I'm of the opinion that we will see a 32 bit processor for th next laptop go around. Considering what IBM did with the 90nm 970 I think they can deliver a 32 bit porcessor with an extremely low thermal profile. A SOC would obviously raise the temperature some but would prevent a great deal of losses else where.
Motorola did accomplish a great deal with the G4 but I have to disagree with the thought that it was leading edge. They simply struggled to much with scaling performance and considering the size of the processor, the thermal performance wasn't all that much to crow about either. Considering the power profile of some other RISC chips the G4 is OK but not outstanding.
It is easy to see what Motorola did with the G4. The issue is what Motorola hasn't done with the G4 and the number of screw ups along the way to today. They failed to compete in the performance desktop chip market and almost took Apple with them. The fact that the G4 was somewhat power frugal is more a result of it being a RISC machine more than anything.
Apple may not have been a big customer of motorola but they were a big customer for the G4. Loosing Apple is going to hurt motorola both in the pocket book and the credibility department.
IBM is indicating that they are about to start shipping prodcution runs of SOC at 90nm. We of course have no idea exactly what those chips are for, but my vote goes to the laptop line. If these do indeed end up in a laptop it will be very interesting to see how well they do power wise.
Quote:
Originally posted by smalM
VIA just announced the Eden ESP 10000.
7W Thermal Design Power.
1GHz
And don't think this CPU to be as fast as a G4 @ 1GHz!
The Moto bashing on this site has kept some people from seeing what Moto really did with the G4: Reduce power consumption. Point no. 1 on the wish list of their big customers. Yes, Apple is not a big customer to Moto.
BTW, ever looked under a G5 mother board?
The system controller gets massive cooling!
I don't think we see a G5 PB before the controller is build on the 90nm process.
The next CPU in the Powerbook has to be a SOC design if the Powerbooks have to compete and trounce the WINTEL competition. A normal update with just a processor speed increase will not cut it nor would plugging in a G5.
Forget heat generation, how long would the battery life be with a G5?
If they have a SOC chip with ethernet and WIFi and some other parts on the CPU fabbed on the 90nm process (which seems to be the new standard), I am sure that even a 2 GHZ G4 class processor (G3 + Altivec) would give back Apple its lead in laptop technology.
MOSR, he-he. I remember they predicted a quad G4 PowerBook as well.
MOSR, certainly a joke in these boards. However, wizard69's comment on SoC systems revived my interest in this discussion. I am waiting for the details...
The Moto bashing on this site has kept some people from seeing what Moto really did with the G4: Reduce power consumption. Point no. 1 on the wish list of their big customers. Yes, Apple is not a big customer to Moto.
Concerning G4 processors, Apple was by far the largest customer (probably still is). That's why I find it so odd that Moto pissed them off so much.
Some are speculating that it is a typo, but I don't see a slip up between an 'I' and a 'C' taking place. It could very well be a reference to the 90nm 970 as some have suggested.
I'm more of th opinion that it is not a mistake and that Apple is about to take delievery of another porcessor/SOC along with the 970. There are a couple of reasons why I lean this way.
One is the 970 itself, if this was a reference to the 970 they would have said 970 as this cat is already out of the bag. The second issue is that Apple desperately needs a laptop CPU and they are known to be working on such a chip with IBM. If you accept that it is a laptop chip then you will certainly accept that it is a 32 bit machine and is most likely desinged to afford the machine it goes into a maximum of power savings.
Since the entire Laptop line needs an upgrade and the 90nm 970 is still to hot for this product, an optimised 32 bit solution is most likely in the works.
Thanks
Dave
Quote:
Originally posted by PB
This reflects my concerns raised in the other thread, back in December.
The second issue is that Apple desperately needs a laptop CPU and they are known to be working on such a chip with IBM.
Do you know something we don't?
Quote:
Since the entire Laptop line needs an upgrade and the 90nm 970 is still to hot for this product, an optimised 32 bit solution is most likely in the works.
For the reasons I explained in the other thread, I expect and prefer too a 32 bit chip for the Powerbooks for the time being. Of course if this is going to be cost and heat-power effective. Still very obscure what exactly one could expect.
Remember the rumor that Apple had a contract to buy G4's at a set price from Moto, and Moto was losing money on every one due to low yields?
Well the flip side of that would be that Apple are committed to buying a minimum number of G4's every quarter. If they moved thier machines to G5's or 750vx's they would still have to buy the G4's, so no point.
I can see this sort of thing having happened, both to get the low price on G4's, and to get Moto's ongoing to support to G4 development.
It would also explain why the iBook went G4, to make up for the lose of the Powermac sales, towards a total target of G4 purchases ( say Apple comitted to buy 3 million ).
Remember the rumor that Apple had a contract to buy G4's at a set price from Moto, and Moto was losing money on every one due to low yields?
.
.
.
It would also explain why the iBook went G4, to make up for the lose of the Powermac sales, towards a total target of G4 purchases ( say Apple comitted to buy 3 million ).
I remember about a contract with Moto that already ran out some time ago, and, supposedly, Apple got rid of Moto's hell; I am not sure if we talk about the same thing. I too remember some rumors wanting Apple to sue Moto because of contract violation (fabulous Moto inability to deliver, especially in the G5 project).
APPLE is about to bring a lawsuit against Motorola:
The law firm Orrick, famous in the difficult battles for the account of large American companies, was engaged by APPLE to launch out in the battle.
APPLE intends to require approximately 500 million dollars of compensation at Motorola, and this by showing 4 essential facts:
1 Motorola is responsible for the fall on behalf of market of APPLE, because the consumers had a perception of inferiority of the APPLE products.
the 2 economic conditions of then did not have an effect more significant on APPLE than on the remainder of the actors of this market.
3 Motorola hid elements and significant facts with APPLE.
4 APPLE could not make differently than to remain to await Motorola because the costs of change would have been exhorbitants.
APPLE will support its arguments by showing that its collaboration with Motorala was very largely degraded since the year 2000, and this because of Motorola."
Interesting how the article repeatedly refers to "future Power Macs" but never specifically says in future "PowerBooks;" just say it could be used in future notebooks, routers, etc. But half the power consumption at today's present peak clock speed is encouraging nonetheless.
I think it is tomorrow the details about PowerTune, no? Stay tuned ...
Today, IBM talked about Power5 and its SMT capabilities. Tomorrow (wed, 18th) they present PowerTune and they have another talk or presentation scheduled for monday.
MacRumors quotes a senior IBM staff member as stating the 970FX uses 50W at 2.5Ghz and 24.5W at 2Ghz, compared to the 970 using 66W at a slower clock speed (was that 1.8 Ghz or 2.0? I believe 1.8 ). This should make a 1.4 - 1.8 Ghz piece suitable for notebooks.
...as stating the 970FX uses 50W at 2.5Ghz and 24.5W at 2Ghz, compared to the 970 using 66W at a slower clock speed (was that 1.8 Ghz or 2.0? I believe 1.8 ). This should make a 1.4 - 1.8 Ghz piece suitable for notebooks.
I am no CPU engineer, but isn't a twofold increase in the energy consumption over 500MHz a bit strange? Doesn't seem like a very smooth curve in power usage.
But if they'll drop these in PowerBooks soon, then woo-hoo!!! I wonder if it's more the other architecture changes that need to made to accomodate the G5 and its pipeline rather than the heat problem that is holding back introduction (or as others have said again and again, Apple wants to milk the PowerBook G4s for everything they can).
From what I've been hearing, it seems as though the 750VX is a G4. I mean, it's all a G4 ever was... a G3 with a vector processing unit (altivec).
Not really (if you mean the 750x G3/74xx G4): they're both 32 bit PowerPCs but internally they're quite different. The G3 uses the same 60x bus as G1 and G2 PowerPCs, which is bandwidth impaired compared to MPX (G4) and "elastic bus" (G5), and bad for Altivec performance. The G3 would need a new bus if Altivec was added. (starts crazy dreams of a eBus G3++ with Altivec...)
Quote:
I am no CPU engineer, but isn't a twofold increase in the energy consumption over 500MHz a bit strange? Doesn't seem like a very smooth curve in power usage.
Well, you've only got 2 datapoints, so it wouldn't be that smooth.
Comments
Motorola did accomplish a great deal with the G4 but I have to disagree with the thought that it was leading edge. They simply struggled to much with scaling performance and considering the size of the processor, the thermal performance wasn't all that much to crow about either. Considering the power profile of some other RISC chips the G4 is OK but not outstanding.
It is easy to see what Motorola did with the G4. The issue is what Motorola hasn't done with the G4 and the number of screw ups along the way to today. They failed to compete in the performance desktop chip market and almost took Apple with them. The fact that the G4 was somewhat power frugal is more a result of it being a RISC machine more than anything.
Apple may not have been a big customer of motorola but they were a big customer for the G4. Loosing Apple is going to hurt motorola both in the pocket book and the credibility department.
IBM is indicating that they are about to start shipping prodcution runs of SOC at 90nm. We of course have no idea exactly what those chips are for, but my vote goes to the laptop line. If these do indeed end up in a laptop it will be very interesting to see how well they do power wise.
Originally posted by smalM
VIA just announced the Eden ESP 10000.
7W Thermal Design Power.
1GHz
And don't think this CPU to be as fast as a G4 @ 1GHz!
The Moto bashing on this site has kept some people from seeing what Moto really did with the G4: Reduce power consumption. Point no. 1 on the wish list of their big customers. Yes, Apple is not a big customer to Moto.
BTW, ever looked under a G5 mother board?
The system controller gets massive cooling!
I don't think we see a G5 PB before the controller is build on the 90nm process.
Forget heat generation, how long would the battery life be with a G5?
If they have a SOC chip with ethernet and WIFi and some other parts on the CPU fabbed on the 90nm process (which seems to be the new standard), I am sure that even a 2 GHZ G4 class processor (G3 + Altivec) would give back Apple its lead in laptop technology.
Originally posted by wizard69
I'm of the opinion that we will see a 32 bit processor for th next laptop go around.
This reflects my concerns raised in the other thread, back in December.
IBM is indicating that they are about to start shipping prodcution runs of SOC at 90nm.
Now, where this comes from?
I guess that poking their stick between 1.25 and 1.53GHz they cover all the range of modern G4s.
*CONFIRMED* the next PowerBook revision will have a hard disk!
Originally posted by costique
MOSR, he-he. I remember they predicted a quad G4 PowerBook as well.
Originally posted by smalM
The Moto bashing on this site has kept some people from seeing what Moto really did with the G4: Reduce power consumption. Point no. 1 on the wish list of their big customers. Yes, Apple is not a big customer to Moto.
Concerning G4 processors, Apple was by far the largest customer (probably still is). That's why I find it so odd that Moto pissed them off so much.
Some are speculating that it is a typo, but I don't see a slip up between an 'I' and a 'C' taking place. It could very well be a reference to the 90nm 970 as some have suggested.
I'm more of th opinion that it is not a mistake and that Apple is about to take delievery of another porcessor/SOC along with the 970. There are a couple of reasons why I lean this way.
One is the 970 itself, if this was a reference to the 970 they would have said 970 as this cat is already out of the bag. The second issue is that Apple desperately needs a laptop CPU and they are known to be working on such a chip with IBM. If you accept that it is a laptop chip then you will certainly accept that it is a 32 bit machine and is most likely desinged to afford the machine it goes into a maximum of power savings.
Since the entire Laptop line needs an upgrade and the 90nm 970 is still to hot for this product, an optimised 32 bit solution is most likely in the works.
Thanks
Dave
Originally posted by PB
This reflects my concerns raised in the other thread, back in December.
Now, where this comes from?
Originally posted by wizard69
Electronic News!
What Electronic News?
The second issue is that Apple desperately needs a laptop CPU and they are known to be working on such a chip with IBM.
Do you know something we don't?
Since the entire Laptop line needs an upgrade and the 90nm 970 is still to hot for this product, an optimised 32 bit solution is most likely in the works.
For the reasons I explained in the other thread, I expect and prefer too a 32 bit chip for the Powerbooks for the time being. Of course if this is going to be cost and heat-power effective. Still very obscure what exactly one could expect.
EDIT: aha, I just saw the other thread.
Originally posted by PB
Now, where this comes from?
"IBM is about to deliver its very first volume production of an SOC on 90nm, said Reeves. This will be a design for Apple."
link
There is a Thread dedicated to this mysterious SOC tool.
Originally posted by Smircle
There is a Thread dedicated to this mysterious SOC tool.
I was already in that thread; thanks anyway.
Stingerman wrote:
The 90NM 970FX is faster at the same clock speed than the 130NM 970.
It is??? I thought it had the same cache and same pipeline. Where would it get the extra speed?
Well the flip side of that would be that Apple are committed to buying a minimum number of G4's every quarter. If they moved thier machines to G5's or 750vx's they would still have to buy the G4's, so no point.
I can see this sort of thing having happened, both to get the low price on G4's, and to get Moto's ongoing to support to G4 development.
It would also explain why the iBook went G4, to make up for the lose of the Powermac sales, towards a total target of G4 purchases ( say Apple comitted to buy 3 million ).
Originally posted by mmmpie
Remember the rumor that Apple had a contract to buy G4's at a set price from Moto, and Moto was losing money on every one due to low yields?
.
.
.
It would also explain why the iBook went G4, to make up for the lose of the Powermac sales, towards a total target of G4 purchases ( say Apple comitted to buy 3 million ).
I remember about a contract with Moto that already ran out some time ago, and, supposedly, Apple got rid of Moto's hell; I am not sure if we talk about the same thing. I too remember some rumors wanting Apple to sue Moto because of contract violation (fabulous Moto inability to deliver, especially in the G5 project).
"2004-02-14 - The future battle of APPLE
APPLE is about to bring a lawsuit against Motorola:
The law firm Orrick, famous in the difficult battles for the account of large American companies, was engaged by APPLE to launch out in the battle.
APPLE intends to require approximately 500 million dollars of compensation at Motorola, and this by showing 4 essential facts:
1 Motorola is responsible for the fall on behalf of market of APPLE, because the consumers had a perception of inferiority of the APPLE products.
the 2 economic conditions of then did not have an effect more significant on APPLE than on the remainder of the actors of this market.
3 Motorola hid elements and significant facts with APPLE.
4 APPLE could not make differently than to remain to await Motorola because the costs of change would have been exhorbitants.
APPLE will support its arguments by showing that its collaboration with Motorala was very largely degraded since the year 2000, and this because of Motorola."
16 February 2004
IBM reveals chip process trinity
PowerPC 970FX chip can be used in PCs, laptops and networking equipment
Originally posted by PB
I think it is tomorrow the details about PowerTune, no? Stay tuned
Today, IBM talked about Power5 and its SMT capabilities. Tomorrow (wed, 18th) they present PowerTune and they have another talk or presentation scheduled for monday.
MacRumors quotes a senior IBM staff member as stating the 970FX uses 50W at 2.5Ghz and 24.5W at 2Ghz, compared to the 970 using 66W at a slower clock speed (was that 1.8 Ghz or 2.0? I believe 1.8 ). This should make a 1.4 - 1.8 Ghz piece suitable for notebooks.
Originally posted by Smircle
...as stating the 970FX uses 50W at 2.5Ghz and 24.5W at 2Ghz, compared to the 970 using 66W at a slower clock speed (was that 1.8 Ghz or 2.0? I believe 1.8 ). This should make a 1.4 - 1.8 Ghz piece suitable for notebooks.
I am no CPU engineer, but isn't a twofold increase in the energy consumption over 500MHz a bit strange? Doesn't seem like a very smooth curve in power usage.
But if they'll drop these in PowerBooks soon, then woo-hoo!!! I wonder if it's more the other architecture changes that need to made to accomodate the G5 and its pipeline rather than the heat problem that is holding back introduction (or as others have said again and again, Apple wants to milk the PowerBook G4s for everything they can).
From what I've been hearing, it seems as though the 750VX is a G4. I mean, it's all a G4 ever was... a G3 with a vector processing unit (altivec).
Not really (if you mean the 750x G3/74xx G4
I am no CPU engineer, but isn't a twofold increase in the energy consumption over 500MHz a bit strange? Doesn't seem like a very smooth curve in power usage.
Well, you've only got 2 datapoints, so it wouldn't be that smooth.