Regarding the veracity of the IBM numbers: those numbers are from 9 months ago and not on shipping product. They most likely are based on estimates from simulation.
Would it make sense, though, that IBM, of all organizations, would be off by 50% or more in their estimates? IBM is typically very conservative in their estimates, and it stands to reason that the power consumption would be less, not more than the estimate.
As I mentioned in the iMac thread, the POWER 4 consumes 100 watts. Would it make sense that the 970, a chip which is based on the POWER 4's design but much simpler (one core instead of two, no 32MB on chip L3 cache, no larger-than-normal elements for increased reliability), would consume just three watts shy of the POWER 4?
In fact the EETimes numbers are absolutely made up... they didn't want anyone to know... so they came up with a odd number like 97 instead of 50 or 100. One guy guessed a number between 1 and 10 and then another guy did the same and they combined these numbers to make the number that they printed in the article. They didn't talk to IBM at all to get those numbers. Amazingly, no matter how wild those numbers are... no retraction was issued... guess no one at IBM has seen those numbers...
Also, I found out that the pre-production from simulation numbers that IBM released a nine months ago are, get this, exactly correct. It seems like the Cadence design tool that IBM used for the 970 is frighteningly dead on when it comes to power consumption numbers!!!
Actually, the Cadence SW was wrong. IBm was conservative and the numbers for the parts are less than what was in the pre-production document.
Also, I found out that Steve Jobs has had a change of heart. The 970 is perfect for running in the new 15 inch PB and there is currently production running, but since he's been told that he's got a big ego and has to release EVERY neat new toy, he's going to save the fastest laptop on the planet release for Greg.
Wait... he wants to save Greg's ego too... so they're just going to quietly release it one Tuesday next month...
And, oh yea!, the nine fans are needed in the PowerMac because nine fans are QUIETER than one. Even though they had fanless designs in the Cube and a VERY quiet fan in a iMac, they decided to use 9 fans because every computer buyer things fan tech is SOOO cool. And they used the largest heat sinks I have ever seen and the most free airflowing case I have ever seen just by chance!
( Note: I too seriously believe the nine fan design would be quieter than one fan; however, that does not negate the fact that the parts are hot and Apple had to design a special cooling system for the devices. )
But believe whatever you want... it doesn't matter to me!
Oh, boy. Apparently, it does matter to you, otherwise you wouldn't have gone off on that rant...
Do you have a cogent response, preferably something that doesn't involve childish ridicule?
48.5 watts seems a hell of a lot more reasonable to me than 97 does, given the original estimate of 42 watts, albeit at a slower speed. If the 97 watt figure is true, then IBM's got some SERIOUS problems with the 970 design and their design process, and there would be a boatload of IBM engineers looking for new jobs...
Think about it:
POWER 4: 2 cores, 32MB L3 cache, larger elements = 100W.
PPC 970: 1 core, no L3 cache, normal elements = 97W.
Does that really make sense to you?
By the way-- the nine fan thing is a red herring. Get over it. Only four are used to cool the processors, and as you admitted yourself, the thing is designed to be a quiet running machine, so larger heat sinks & more fans aren't really surprising, are they?
Actually, it really doesn't matter to me... I just think it's funny that people basically try to convince other people that their reading of the tea leaves is correct...
I too am surprised by the 97 watt numbers. However, those numbers had to come from somewhere and most likely they came from an IBM engineer/marketing rep at the WWDC. If they had been misquoted or misstated to such a degree, well, I think that IBM would have asked for a retraction by now.
Of course they could be wrong - but speculating on that is idle speculation since neither of us can offer up proof to the contrary. I'm just taking the numbers at face value.
You choose to look at the published documents by IBM. I think those documents are very informative; however, I do give the chance that they are incorrect on power consumption. I'm taking the latest numbers - you're taking the older numbers - it's just what we choose to believe.
Your comparison on the 970 and the Power4 numbers make sense - I'd also love to see another test. But again I just choose to believe what the EETimes is saying until I hear otherwise. Again believe what you will...
I also do not think that the fan design is a "red herring." I've been an engineer and a program manager for 17 years. In the G5 there has been alot of design work, manpower, time, etc. in designing the unit to keep cool. The chassis is absolutely optimized to maximize airflow. The heat sinks are over-sized. The fan system is an extra system that needs to be designed and controlled. All of this had to be designed and was done not arbitrarily, but for a reason. If quietness was the SOLE reason, then there are other solution to that - the Cube was quiet and so is the iMac; however, I say that one of the main solutions was heat. In any project there are limited resources and cost constraints, therefore money, time, and manpower spent on this solution was money, time, and manpower taken from elsewhere. Even if the engineers had nothing else to do, the cost incurred ( even from the materials - those heat sinks are more costly than smaller ones ) impacts the return on the project. There was a conscious decision to deal with a high source of heat in the chassis.
There's the cogent response... but the childish ridicule is so much more fun! I DO think there were cogent points in the childish ridicule...
The idea that IBM's estimate would be off by roughly a factor of two doesn't make sense.
The idea that the 970 would consume nearly as much power as the MUCH more complicated POWER 4 doesn't make sense.
Conclusion: the eeTimes writer mistakenly put down the power consumption for both processors as the power consumption of one processor.
I don't know how much more simply I can put it.
--
Regarding the fans: The Cube's cooling system is irrelevant. Just assume for a moment that the 97 watt figure for both processors is correct. The processor in the Cube uses quite a bit less power than that... I don't know the figures for certain, but I'd be surprised if they were more than about 18-20 watts, probably closer to 15. Either way, the G5's use five or six times as much power as the Cube. There's no way Apple would consider using the same cooling method in the G5 as they used in the Cube. Besides, I wouldn't hold my breath waiting for a machine that was cooled like the Cube. Frying the processor because you stacked a book on top of the machine doesn't really sound like good industrial design, to me.
As far as the iMac goes, why is it quieter than other computers? Perhaps because it uses a fan running at a lower speed? Sound familiar?
The G5's design addresses some of the shortcomings the El Cap case had. The first iteration of the MDD case, especially, had problems with PCI cards overheating, and the name "windtunnel" came from more than the fact that it has a huge fan, but because it is LOUD. It's rather obvious that the G5's design addresses these problems, and is not just to deal with a single source of heat within the chasis.
On topic: I'd imagine it's possible for Apple to shoe-horn a G5 into a Powerbook case, but I suspect they'll go with the 7457 instead. I'd imagine that the G5 along with a new system controller would draw enough power that Apple's claims of long battery life would go out the window (like from 5 hours down to 3 or 2 1/2, which means 1 1/2 to 2 1/2 realistically). I think Apple realizes that people value battery life above performance in a laptop, and will go with Moto for another iteration, at least.
I brought this up in the iMac thread, but this graph seems to be getting a lot of discussion here.
Can someone explain this graph to me so that it makes sense, or post a link to the original source document?
Why does the caption reading "97 Watts: Source: IBM 970 doc" have two arrows pointing at blue dots on the "Predicted" graph that are at about 20 and 42 watts? If the caption's not pointing out where the graph hits 97, what is it trying to point out?
Why are there two red lines for "Actual"? eeTimes data vs. IBM data? If so, which one is which? How did anyone come up with a curve for the top red line from only one data point on one end?
Is there any way to know if these figures had anything to do with a dual processor system rather than a single processor?
rickag... what corrections would you have me make??? Considering everything we are talking about here is an assumption... BTW, what other website has used this chart?
Geez, I'm down to explainning sarcasm!
Gamblor, you assume that the EETimes article was wrong and that the IBM marketing docs were correct. I assume otherwise. THAT is as simple as it gets.
Your fan explanation does not convince me.
Shetline, the 97 watts is from the EETimes article. The 42 watt/ 19 watt data points are from the three IBM marketing documents mentioned in other threads. Those points are mislabled with the "97 watts." As I said in the original post the graph is an extrapolation. It is based on the current number ( 97 watts ), the numbers from the IBM marketing doc, and SOI power consumption curves. It is a swag but so far we only have three numbers... wish there was more.
rickag... what corrections would you have me make???
You might remake your graph based on this IBM document released on 7/2/03.
IBM 970 @ 1.8GHz uses 47 watts.
Just need to add that I don't doubt your knowledge, but I tend to believe documents from IBM rather than second hand sources such as eetimes(to easy for reporters to misunderstand information they gather).
Oh, and as to other sites, I can't recall, could be Thinksecret or Spymac or MacNN, but your graph has definitely been used elsewhere. It is a very nice graph and very well thought out.
Promise this is the last edit. I forgot to mention that I stole this link from a post by ADDRAGYN at Arstechnica in the thread,"G3 and G4 Processors and news" in Machintosian Achaia Forum.
Gamblor, you assume that the EETimes article was wrong and that the IBM marketing docs were correct. I assume otherwise. THAT is as simple as it gets.
Uh, no. I concluded that the eeTimes article was wrong, and provided the reasoning behind that conclusion. The fact that you rejected my reasoning does not make it an assumption.
Quote:
Your fan explanation does not convince me.
Ya know, Chazwazzer, I'm begining to suspect that you've been a manager for too long. You seem to rely on "proof via decree" quite a bit...
...None of which really matters, though, since rickag provided a link to current numbers from IBM. Thanks Rick!
I wonder how the eeTimes arrived at 97 watts? It's too high for a single processor, and I doubt the 2GHz 970 only consumes 1.5 watts more than the 1.8GHz...
PowerPage is on board with the world's most lukewarm rumor confirmation ever:
"A PowerPage source has indicated that the feature presentation by Greg 'Joz' Joswiak, Vice President of Hardware Product Marketing at Apple, may include a surprise announcement of the new Aluminum PowerBook 15-inch. It is unconfirmed and frankly, a long shot, but stranger things have happened. Joz is actively involved in PowerBooks and has given most of the quotes about the new 'Books to the media, so it seems plausible."
Rickag! Thanks for the new data point!!! Now that this is more recent, I believe this over the EE Times number.
I have a email in to the writer of the EE Times article on how he got the number.
I DO NOT see a power consumption rise of 50 watts from 1.8 GHz to 2 GHz. Could at most see a 10 watt rise and, yes, 1.5 watts seems a little low, but it is more in the ballpark than 50.
Gamblor, it's not that I believed the EETimes number; however, since it was the only number around it was the only data point I had. I've seen too many times where pre-production numbers turn out wrong. With the Power 4 numbers where they are it would seem unlikely and I agree that the reporter could have gotten it wrong; however, I don't KNOW that... so with only one post-production data point to go with...
PowerPage is on board with the world's most lukewarm rumor confirmation ever:
"A PowerPage source has indicated that the feature presentation by Greg 'Joz' Joswiak, Vice President of Hardware Product Marketing at Apple, may include a surprise announcement of the new Aluminum PowerBook 15-inch. It is unconfirmed and frankly, a long shot, but stranger things have happened. Joz is actively involved in PowerBooks and has given most of the quotes about the new 'Books to the media, so it seems plausible."
That's about as much of a "kinda, maybe, sorta, not sure, could be, hmmm..." "rumor" as I've ever heard. They're trying to cover their backside, just in case it doesn't come to be.
They are, apparently, trying to be the salt to Macwhisper.com's pepper. The peanut butter to their jelly. The lemon to their lime, as it were.
Where macwhispers.com pretty much goes "yeah, 970 PowerBooks are DEFINITE at WWDC...we've seen the production numbers!!!" (and then gets all offended and pouty when people called him on his crap), powerpage.org seems to be going out of their way to make sure that, come next week if things don't pan out, they can always say "well, we SAID it was a longshot...".
Exactly, and Jack Campbell wonders why so many people get so pissed off! I mean he really says things like "two shifts have been working on the new 15 PB and Apple is stocking them in their warehouse." And then it doesn't happen and he's like "uhhh, I wasn't lying... they'll be out next week.."
Sorry to basically repeat your post, but the guy just drives me nuts!
It's like the Liar guy and Emily Lattela from SNL had a kid...
PowerPage is on board with the world's most lukewarm rumor confirmation ever
It sounds like they are trying to have their cake and eat it too; if it's right they take credit, if it's wrong, "it's not like we didn't warn you"-type of thing. Lame ass.
Comments
anyone have any idea FOR REAL when the 15 incher is coming out?
DIE! DIE! DIE!
Regarding the veracity of the IBM numbers: those numbers are from 9 months ago and not on shipping product. They most likely are based on estimates from simulation.
Would it make sense, though, that IBM, of all organizations, would be off by 50% or more in their estimates? IBM is typically very conservative in their estimates, and it stands to reason that the power consumption would be less, not more than the estimate.
As I mentioned in the iMac thread, the POWER 4 consumes 100 watts. Would it make sense that the 970, a chip which is based on the POWER 4's design but much simpler (one core instead of two, no 32MB on chip L3 cache, no larger-than-normal elements for increased reliability), would consume just three watts shy of the POWER 4?
In fact the EETimes numbers are absolutely made up... they didn't want anyone to know... so they came up with a odd number like 97 instead of 50 or 100. One guy guessed a number between 1 and 10 and then another guy did the same and they combined these numbers to make the number that they printed in the article. They didn't talk to IBM at all to get those numbers. Amazingly, no matter how wild those numbers are... no retraction was issued... guess no one at IBM has seen those numbers...
Also, I found out that the pre-production from simulation numbers that IBM released a nine months ago are, get this, exactly correct. It seems like the Cadence design tool that IBM used for the 970 is frighteningly dead on when it comes to power consumption numbers!!!
Actually, the Cadence SW was wrong. IBm was conservative and the numbers for the parts are less than what was in the pre-production document.
Also, I found out that Steve Jobs has had a change of heart. The 970 is perfect for running in the new 15 inch PB and there is currently production running, but since he's been told that he's got a big ego and has to release EVERY neat new toy, he's going to save the fastest laptop on the planet release for Greg.
Wait... he wants to save Greg's ego too... so they're just going to quietly release it one Tuesday next month...
And, oh yea!, the nine fans are needed in the PowerMac because nine fans are QUIETER than one. Even though they had fanless designs in the Cube and a VERY quiet fan in a iMac, they decided to use 9 fans because every computer buyer things fan tech is SOOO cool. And they used the largest heat sinks I have ever seen and the most free airflowing case I have ever seen just by chance!
( Note: I too seriously believe the nine fan design would be quieter than one fan; however, that does not negate the fact that the parts are hot and Apple had to design a special cooling system for the devices. )
But believe whatever you want... it doesn't matter to me!
Do you have a cogent response, preferably something that doesn't involve childish ridicule?
48.5 watts seems a hell of a lot more reasonable to me than 97 does, given the original estimate of 42 watts, albeit at a slower speed. If the 97 watt figure is true, then IBM's got some SERIOUS problems with the 970 design and their design process, and there would be a boatload of IBM engineers looking for new jobs...
Think about it:
POWER 4: 2 cores, 32MB L3 cache, larger elements = 100W.
PPC 970: 1 core, no L3 cache, normal elements = 97W.
Does that really make sense to you?
By the way-- the nine fan thing is a red herring. Get over it. Only four are used to cool the processors, and as you admitted yourself, the thing is designed to be a quiet running machine, so larger heat sinks & more fans aren't really surprising, are they?
I too am surprised by the 97 watt numbers. However, those numbers had to come from somewhere and most likely they came from an IBM engineer/marketing rep at the WWDC. If they had been misquoted or misstated to such a degree, well, I think that IBM would have asked for a retraction by now.
Of course they could be wrong - but speculating on that is idle speculation since neither of us can offer up proof to the contrary. I'm just taking the numbers at face value.
You choose to look at the published documents by IBM. I think those documents are very informative; however, I do give the chance that they are incorrect on power consumption. I'm taking the latest numbers - you're taking the older numbers - it's just what we choose to believe.
Your comparison on the 970 and the Power4 numbers make sense - I'd also love to see another test. But again I just choose to believe what the EETimes is saying until I hear otherwise. Again believe what you will...
I also do not think that the fan design is a "red herring." I've been an engineer and a program manager for 17 years. In the G5 there has been alot of design work, manpower, time, etc. in designing the unit to keep cool. The chassis is absolutely optimized to maximize airflow. The heat sinks are over-sized. The fan system is an extra system that needs to be designed and controlled. All of this had to be designed and was done not arbitrarily, but for a reason. If quietness was the SOLE reason, then there are other solution to that - the Cube was quiet and so is the iMac; however, I say that one of the main solutions was heat. In any project there are limited resources and cost constraints, therefore money, time, and manpower spent on this solution was money, time, and manpower taken from elsewhere. Even if the engineers had nothing else to do, the cost incurred ( even from the materials - those heat sinks are more costly than smaller ones ) impacts the return on the project. There was a conscious decision to deal with a high source of heat in the chassis.
There's the cogent response... but the childish ridicule is so much more fun! I DO think there were cogent points in the childish ridicule...
Originally posted by chazmox
Ok! Ok! You got me! I'll admit I am wrong!
Thank you, now will you go to all the other websites that have used your graph and make all necessary corrections.
Originally posted by soulcrusher
I am almost certain.
what kind of deal is it?
The idea that the 970 would consume nearly as much power as the MUCH more complicated POWER 4 doesn't make sense.
Conclusion: the eeTimes writer mistakenly put down the power consumption for both processors as the power consumption of one processor.
I don't know how much more simply I can put it.
--
Regarding the fans: The Cube's cooling system is irrelevant. Just assume for a moment that the 97 watt figure for both processors is correct. The processor in the Cube uses quite a bit less power than that... I don't know the figures for certain, but I'd be surprised if they were more than about 18-20 watts, probably closer to 15. Either way, the G5's use five or six times as much power as the Cube. There's no way Apple would consider using the same cooling method in the G5 as they used in the Cube. Besides, I wouldn't hold my breath waiting for a machine that was cooled like the Cube. Frying the processor because you stacked a book on top of the machine doesn't really sound like good industrial design, to me.
As far as the iMac goes, why is it quieter than other computers? Perhaps because it uses a fan running at a lower speed? Sound familiar?
The G5's design addresses some of the shortcomings the El Cap case had. The first iteration of the MDD case, especially, had problems with PCI cards overheating, and the name "windtunnel" came from more than the fact that it has a huge fan, but because it is LOUD. It's rather obvious that the G5's design addresses these problems, and is not just to deal with a single source of heat within the chasis.
On topic: I'd imagine it's possible for Apple to shoe-horn a G5 into a Powerbook case, but I suspect they'll go with the 7457 instead. I'd imagine that the G5 along with a new system controller would draw enough power that Apple's claims of long battery life would go out the window (like from 5 hours down to 3 or 2 1/2, which means 1 1/2 to 2 1/2 realistically). I think Apple realizes that people value battery life above performance in a laptop, and will go with Moto for another iteration, at least.
Can someone explain this graph to me so that it makes sense, or post a link to the original source document?
Why does the caption reading "97 Watts: Source: IBM 970 doc" have two arrows pointing at blue dots on the "Predicted" graph that are at about 20 and 42 watts? If the caption's not pointing out where the graph hits 97, what is it trying to point out?
Why are there two red lines for "Actual"? eeTimes data vs. IBM data? If so, which one is which? How did anyone come up with a curve for the top red line from only one data point on one end?
Is there any way to know if these figures had anything to do with a dual processor system rather than a single processor?
Geez, I'm down to explainning sarcasm!
Gamblor, you assume that the EETimes article was wrong and that the IBM marketing docs were correct. I assume otherwise. THAT is as simple as it gets.
Your fan explanation does not convince me.
Shetline, the 97 watts is from the EETimes article. The 42 watt/ 19 watt data points are from the three IBM marketing documents mentioned in other threads. Those points are mislabled with the "97 watts." As I said in the original post the graph is an extrapolation. It is based on the current number ( 97 watts ), the numbers from the IBM marketing doc, and SOI power consumption curves. It is a swag but so far we only have three numbers... wish there was more.
Originally posted by chazmox
rickag... what corrections would you have me make???
You might remake your graph based on this IBM document released on 7/2/03.
IBM 970 @ 1.8GHz uses 47 watts.
Just need to add that I don't doubt your knowledge, but I tend to believe documents from IBM rather than second hand sources such as eetimes(to easy for reporters to misunderstand information they gather).
Oh, and as to other sites, I can't recall, could be Thinksecret or Spymac or MacNN, but your graph has definitely been used elsewhere. It is a very nice graph and very well thought out.
Promise this is the last edit. I forgot to mention that I stole this link from a post by ADDRAGYN at Arstechnica in the thread,"G3 and G4 Processors and news" in Machintosian Achaia Forum.
Gamblor, you assume that the EETimes article was wrong and that the IBM marketing docs were correct. I assume otherwise. THAT is as simple as it gets.
Uh, no. I concluded that the eeTimes article was wrong, and provided the reasoning behind that conclusion. The fact that you rejected my reasoning does not make it an assumption.
Your fan explanation does not convince me.
Ya know, Chazwazzer, I'm begining to suspect that you've been a manager for too long. You seem to rely on "proof via decree" quite a bit...
...None of which really matters, though, since rickag provided a link to current numbers from IBM. Thanks Rick!
I wonder how the eeTimes arrived at 97 watts? It's too high for a single processor, and I doubt the 2GHz 970 only consumes 1.5 watts more than the 1.8GHz...
"A PowerPage source has indicated that the feature presentation by Greg 'Joz' Joswiak, Vice President of Hardware Product Marketing at Apple, may include a surprise announcement of the new Aluminum PowerBook 15-inch. It is unconfirmed and frankly, a long shot, but stranger things have happened. Joz is actively involved in PowerBooks and has given most of the quotes about the new 'Books to the media, so it seems plausible."
I have a email in to the writer of the EE Times article on how he got the number.
I DO NOT see a power consumption rise of 50 watts from 1.8 GHz to 2 GHz. Could at most see a 10 watt rise and, yes, 1.5 watts seems a little low, but it is more in the ballpark than 50.
Gamblor, it's not that I believed the EETimes number; however, since it was the only number around it was the only data point I had. I've seen too many times where pre-production numbers turn out wrong. With the Power 4 numbers where they are it would seem unlikely and I agree that the reporter could have gotten it wrong; however, I don't KNOW that... so with only one post-production data point to go with...
Originally posted by jante99
Here is what I say about this thread:
DIE! DIE! DIE!
I think this thread has been neutered, so the proper German definite article would be:
Das! Das! Das!
Originally posted by Ensign Pulver
PowerPage is on board with the world's most lukewarm rumor confirmation ever:
"A PowerPage source has indicated that the feature presentation by Greg 'Joz' Joswiak, Vice President of Hardware Product Marketing at Apple, may include a surprise announcement of the new Aluminum PowerBook 15-inch. It is unconfirmed and frankly, a long shot, but stranger things have happened. Joz is actively involved in PowerBooks and has given most of the quotes about the new 'Books to the media, so it seems plausible."
That's about as much of a "kinda, maybe, sorta, not sure, could be, hmmm..." "rumor" as I've ever heard. They're trying to cover their backside, just in case it doesn't come to be.
They are, apparently, trying to be the salt to Macwhisper.com's pepper. The peanut butter to their jelly. The lemon to their lime, as it were.
Where macwhispers.com pretty much goes "yeah, 970 PowerBooks are DEFINITE at WWDC...we've seen the production numbers!!!" (and then gets all offended and pouty when people called him on his crap), powerpage.org seems to be going out of their way to make sure that, come next week if things don't pan out, they can always say "well, we SAID it was a longshot...".
Sorry to basically repeat your post, but the guy just drives me nuts!
It's like the Liar guy and Emily Lattela from SNL had a kid...
"Oh.... never mind.... that's the ticket!"
Originally posted by Ensign Pulver
PowerPage is on board with the world's most lukewarm rumor confirmation ever
It sounds like they are trying to have their cake and eat it too; if it's right they take credit, if it's wrong, "it's not like we didn't warn you"-type of thing. Lame ass.