errr high definition wide screen is probably the best thing on the list. ddr2 ram isnt really that great because its latency is way higher than ddr and the bus speed jumps are good but its not like theyll help that much. i think it is a much needed update for the powerbook line though. there is basically no real reason to spend 500 more on a powerbook instead of an iBook.
It's not that simple. The performance depends on the difference between what the current memory is and what they will use. Dual channel adds between 5-10% to the speed of the memory
There is a BIG difference between a PowerBook and an iBook. Have you ever gone to their site and looked at the difference?
Better GPU, better HD, better I/O options, higher speed, better quality screens with much higher rez,better memory options, etc.
This is like the argument about the Bug vs the Mercedes when I was in college in the '70's. They both got you there but the Bug gave a much bumpier ride.
The costs of ramping up a new product for one revision would be a lot higher than that spread out over 2-3+ years that a normal product exists. With the G5 they would have to contract out the tooling of a new MotherBoard, internals and case and recoup that cost in the (likely) 12 months that the product would be offered. Then there would be the cost of supporting that product with repair parts, documentation and training for repair techs, for the a few years as well. Even if all the R&D was done, and the chips ready to be delivered a wise person would scrap the release before all the other costs associated with the release were incured and go without an upgrade, opting instead to push foreword the timetable for the Intel switch. G5 isn't going to happen.
As for the G4, well PowerLogix has released a 2 Ghz 7447 upgrade for PM's, and with the smaller process and other improvements that the 7448 has one would logically assume that even if it is not anounced there is a faster chip available.
I seam to remember that there was a scandal with one of the upgrade manufacturers a number of years ago about processor speeds, and the company pointed out that the chips are are tested and certified to meet the requirements of the contract. This particular case was an issue of the upgrade company buying a lot that originally ordered by another customer and the chips were certified at one speed, but in and environment with a higher operating temperature than the chips would be used in. Given the lower temperature of the Mac case they meet the speed requirements that they were sold as and were not clocked up, however IBM (i think) had printed the clock speed from the original contract on them so there were complaints from customers about thier purchase.
Sure, but that doesn't mean that it wouldn't be worth it. Apple might wery well find its Powerbook sales lagging too much here before the switch. Look at the numbers. While iMacG5's sales were up 48%, the sales of the laptop line was only up 16% with most of that being iBooks because of the school contracts. So the PB's are seriously lagging even now. With all the waiting for the G5, it might pay if what I said could be done. If Apple could up the sales by a third by doing this it would be worth while. I know what it costs to start up a production line. I'v gone through it numerous times with my own company.
There is also the political victory. If they can get it out, and its performance is fairly good it would be a plus while it lasted. When a new x86 machine comes out with a new chip line and is even better, ther will be applause, and a rush to buy.
It's all marketing. Companies look at costs in more ways than one. some of those costs would be absorbed as publicity. The startup costs are high, but not as high as you might think. If no new case is required, it's not very high at all. A new mobo design is cheap. Third party PC board makers come out with new ones every three months, and they don't all sell as well as you might think.
The startup costs could be a few 10's of millions at most. If Apple could sell 100 thousand of them in the next 7 months, not really that many after all, at a price of $2,000 to Apple, that's $200 million. Not too shabby.
Freescale, nee Motorola, has had a reputation for dirty fabs for years on end now, all the way back to 220 nm and the 7400 "500 MHz" debacle, and have gradually fallen further behind every year after that. Schedules and roadmaps for fab rampups were late at every node. AMD essentially abandoned them for IBM for their 90 nm tech. Lastly, they did not have the funds to go at it alone at 90 nm, and had to combine forces with Philips and STMicro to get there. Crolles may actually be clean given that it seems to be a Philips fab, but Freescale has a lot of work to do to get their circuits working well at 90 nm.
No, they are not behind by choice. It is in their best interest to have the most advanced process node as possible with mutlicore embedded processors being offered by all of Freescale's competition. They are in a precarious situation right now, and will be in the foreseeable future.
Say what you want, but it's not true. Intel had a bad rep throughout the '80's and '90's. AMD has been a loser as far back as I can remember. IBM was known for great R&D but poor production.
So where are we now? Intel didn't go out of business in the '80's as IBM though they might, and dominate. AMD has has a streak for a couple of years, which they might not keep up. IBM still has great R&D but poor production.Moto had problems because they put all their chips (oh, a pun!) in the wrong basket; the PPC. It does great in the embedded markets, but not so good for general purpose machines. Most of the R&D costs come from that area. Cut it out and the company is much healther.
As of this moment, id much prefer a new machine with a far higher resolution display. IMO the UI is far too large with the current res on the 15" 1280x854, i want everything much smaller. Considering the so called Resolution Independent UI is still a long way off, this is a reasonable request.
Of course any additional mhz would always be appreciated.
I certainly will NOT be impressed if Apple end up offering Intel Powerbooks with PC BIOS's.
Say what you want, but it's not true. Intel had a bad rep throughout the '80's and '90's. AMD has been a loser as far back as I can remember. IBM was known for great R&D but poor production.
So where are we now? Intel didn't go out of business in the '80's as IBM though they might, and dominate. AMD has has a streak for a couple of years, which they might not keep up. IBM still has great R&D but poor production.Moto had problems because they put all their chips (oh, a pun!) in the wrong basket; the PPC. It does great in the embedded markets, but not so good for general purpose machines. Most of the R&D costs come from that area. Cut it out and the company is much healther.
Cost of entry for every successive node doubles, and companies who fall behind are at risk of not having the money to catch up. Freescale will not be catching up. The only companies to rely on into the future are Intel, IBM, and East Asian conglomerates, and I'm sticking to my prediction that only 2 entities will have 45 nm fabs: Intel and a conglomerate of IBM, AMD, east Asian companies.
But back to the original statement: a 90nm 1.7 GHz 7448 can have just the same power consumption as a 130 nm 1.7 GHz 7447. I can believe that and wouldn't be surprised if it was true, especially on the first run of the chip.
There is a BIG difference between a PowerBook and an iBook. Have you ever gone to their site and looked at the difference?
Better GPU, better HD, better I/O options, higher speed, better quality screens with much higher rez,better memory options, etc.
This is like the argument about the Bug vs the Mercedes when I was in college in the '70's. They both got you there but the Bug gave a much bumpier ride.
I don't think the small differences are worth the $1000 or so premium for the PBook right now. Then again I'm one who doubts that the premium for the Mercedes is worth it either.
Up until two years ago, the iBook was G3 and the PowerBook G4, but then Apple ran into this problem where they couldn't differentiate as clearly as I'm sure they'd like. My guess is that once the Intel transition is in place, the iBook will be single core and the PBook dual.
Cost of entry for every successive node doubles, and companies who fall behind are at risk of not having the money to catch up. Freescale will not be catching up. The only companies to rely on into the future are Intel, IBM, and East Asian conglomerates, and I'm sticking to my prediction that only 2 entities will have 45 nm fabs: Intel and a conglomerate of IBM, AMD, east Asian companies.
But back to the original statement: a 90nm 1.7 GHz 7448 can have just the same power consumption as a 130 nm 1.7 GHz 7447. I can believe that and wouldn't be surprised if it was true, especially on the first run of the chip.
Not according to the figures. Check the chart again;
I don't think the small differences are worth the $1000 or so premium for the PBook right now. Then again I'm one who doubts that the premium for the Mercedes is worth it either.
Up until two years ago, the iBook was G3 and the PowerBook G4, but then Apple ran into this problem where they couldn't differentiate as clearly as I'm sure they'd like. My guess is that once the Intel transition is in place, the iBook will be single core and the PBook dual.
If you need the features of the PB, the iBk isn't useful at all. It's that simple.
I'm not saying that the PB is for everyone. That's why it's the pro line and the iBk is the consumer line. You buy what you need. The iBk is a great machine. but it's simply not as fast or versatile as the PB .
I certainly don't criticize someone for saying that "for me, the iBk is just as good". But the statement "the iBk is just as good" just doesn't wash.
I think THT's point is those are design goals and manufacturing may tell a different tale.
No, that's not what they are at all. These are product specs for production units. AMD, IBM, and Intel also release product specs.
How else do you think a manufacturer can design a product around them? They must know the true, worst case, power envelope, as well as the average, and minimum power requirements, as well as a host of other specs that are released.
Without that they can't design heatsinks, predict battery life, etc.
For example; PPC chips are rated at 10 year lifetime at 105° C. That's not a design goal, it's the minimum lifetime at that temp.
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
Your numbers are so far off it's not funny. You really don't understand this, do you? The power figures include all the areas of concern. If you want to look further on their site you can find the figures. I'm going to dinner.
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
Sorry, as melgross said, Freescale cannot hand out these power specs to its customers who start to design their (embedded) applications based on the 15 W number and then when the chips ship in volume, say 'Oups, sorry folks the chips consume 25 W'.
And in case you understood this differently, these Freescale numbers mean that all processors up to (and including) 1.5 Ghz should consume a maximum of 10 W. And all those over 1.5 Ghz, i.e. 1.6 and 1.7 Ghz processors, should consume a maxium of 15 W.
Now, plotting Freescales numbers for the 7447As one can clearly see that there are two different slopes (power cons. vs. frequency) indicating two different voltages, perhaps 1.1 and 1.3 V. One can therefore speculate that the '< 10 W' and '< 15 W' also refer to two different voltages.
The current 1.67 Ghz processor is 17% faster than the official 1.42 Ghz limit for the 7447A, 2 Ghz would also be 17% faster than the official 1.7 Ghz limit for the 7448. At their respective official limits the 7447A consumes 21 W and the 7448 only 15 W. Under the assumption that power consumption scales the same for the 7447A and the 7448, a 2 Ghz Powerbook should therefore be quite feasible, there is even a 25% savety margin (21 W vs. 15 W), to compensate for the faster system bus and kinks in 90 nm like the leakage you mentioned.
Sorry, as melgross said, Freescale cannot hand out these power specs to its customers who start to design their (embedded) applications based on the 15 W number and then when the chips ship in volume, say 'Oups, sorry folks the chips consume 25 W'.
Heh. Freescale hasn't handed out any power numbers. The table in their product page are projections as far as I concerned. If they can achieve the values, then it's good for them. But I'll wait and see.
If you think it is impossible for a company to be wrong with their projections, just ask Apple about the 39 W max 2 GHz 970fx CPUs they were expecting last year.
Quote:
And in case you understood this differently, these Freescale numbers mean that all processors up to (and including) 1.5 Ghz should consume a maximum of 10 W. And all those over 1.5 Ghz, i.e. 1.6 and 1.7 Ghz processors, should consume a maxium of 15 W.
Not maximum, but typical power usage, likely running at 2.3 Dhrystone MIPS/MHz.
It's like you guys think I don't understand the power box Freescale is presenting. That's not it at all. I understand it, it's just that Freescale has a lot of history to overcome for me to believe that. Not to mention the known 90 nm transition troubles from arguably the best fabs in the world, either. You trust it at your peril.
Quote:
Now, plotting Freescales numbers for the 7447As one can clearly see that there are two different slopes (power cons. vs. frequency) indicating two different voltages, perhaps 1.1 and 1.3 V. One can therefore speculate that the '< 10 W' and '< 15 W' also refer to two different voltages.
Yes probably. But look at the 7448 column. For the N-spec, it is <10 W @ > 1 GHz. What GHz do you think that will hold true for? 1.1 GHz? 1.2 GHz? 1.4 GHz? 1.8 GHz? Likewise for the L-spec, it is <15 W @ > 1.5 GHz. What GHz do you think that will hold true? 1.6 GHz? 1.7 GHz? 1.8 GHz? 2 GHz?
It's betting game for what that frequency would be. 1.7 GHz is pretty close to that edge I think.
Quote:
The current 1.67 Ghz processor is 17% faster than the official 1.42 Ghz limit for the 7447A, 2 Ghz would also be 17% faster than the official 1.7 Ghz limit for the 7448. At their respective official limits the 7447A consumes 21 W and the 7448 only 15 W. Under the assumption that power consumption scales the same for the 7447A and the 7448, a 2 Ghz Powerbook should therefore be quite feasible, there is even a 25% savety margin (21 W vs. 15 W), to compensate for the faster system bus and kinks in 90 nm like the leakage you mentioned.
It all depends on what voltages are required to drive a 7448 from 1.5 GHz to 1.7 GHz. If no voltage change is required, power consumption is a straight linear scale with clock rate. If increased voltage is required, power consumption goes up be the square of the voltage ratio. And if it is increased voltage, it's more than quadratic penalty, since the power consumption due to leakage can be cubic (assuming Crolles haven't overcome those problems).
Heh. Freescale hasn't handed out any power numbers. The table in their product page are projections as far as I concerned. If they can achieve the values, then it's good for them. But I'll wait and see.
If you think it is impossible for a company to be wrong with their projections, just ask Apple about the 39 W max 2 GHz 970fx CPUs they were expecting last year.
Not maximum, but typical power usage, likely running at 2.3 Dhrystone MIPS/MHz.
It's like you guys think I don't understand the power box Freescale is presenting. That's not it at all. I understand it, it's just that Freescale has a lot of history to overcome for me to believe that. Not to mention the known 90 nm transition troubles from arguably the best fabs in the world, either. You trust it at your peril.
Yes probably. But look at the 7448 column. For the N-spec, it is <10 W @ > 1 GHz. What GHz do you think that will hold true for? 1.1 GHz? 1.2 GHz? 1.4 GHz? 1.8 GHz? Likewise for the L-spec, it is <15 W @ > 1.5 GHz. What GHz do you think that will hold true? 1.6 GHz? 1.7 GHz? 1.8 GHz? 2 GHz?
It's betting game for what that frequency would be. 1.7 GHz is pretty close to that edge I think.
It all depends on what voltages are required to drive a 7448 from 1.5 GHz to 1.7 GHz. If no voltage change is required, power consumption is a straight linear scale with clock rate. If increased voltage is required, power consumption goes up be the square of the voltage ratio. And if it is increased voltage, it's more than quadratic penalty, since the power consumption due to leakage can be cubic (assuming Crolles haven't overcome those problems).
You're making several mistakes. What you're doing is taking the numbers and deciding that you don't want to believe them. Then to reinforce your opinion, you say that they aren't real, but projections. Thern you say that projections aren't always met, and so on.
These aren't projections. They are real numbers. you don't have to be happy about them..
Also, the voltage doesn't matter. The watts matter. when they give a figure in watts, they are giving the power number. Volts times amps equals watts. Watts is power. However they get there doesn't matter. If the voltage is raised then the amperage is less. The watts used is the total of every area in the chip where it's consumed, including leakage cuttents etc.
At 1.7GHz the wattage is probably less than 18. Voltage is used to push through the gate. It's pressure. In and of itself, it doesn't mean that higher wattage is needed. The greater the voltage, the faster it can go. But that doesn't mean that a higher voltage is necessary to get to the highest currently rated speed. Usually it means that the chip can be run at that voltage, as when board makers clock the chips higher than the ratings call for. Such as the 1.7, 1.8, and 2GHz boards now being offered for the 7447a's. Some companies overclock the chips even they know they are out of spec. This tells them how high a voltage they can use and still be safe.
It's also used for later runs of the chips when they get the process working better and up the speeds. so maybe these chips will eventually be rated at 1.8-2.0GHz.
They've been available for months. Damn good ones too. It's just Apple's refusal to use the latest and greatest. They may not have a choice this time if these PowerBooks are to tide us over for another 8 to 12 months.
They've been available for months. Damn good ones too. It's just Apple's refusal to use the latest and greatest. They may not have a choice this time if these PowerBooks are to tide us over for another 8 to 12 months.
Yeah, and that's a nice drive. It came out after Apple's last PB update. We'll see what happens with this one.
Comments
Originally posted by SteveGTA
errr high definition wide screen is probably the best thing on the list. ddr2 ram isnt really that great because its latency is way higher than ddr and the bus speed jumps are good but its not like theyll help that much. i think it is a much needed update for the powerbook line though. there is basically no real reason to spend 500 more on a powerbook instead of an iBook.
It's not that simple. The performance depends on the difference between what the current memory is and what they will use. Dual channel adds between 5-10% to the speed of the memory
There is a BIG difference between a PowerBook and an iBook. Have you ever gone to their site and looked at the difference?
Better GPU, better HD, better I/O options, higher speed, better quality screens with much higher rez,better memory options, etc.
This is like the argument about the Bug vs the Mercedes when I was in college in the '70's. They both got you there but the Bug gave a much bumpier ride.
Originally posted by @homenow
The costs of ramping up a new product for one revision would be a lot higher than that spread out over 2-3+ years that a normal product exists. With the G5 they would have to contract out the tooling of a new MotherBoard, internals and case and recoup that cost in the (likely) 12 months that the product would be offered. Then there would be the cost of supporting that product with repair parts, documentation and training for repair techs, for the a few years as well. Even if all the R&D was done, and the chips ready to be delivered a wise person would scrap the release before all the other costs associated with the release were incured and go without an upgrade, opting instead to push foreword the timetable for the Intel switch. G5 isn't going to happen.
As for the G4, well PowerLogix has released a 2 Ghz 7447 upgrade for PM's, and with the smaller process and other improvements that the 7448 has one would logically assume that even if it is not anounced there is a faster chip available.
I seam to remember that there was a scandal with one of the upgrade manufacturers a number of years ago about processor speeds, and the company pointed out that the chips are are tested and certified to meet the requirements of the contract. This particular case was an issue of the upgrade company buying a lot that originally ordered by another customer and the chips were certified at one speed, but in and environment with a higher operating temperature than the chips would be used in. Given the lower temperature of the Mac case they meet the speed requirements that they were sold as and were not clocked up, however IBM (i think) had printed the clock speed from the original contract on them so there were complaints from customers about thier purchase.
Sure, but that doesn't mean that it wouldn't be worth it. Apple might wery well find its Powerbook sales lagging too much here before the switch. Look at the numbers. While iMacG5's sales were up 48%, the sales of the laptop line was only up 16% with most of that being iBooks because of the school contracts. So the PB's are seriously lagging even now. With all the waiting for the G5, it might pay if what I said could be done. If Apple could up the sales by a third by doing this it would be worth while. I know what it costs to start up a production line. I'v gone through it numerous times with my own company.
There is also the political victory. If they can get it out, and its performance is fairly good it would be a plus while it lasted. When a new x86 machine comes out with a new chip line and is even better, ther will be applause, and a rush to buy.
It's all marketing. Companies look at costs in more ways than one. some of those costs would be absorbed as publicity. The startup costs are high, but not as high as you might think. If no new case is required, it's not very high at all. A new mobo design is cheap. Third party PC board makers come out with new ones every three months, and they don't all sell as well as you might think.
The startup costs could be a few 10's of millions at most. If Apple could sell 100 thousand of them in the next 7 months, not really that many after all, at a price of $2,000 to Apple, that's $200 million. Not too shabby.
Originally posted by THT
Freescale, nee Motorola, has had a reputation for dirty fabs for years on end now, all the way back to 220 nm and the 7400 "500 MHz" debacle, and have gradually fallen further behind every year after that. Schedules and roadmaps for fab rampups were late at every node. AMD essentially abandoned them for IBM for their 90 nm tech. Lastly, they did not have the funds to go at it alone at 90 nm, and had to combine forces with Philips and STMicro to get there. Crolles may actually be clean given that it seems to be a Philips fab, but Freescale has a lot of work to do to get their circuits working well at 90 nm.
No, they are not behind by choice. It is in their best interest to have the most advanced process node as possible with mutlicore embedded processors being offered by all of Freescale's competition. They are in a precarious situation right now, and will be in the foreseeable future.
Say what you want, but it's not true. Intel had a bad rep throughout the '80's and '90's. AMD has been a loser as far back as I can remember. IBM was known for great R&D but poor production.
So where are we now? Intel didn't go out of business in the '80's as IBM though they might, and dominate. AMD has has a streak for a couple of years, which they might not keep up. IBM still has great R&D but poor production.Moto had problems because they put all their chips (oh, a pun!) in the wrong basket; the PPC. It does great in the embedded markets, but not so good for general purpose machines. Most of the R&D costs come from that area. Cut it out and the company is much healther.
Of course any additional mhz would always be appreciated.
I certainly will NOT be impressed if Apple end up offering Intel Powerbooks with PC BIOS's.
Originally posted by satchmo
Could we see the introduction of dual layer SuperDrives in Powerbooks at Paris?
That's just a matter of the drive manufacrurers.
Originally posted by melgross
That's just a matter of the drive manufacrurers.
and how apple feels about including such drives
Originally posted by melgross
Say what you want, but it's not true. Intel had a bad rep throughout the '80's and '90's. AMD has been a loser as far back as I can remember. IBM was known for great R&D but poor production.
So where are we now? Intel didn't go out of business in the '80's as IBM though they might, and dominate. AMD has has a streak for a couple of years, which they might not keep up. IBM still has great R&D but poor production.Moto had problems because they put all their chips (oh, a pun!) in the wrong basket; the PPC. It does great in the embedded markets, but not so good for general purpose machines. Most of the R&D costs come from that area. Cut it out and the company is much healther.
Cost of entry for every successive node doubles, and companies who fall behind are at risk of not having the money to catch up. Freescale will not be catching up. The only companies to rely on into the future are Intel, IBM, and East Asian conglomerates, and I'm sticking to my prediction that only 2 entities will have 45 nm fabs: Intel and a conglomerate of IBM, AMD, east Asian companies.
But back to the original statement: a 90nm 1.7 GHz 7448 can have just the same power consumption as a 130 nm 1.7 GHz 7447. I can believe that and wouldn't be surprised if it was true, especially on the first run of the chip.
Originally posted by melgross
There is a BIG difference between a PowerBook and an iBook. Have you ever gone to their site and looked at the difference?
Better GPU, better HD, better I/O options, higher speed, better quality screens with much higher rez,better memory options, etc.
This is like the argument about the Bug vs the Mercedes when I was in college in the '70's. They both got you there but the Bug gave a much bumpier ride.
I don't think the small differences are worth the $1000 or so premium for the PBook right now. Then again I'm one who doubts that the premium for the Mercedes is worth it either.
Up until two years ago, the iBook was G3 and the PowerBook G4, but then Apple ran into this problem where they couldn't differentiate as clearly as I'm sure they'd like. My guess is that once the Intel transition is in place, the iBook will be single core and the PBook dual.
Originally posted by THT
Cost of entry for every successive node doubles, and companies who fall behind are at risk of not having the money to catch up. Freescale will not be catching up. The only companies to rely on into the future are Intel, IBM, and East Asian conglomerates, and I'm sticking to my prediction that only 2 entities will have 45 nm fabs: Intel and a conglomerate of IBM, AMD, east Asian companies.
But back to the original statement: a 90nm 1.7 GHz 7448 can have just the same power consumption as a 130 nm 1.7 GHz 7447. I can believe that and wouldn't be surprised if it was true, especially on the first run of the chip.
Not according to the figures. Check the chart again;
http://www.freescale.com/webapp/sps/...468rH3bTdG8653
Originally posted by BRussell
I don't think the small differences are worth the $1000 or so premium for the PBook right now. Then again I'm one who doubts that the premium for the Mercedes is worth it either.
Up until two years ago, the iBook was G3 and the PowerBook G4, but then Apple ran into this problem where they couldn't differentiate as clearly as I'm sure they'd like. My guess is that once the Intel transition is in place, the iBook will be single core and the PBook dual.
If you need the features of the PB, the iBk isn't useful at all. It's that simple.
I'm not saying that the PB is for everyone. That's why it's the pro line and the iBk is the consumer line. You buy what you need. The iBk is a great machine. but it's simply not as fast or versatile as the PB .
I certainly don't criticize someone for saying that "for me, the iBk is just as good". But the statement "the iBk is just as good" just doesn't wash.
Originally posted by melgross
Not according to the figures. Check the chart again;
http://www.freescale.com/webapp/sps/...468rH3bTdG8653
I think THT's point is those are design goals and manufacturing may tell a different tale.
Originally posted by ChevalierMalFet
I think THT's point is those are design goals and manufacturing may tell a different tale.
No, that's not what they are at all. These are product specs for production units. AMD, IBM, and Intel also release product specs.
How else do you think a manufacturer can design a product around them? They must know the true, worst case, power envelope, as well as the average, and minimum power requirements, as well as a host of other specs that are released.
Without that they can't design heatsinks, predict battery life, etc.
For example; PPC chips are rated at 10 year lifetime at 105° C. That's not a design goal, it's the minimum lifetime at that temp.
Originally posted by melgross
Not according to the figures. Check the chart again;
http://www.freescale.com/webapp/sps/...468rH3bTdG8653
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
Originally posted by THT
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
Your numbers are so far off it's not funny. You really don't understand this, do you? The power figures include all the areas of concern. If you want to look further on their site you can find the figures. I'm going to dinner.
Originally posted by THT
I don't see any product specifications for the 7448 on the website. All I see are:
N-Spec "<10W @ >1.0 GHz"
L-Spec "<15W @ >1.5 GHz"
For all we know, those are the numbers Freescale is expecting to see next year. I'll wait for the hardware spec documentation to come out for the 7448 to see what the real numbers are.
Just some basic math using a 1.5 GHz 7448 at 1.1 V and 15 Watts would mean a 1.7 GHz 7448 at 1.3 V would be consuming 24 Watts. With increased leakage at 90 nm, yeah, I can see a 1.7 GHz 7448 having the same power consumption as a 1.7 GHz 7447A.
Sorry, as melgross said, Freescale cannot hand out these power specs to its customers who start to design their (embedded) applications based on the 15 W number and then when the chips ship in volume, say 'Oups, sorry folks the chips consume 25 W'.
And in case you understood this differently, these Freescale numbers mean that all processors up to (and including) 1.5 Ghz should consume a maximum of 10 W. And all those over 1.5 Ghz, i.e. 1.6 and 1.7 Ghz processors, should consume a maxium of 15 W.
Now, plotting Freescales numbers for the 7447As one can clearly see that there are two different slopes (power cons. vs. frequency) indicating two different voltages, perhaps 1.1 and 1.3 V. One can therefore speculate that the '< 10 W' and '< 15 W' also refer to two different voltages.
The current 1.67 Ghz processor is 17% faster than the official 1.42 Ghz limit for the 7447A, 2 Ghz would also be 17% faster than the official 1.7 Ghz limit for the 7448. At their respective official limits the 7447A consumes 21 W and the 7448 only 15 W. Under the assumption that power consumption scales the same for the 7447A and the 7448, a 2 Ghz Powerbook should therefore be quite feasible, there is even a 25% savety margin (21 W vs. 15 W), to compensate for the faster system bus and kinks in 90 nm like the leakage you mentioned.
Originally posted by noirdesir
Sorry, as melgross said, Freescale cannot hand out these power specs to its customers who start to design their (embedded) applications based on the 15 W number and then when the chips ship in volume, say 'Oups, sorry folks the chips consume 25 W'.
Heh. Freescale hasn't handed out any power numbers. The table in their product page are projections as far as I concerned. If they can achieve the values, then it's good for them. But I'll wait and see.
If you think it is impossible for a company to be wrong with their projections, just ask Apple about the 39 W max 2 GHz 970fx CPUs they were expecting last year.
And in case you understood this differently, these Freescale numbers mean that all processors up to (and including) 1.5 Ghz should consume a maximum of 10 W. And all those over 1.5 Ghz, i.e. 1.6 and 1.7 Ghz processors, should consume a maxium of 15 W.
Not maximum, but typical power usage, likely running at 2.3 Dhrystone MIPS/MHz.
It's like you guys think I don't understand the power box Freescale is presenting. That's not it at all. I understand it, it's just that Freescale has a lot of history to overcome for me to believe that. Not to mention the known 90 nm transition troubles from arguably the best fabs in the world, either. You trust it at your peril.
Now, plotting Freescales numbers for the 7447As one can clearly see that there are two different slopes (power cons. vs. frequency) indicating two different voltages, perhaps 1.1 and 1.3 V. One can therefore speculate that the '< 10 W' and '< 15 W' also refer to two different voltages.
Yes probably. But look at the 7448 column. For the N-spec, it is <10 W @ > 1 GHz. What GHz do you think that will hold true for? 1.1 GHz? 1.2 GHz? 1.4 GHz? 1.8 GHz? Likewise for the L-spec, it is <15 W @ > 1.5 GHz. What GHz do you think that will hold true? 1.6 GHz? 1.7 GHz? 1.8 GHz? 2 GHz?
It's betting game for what that frequency would be. 1.7 GHz is pretty close to that edge I think.
The current 1.67 Ghz processor is 17% faster than the official 1.42 Ghz limit for the 7447A, 2 Ghz would also be 17% faster than the official 1.7 Ghz limit for the 7448. At their respective official limits the 7447A consumes 21 W and the 7448 only 15 W. Under the assumption that power consumption scales the same for the 7447A and the 7448, a 2 Ghz Powerbook should therefore be quite feasible, there is even a 25% savety margin (21 W vs. 15 W), to compensate for the faster system bus and kinks in 90 nm like the leakage you mentioned.
It all depends on what voltages are required to drive a 7448 from 1.5 GHz to 1.7 GHz. If no voltage change is required, power consumption is a straight linear scale with clock rate. If increased voltage is required, power consumption goes up be the square of the voltage ratio. And if it is increased voltage, it's more than quadratic penalty, since the power consumption due to leakage can be cubic (assuming Crolles haven't overcome those problems).
Originally posted by THT
Heh. Freescale hasn't handed out any power numbers. The table in their product page are projections as far as I concerned. If they can achieve the values, then it's good for them. But I'll wait and see.
If you think it is impossible for a company to be wrong with their projections, just ask Apple about the 39 W max 2 GHz 970fx CPUs they were expecting last year.
Not maximum, but typical power usage, likely running at 2.3 Dhrystone MIPS/MHz.
It's like you guys think I don't understand the power box Freescale is presenting. That's not it at all. I understand it, it's just that Freescale has a lot of history to overcome for me to believe that. Not to mention the known 90 nm transition troubles from arguably the best fabs in the world, either. You trust it at your peril.
Yes probably. But look at the 7448 column. For the N-spec, it is <10 W @ > 1 GHz. What GHz do you think that will hold true for? 1.1 GHz? 1.2 GHz? 1.4 GHz? 1.8 GHz? Likewise for the L-spec, it is <15 W @ > 1.5 GHz. What GHz do you think that will hold true? 1.6 GHz? 1.7 GHz? 1.8 GHz? 2 GHz?
It's betting game for what that frequency would be. 1.7 GHz is pretty close to that edge I think.
It all depends on what voltages are required to drive a 7448 from 1.5 GHz to 1.7 GHz. If no voltage change is required, power consumption is a straight linear scale with clock rate. If increased voltage is required, power consumption goes up be the square of the voltage ratio. And if it is increased voltage, it's more than quadratic penalty, since the power consumption due to leakage can be cubic (assuming Crolles haven't overcome those problems).
You're making several mistakes. What you're doing is taking the numbers and deciding that you don't want to believe them. Then to reinforce your opinion, you say that they aren't real, but projections. Thern you say that projections aren't always met, and so on.
These aren't projections. They are real numbers. you don't have to be happy about them..
Also, the voltage doesn't matter. The watts matter. when they give a figure in watts, they are giving the power number. Volts times amps equals watts. Watts is power. However they get there doesn't matter. If the voltage is raised then the amperage is less. The watts used is the total of every area in the chip where it's consumed, including leakage cuttents etc.
At 1.7GHz the wattage is probably less than 18. Voltage is used to push through the gate. It's pressure. In and of itself, it doesn't mean that higher wattage is needed. The greater the voltage, the faster it can go. But that doesn't mean that a higher voltage is necessary to get to the highest currently rated speed. Usually it means that the chip can be run at that voltage, as when board makers clock the chips higher than the ratings call for. Such as the 1.7, 1.8, and 2GHz boards now being offered for the 7447a's. Some companies overclock the chips even they know they are out of spec. This tells them how high a voltage they can use and still be safe.
It's also used for later runs of the chips when they get the process working better and up the speeds. so maybe these chips will eventually be rated at 1.8-2.0GHz.
Originally posted by melgross
That's just a matter of the drive manufacrurers.
They've been available for months. Damn good ones too. It's just Apple's refusal to use the latest and greatest. They may not have a choice this time if these PowerBooks are to tide us over for another 8 to 12 months.
Originally posted by 1984
They've been available for months. Damn good ones too. It's just Apple's refusal to use the latest and greatest. They may not have a choice this time if these PowerBooks are to tide us over for another 8 to 12 months.
Yeah, and that's a nice drive. It came out after Apple's last PB update. We'll see what happens with this one.