The best world would be if Apple released a large 970 notebook for those who want speed and a smaller 7457 notebook for those who want a light, long battery life notebook (that is still reasonably fast). But it is unclear whether they can afford to support two totally different motherboards at the same time for the PowerBook lineup (plus the iBook). And if they went all-970 for the PBs, then they would have no answer for the low power consumption of Centrino, which, if Intel has its way, will soon become the primary mobile competition. So that is a very dangerous move strategically.

Again, the best of all worlds would be to offer both 7457 and 970, but I don't know if it is feasible from a cost perspective. And maybe the average consumer would find the speed gain of the 970 to be worth the tradeoff of higher power consumption (vis a vis the 7457, and even vis a vis the latest 7455). But the point is that THERE IS A TRADEOFF. So people really need to let go of this ridiculous Panglossian idea that somehow the 970 will swoop in with the performance of the fastest P4-M AND the power consumption of the most efficient PPC 750FX. That is NOT the way the world works. (I think you know that, but many other people on this site seem to be blissfully unaware of it)

 

[mathiasr]

Originally posted by Rincewind42
I agreed with most of your post, but here you aren't entirely correct. You are correct in that OS X is 64-bit clean, and would probably only require a recompile (for the most part) to run cleanly on a 64-bit processor. However, this is not the only concern. You will want the OS to take advantage of 64-bits, but unless they have been really good, just recompiling will only get you as far as running, not necessarily bug free. Additionally, that still only gets the OS on a 64-bit processor. You still have to build APIs for developers to use to develop 64-bit apps and tools to create 64-bit apps. These concerns may have already been handled, but without actual 64-bit hardware to run on these capabilities are only a little better than being only on paper.

Don't forget that the kernel has to handle the supervisor part of the CPU, 64 bits supervisor-level registers are quite different compared to the 32 bits one. Look at virtual memory implementation: segment registers are used in the 32 bits PowerPCs whereas the address space register is used in 64 bits ones.

On the other hand Apple could have made a 64 bits version of Mac OS X and tested it against IBM POWER3 or POWER4 RS/6000 hardware for years, a 64 bits PowerPC CPU is nothing new*. Linux PowerPC64 runs on these boxes : http://linuxppc64.org/history.shtml

*Bringing it to the mainstream market and adding an AltiVec unit, definitely is.

 

[pgwalsh]

Originally posted by mathiasr
Real developers would look at what Cocoa has to offer !

One of them is a very young and talented CGI developer. He loves OS X too.

 

[ffakr]

Re: Intel 4-5 ghz chips

Originally posted by Macpoops
Last time i checked the P4 was still at 3.06 with hyperthreading. Also their are rumors that they are stuck at that clockspeed or very close to it until they shrink their Fab to .09 um.

Intel could probably release faster P4s right now but they don't need to because they are beating everyone else.

The processors have an overall yield.. so many good, so many crap... and of the good ones, certain processors will run reliably at different speeds.

Intel can make faster processors right now.. they might be able to release the P4 up to 3.6 GHz or so.. buy why should they?
lets say market forces allow them to fit a processor into these price points {$100, $125, $180, $300, $550} .. note I put a premium on the high end...
Now, assume that they are getting a sort like this.....
{2.5GHz, 2.8GHz, 3GHz, 3.2GHz, 3.4GHz, 3.6GHz}... but the 3.6GHz and 3.4GHz parts are constrained (not too many of them).

Intel can give you the latest greatest.. and only have a limited supply of the better processors... (remember they are pretty much locked into max price points by market forces)
Or...
they can release them all at lower speeds... the 2.5GHz can be sold as anything at 2.5 or less... everything shifts down... and all the processors that currently sort at 3-3.6 GHz can be sold for the same premium price ... but the supply is MUCH higher.
It makes good business sense for the leader in the chip race to only sell processors that are slightly faster than the competetors.

Now if the 970 debuts... hopefully Intel won't consider it a real competitor to the Pentium4.. and they won't immediatly raise the clock speeds. You probably won't see significantly faster P4s ship until AMD begins to threaten the P4 3.06 with HT.

 

[Frobozz]

Originally posted by macrumors12345
Did you even LOOK at the IBM document?!?

Yes. And it's information has been superceeded by recent developments. That same doc also mentions that they only go to 1.8 Ghz, which has since been trumped by a 2.5 Ghz chip. Simple enough.

Also, the current G4 in the Powerbooks DOES NOT dissipate 30 watts.

Actually, it does:

http://arstechnica.com/cpu/02q2/ppc970/ppc970-1.html

 

[BenRoethig]

Originally posted by macrumors12345
Yes, I agree, we don't know for certain which version of the G4 is using. However, we can be sure that it is not dissipating 30 watts, because if it were the battery life of the PB G4 would be even worse than that of the Intel P4-M notebooks! In general it is safe to assume that Apple would use the lowest power version available at a given clock speed (for mobile applications), but as you said, if they screwed up when designing their mobos and didn't anticipate lower voltage chips being available, then you never know.



Yes, I agree (though it still is more). Again, the 1.2 Ghz part is a lower voltage part than the 1.8 Ghz part. Apple could certainly use it in the PB, and the battery life would probably be somewhat worse than today's G4 but still better than a P4-M.

The thing is that the comparison is more likely to be between 7457 and 970, and the 7457 clearly dissipates significantly less heat than the 970. Nevertheless, the 970 is much more powerful, so there is a classic speed vs. power consumption tradeoff. The best world would be if Apple released a large 970 notebook for those who want speed and a smaller 7457 notebook for those who want a light, long battery life notebook (that is still reasonably fast). But it is unclear whether they can afford to support two totally different motherboards at the same time for the PowerBook lineup (plus the iBook). And if they went all-970 for the PBs, then they would have no answer for the low power consumption of Centrino, which, if Intel has its way, will soon become the primary mobile competition. So that is a very dangerous move strategically.

Again, the best of all worlds would be to offer both 7457 and 970, but I don't know if it is feasible from a cost perspective. And maybe the average consumer would find the speed gain of the 970 to be worth the tradeoff of higher power consumption (vis a vis the 7457, and even vis a vis the latest 7455). But the point is that THERE IS A TRADEOFF. So people really need to let go of this ridiculous Panglossian idea that somehow the 970 will swoop in with the performance of the fastest P4-M AND the power consumption of the most efficient PPC 750FX. That is NOT the way the world works. (I think you know that, but many other people on this site seem to be blissfully unaware of it)

Of course this is all assuming that Motorola is developing a desktop version of the 7457. With the 970 on the horizion and Motorola offically out of the desktop processor after the G4, the PPC7455A could have been their last desktop processor.

 

[type_r503]

Reading all these posts I keep thinking, "haven't I read this before?" Every time there is a rumor the same arguments come up. All I am waiting for is the "Why do you people think 64 bits will make it faster?" or the always popular "I just hope apple uses the IBM chip instead of the Motorola G5."

Hurry up apple!

 

[Mr. MacPhisto]

This goes with what I've been told by IBMers who are friends. They insisted Apple could have machines shipping by June. I haven't talked about the 970 with them for a while, so I'll have to ask if they know about the current production schedule.

 

[rjstanford]

Originally posted by macrumors12345
At any rate, it EXPLICITLY states that the 970 at 1.2 Ghz dissipates 19 watts. There is nothing to debate here: I have no idea where your made-up numbers are coming from (but they're certainly not from IBM), and they are just plain wrong.

Which gives us a couple of interesting possibilities, considering the performance difference (current estimates give the 970 about a 2X performance benefit per clock cycle over the G4).

Lets think about that for a second. We have the G4 (1ghz) at a conservative 14-15 watts, and the 1.2ghz 970 at 19 watts (1.3x the power but with approximately 2.5x the speed).

( can you see where I'm heading with this? )

Now, assume that the 970 power requirements decrease as its clock speed does, just like the current processors. Optimize for battery life and all that. How much faster do you think it would be at the G4's 15 watt level? Especially if it could still burst to full speed (maybe more conservatively) when needed? How much power would it take to run at the G4's normal "reduced" levels for battery longevity?

Hmm...

 

[rjstanford]

Interesting tie-in

I was just browsing around and came across this little item:

Reports I have seen (just the usual "I have a friend at IBM" kind of things) would indicate that this thing taped out about year ago, and sampled in March. So, no way would Apple have had hardware a year ago, but they may have had software emulation and samples the following quarter. So that would put the 1 year anniversary around 1Q '03.

Why interesting? I dug it off one of the ArsTechnica discussion boards ( http://arstechnica.infopop.net/OpenTopic/page?q=Y&a=tpc&s=50009562&f=8300945231&m=8560987925&p=8 ) dated 9/23/02. Seems pretty timely...

 

[macrumors12345]

Originally posted by Frobozz
Yes. And it's information has been superceeded by recent developments. That same doc also mentions that they only go to 1.8 Ghz, which has since been trumped by a 2.5 Ghz chip. Simple enough.

And, in fact, we have no reason to believe at this point that the initial clock speeds will be above 1.8 Ghz. The latest word from IBM in March at CeBit reiterated that they would be released initially at speeds up to 1.8 Ghz. See the earlier MacRumors story about this.

Furthermore, IBM has NEVER claimed that the PPC 970 would consume less power than their documents currently say. I have no idea where you got that idea. I would LOVE to see a credible source saying that the chip will only consume 10 watts. And no, some random guy posting on the MacRumors forum that his friend's cousin said it would be a low power chip does NOT count as confirmation of this rumor.

Actually, it does:

http://arstechnica.com/cpu/02q2/ppc970/ppc970-1.html

Siracusa does some nice writing, but quite frankly he is simply wrong on this point. Do you honestly think that he has better information about the "G4e" (which isn't really a chip) than Motorola, the company that produces it? The article is a bit dated (it's from last fall), but even the earlier version of the 7455 didn't dissipate 30 watts at 1 Ghz. As you can see from the following Motorola document, the "G4e" that he was referring to, i.e. the 7455 with a core voltage of 1.6v, dissipates 21.3 watts at 1 Ghz. Maybe he was referring to maximum power dissipation for the G4 rather than typical power dissipation, but that would be an invalid comparison since the 42 watt figure quoted for the 1.8 Ghz 970 is typical, not maximum.

And even if Siracusa's facts were right, the processor he is referring to would not be "the current one that is shipping in the Powerbook G4", which is explicitly what we were talking about earlier.

However, I will congratulate you for at least having SOME quasi-credible source for this claim (even if it was wrong), which is more than I can say for many of the rumors about the 970 that are being thrown around so casually.

Motorola specifications document:
http://e-www.motorola.com/brdata/PDFDB/docs/MPC7455FACT.pdf

 

[macrumors12345]

Originally posted by BenRoethig
Of course this is all assuming that Motorola is developing a desktop version of the 7457. With the 970 on the horizion and Motorola offically out of the desktop processor after the G4, the PPC7455A could have been their last desktop processor.

Nobody is suggesting that they use the 7457 in the desktop. Except for with the iMac. I think a 1.5+ Ghz 7457 iMac with a 167-200 mhz bus (and some L3 cache for the higher models) would be great! It would certainly give the Celerons a run for their money. But I won't hold my breath for it appearing *too* soon...

No, the main discussion about the 7457 is wrt mobile processors (i.e. will Apple use the 7457 or the 970 in the next generation of Powerbooks?).

 

[barkmonster]

Originally posted by Rincewind42
For the G4 we have confirmed 15 watts at 1.067 Ghz (from Motorola's PDF) and a 'street' quote of 30 watts at 1Ghz.
------------------------------------------------------------------------

Yes, I agree, we don't know for certain which version of the G4 is using. However, we can be sure that it is not dissipating 30 watts, because if it were the battery life of the PB G4 would be even worse than that of the Intel P4-M notebooks! In general it is safe to assume that Apple would use the lowest power version available at a given clock speed (for mobile applications), but as you said, if they screwed up when designing their mobos and didn't anticipate lower voltage chips being available, then you never know.

This review of the powerbook 17" mentions a realworld battery life of 2 hours, 38 minutes (In their stress test)

In my usual harsh battery test, where I turn off all power-saving features and play a continuous loop of music, the 17-inch PowerBook lasted two hours, 38 minutes. That suggests that with power-saving on, and a more normal usage pattern, you'd get between 3.5 and four hours, which is pretty good, though shy of the "up to 4.5 hours" that Apple claims.

However, there's a new crop of Windows laptops using the new power-saving Intel Pentium-M processor that typically do a lot better. In my tests of four Pentium-M laptops a few weeks back, three of the four showed significantly longer battery life than the new Apple.

I think we can safely say anything that improves on the battery life of a powerbook whether it's lower powered G4s from motorola or PPC970 chips from IBM are needed soon. The Centrino means PC laptops can have better battery life than the Pentium-M or G4, thin, lightweight design like the G4 and wireless networking that the mac's had for 4 years (they should have advertised that as the main feature of instead of neon toilet seats!!)

 

[law guy]

Re: Intel 4-5 ghz chips

Originally posted by Macpoops
Didn't IBM say the Blade servers would ship Q3 this year? If the chips are going into production that would seem to be right.

Well, the article says that IBM will deliver the chips for IBM Blades and Apple computers at the end of June. With the motherboard design complete - as rumored - and mass shipments of chips in July, it seems like Apple could start getting these machines out soon. Heck, Apple could even announce in June given their propensity to announce before ready for shipping.

 

[tychay]

Originally posted by j33pd0g
I agree. And I am really excited too! Plus I got burned with rushing out to get the first g4s when they came out (I got the "350 Yikes") and then they discontinued that like 3 months later and started with the AGP motherboards. If these rumors are true I think I'll wait to buy a 970 mac until they have been well tested by the mac community. For now I'll settle for a dual 1.25 and I will like it.

The 350 Yikes lasted from mid October to the beginning of December of 1999, so I believe it is more accurate to say you've been using it for 3.5 years instead of four. While the "Yikes" PCI motherboard lasted about 3 months (end of August to beginning of December), the 350 Yikes was half that life and the computer was simply introduced because Motorola was having yield problems with the G4/500--imagine waiting half a year for a computer you preordered!

I'm not too certain you should say you got "burned". At the time, the 400 Yikes had already shipped and Apple was raising prices and lowering clockspeeds contrary to whatever has ever happenned in the industry. People knew that it was inevitable that they would at least lower them back to announced price and configuration (which they did). Also, Sawtooth (AGP) boards were announced the same time the Yikes was. Apple made it clear, with the name based on "Yosemite" that these PCI configuration G4's were EOLed when they launched. This is similar to what happenned when LCD panel prices went up and iMac G4s jumped $100 in price.

Small historical note: PowerMac G4s seem to be refreshed twice a year. While the Yikes was very stunted (at around 3 months), the Sawtooth announced at the same time lasted almost a full year.

As one of the lucky enough to get a Yikes at 400Mhz, I can agree that they still run like a champ.

It looks like Apple is trying to keep that sort of chip supply fiasco from happening with the 970 with a rumored release date well after IBM is rumored to have production ready.

Take care,

terry

 

[macrumors12345]

Originally posted by rjstanford


Lets think about that for a second. We have the G4 (1ghz) at a conservative 14-15 watts, and the 1.2ghz 970 at 19 watts (1.3x the power but with approximately 2.5x the speed).

( can you see where I'm heading with this? )

Yes, but I don't think it's going to work.

You have to realize that the reason the 1.2 Ghz 970 dissipates less than have the power of the 1.8 Ghz 970 is that its core voltage is 1.1v instead of 1.3v. A lot of the decrease in power consumption comes from dropping the core voltage. Now, as a result of dropping the voltage, the maximum clock speed also has to drop. So the drop in clock speed is correlated with the drop in power consumption, but it's not necessarily causal.

The point is that if you were to use a 800 mhz 970 with a core voltage of 1.1v, it would probably consume less power than the 1.2 Ghz, but not that much less. For example, maybe 13 or 14 watts (just a guess), but definitely not under 10 watts. Now, if you dropped the core voltage again, from 1.1v to 1.0v or 0.9v, then you might get the power consumption under 10 watts at low clock speeds. But I have no idea whether or not that's feasible (I'm not an electrical engineer, nor do I work for IBM!).

At any rate, even if you could get the power consumption down to, say, 10 watts by lowering the core voltage and throttling the clock down to 750 mhz, would you really want to buy a 750 mhz PB? I mean, I'm not saying I wouldn't (buy it), but would the general public? It would be an interesting psychology experiment. Because really, that just sounds pathetic, even if the processing power is comparable to a 1.5 Ghz G4 or 1.5 Ghz Centrino. I suspect Apple would have trouble selling 750 mhz 970 based Powerbooks if they were also selling 1.2 Ghz G3 based iBooks! I mean, far be it from me to defend the Megahertz Myth, but still, it would have to hurt sales....

I still like offering a 1.3+ Ghz 7457 Powerbook (say, 12"/15") and a 1.2 Ghz 970 Powerbook (15"/17"). That would be really nice. But I think it might be too expensive...unclear...

 

[tazznb]

The G3-4 chips still have some life left, but...

Originally posted by minux
Do not be surprised to see the G5 go only into a high level PM in its initial run. Additionally, do not be surprised to see the low end PM sporting a G4 for some time to come. The IBM's production can not match the demand for the G5, nor can the price tag of the G5 find its way into many consumers pockets.


Mike

If Apple only puts the 970 in its highest end PM, no one will buy the other PMs without it, and at a realistic price point.

"The IBM's production can not match the demand for the G5".

This is IBM taking up the chip process, now. Not Moto; If Moto were doing this I can agree the chips not being readily available to use for all users for another 2-5yrs

 

[macrumors12345]

Originally posted by barkmonster
This review of the powerbook 17" mentions a realworld battery life of 2 hours, 38 minutes (In their stress test)

I believe it. But it would not even last that long with a 30 watt G4 under "stress testing". The other components (e.g. display, hard drive, etc.) consume at least 10-15 watts themselves. Assume for the moment that it was only 10. Then total power consumption would be 40 watts. So to last 2.6 hours, they PB would need an battery with a capacity of over 100 watt-hours. But the 17" PB has a 60 watt-hour battery. So we can reject the hypothesis that the G4 is dissipating 30 watts.

I think we can safely say anything that improves on the battery life of a powerbook whether it's lower powered G4s from motorola or PPC970 chips from IBM are needed soon

Certainly (I'm just saying the 130 nm 970 won't be that chip - see the 7457).

. The Centrino means PC laptops can have better battery life than the Pentium-M or G4,

Ummm...you realize that the Centrino IS the Pentium-M, right? (well, technically to use the Centrino brandname you have to use a Pentium-M *and* buy Intel's own chipset and wireless kit and so forth)

 

[Rincewind42]

Originally posted by macrumors12345
So to last 2.6 hours, they PB would need an battery with a capacity of over 100 watt-hours. But the 17" PB has a 60 watt-hour battery. So we can reject the hypothesis that the G4 is dissipating 30 watts.

Actually the 17" PowerBook uses a 55 watt hour battery. The 15" PowerBook uses a 61 watt hour battery. That leads me to believe that as part of the mother board redesign, the 17" PowerBook was made to take advantage of a 1.1v CPU, while the 15" might use a 1.3v chip, which would be estimated to disappate 20 watts.

 

[macrumors12345]

Originally posted by Rincewind42
Actually the 17" PowerBook uses a 55 watt hour battery. The 15" PowerBook uses a 61 watt hour battery. That leads me to believe that as part of the mother board redesign, the 17" PowerBook was made to take advantage of a 1.1v CPU, while the 15" might use a 1.3v chip, which would be estimated to disappate 20 watts.

Thanks for the exact numbers. It's quite possible that your conjecture is true. Although the PB G4 17" also has a proportionally lower quoted runtime ("up to 4.5 hours" for 17" vs. "up to 5 hours" for 15"). Maybe the larger display uses more power, and that is why the 17" doesn't have comparable battery life to the 15" if it is using a more efficient processor? Unclear... (the other hypothesis of course being that they're both using the same processor...quite possibly the low voltage version, but who knows)

 

[Rocketman]

Re: Re: Heaven in September?

Originally posted by mcs37
I can't wait! I am still hoping for a dual 970 17" PowerBook. I refuse to believe Apple will not release a dual 970 17" PB this January. They have plenty of area to spare inside that massive laptop. Just imagine the insane power of a 17" PB, dual 970 @ 1.5 GHz, 1 GB RAM, USB 2.0, SuperDrive, Airport Extreme.... mmmmm I'm drooling already.

I still hope for a built-in HDTV tuner. But I know that's a long shot, as is the booze-powered fuel cell to power it for 10+ hours.

Our copmpany bought 2 AL 17's on speculation. We will probably buy more.

If they do release a dualie 970 Powerbook (and steviepoo DID say this is the year of the laptop), then I suspect we will have to redirect resources to make room for them. Can you say portable HD editing station?

For those that note they willl wait for a year or so for desktop issues to settle on a new 970, I agree. For those who note that their freinds have has three peecees during the timeframe of a single older G4 box, of course I agree. For those who note that Apple redated WWDC because of likely 970 news, I hope you are right because the news would be a predating event for a virtually guaranteed CONSUMER August release of new 970 hardware.

I still claim it willl be x-serve first and HOPE it will be PowerPowerBook first.

Rocketman

 

[MacQuest]

Originally posted by springscansing
Holy crap, 2mhz! That's twice my IIe man! Sweet!

ROTFL!! ROTFL!! ROTFL!!

 

[MacQuest]

Originally posted by GPTurismo
Time to start saving my pennies!

BTW Intel = LOL

Yeah, I'm gonna start saving my *****' too!!

 

[woodsey]

Apple Sponsort MacBidouille!

Anyone else think it is intersting that there is an official apple advertisement, linked to the french apple store, at the top of the macbidouille site?

I didn't know Apple supported rumor sites!!!

Whats up here???