[quote]As the first in a new family of high-end PowerPC processors, the chip is designed for initial speeds of up to 1.8 gigahertz, manipulating data in larger, 64-bit chunks and accelerating compute-intensive workloads like multimedia and graphics through specialized circuitry known as a single instruction multiple data (SIMD) unit.<hr></blockquote>
[quote]IBM plans to pack performance and new features into the chip using ultra-thin 0.13-micron circuitry (nearly 800 times thinner than a human hair), constructed of copper wiring and about 52 million transistors based on IBM?s efficient silicon-on-insulator (SOI) technology.<hr></blockquote>
[quote]IBM plans to make the PowerPC 970 chip available next year.<hr></blockquote>
[quote]Running at a speed of up to 900 megahertz, the bus can deliver information to the processor at up to 6.4 gigabytes per second, to help ensure that the high-performance processor is fed data at sufficient speeds.<hr></blockquote>
The PPC 970 will begin on a 130 nm process. I remember that this chip foundry it's being fabbed at will be starting at 130 nm, but will soon migrate to 90 nm. If the PPC 970 is included in the migration to 90 nm, then the speed should scale VERY nicely, and we may even see this chip in Powerbooks.
I'm still down about the timeline for it's introduction. MWSF04 is just too damn long, I can't imagine how Apple will keep their Powermac sales up for much longer with the ever-increasing performance gap. Something must give sooner or later.
I'm very hopeful that IBM is giving a pessimistic timeline for the PPC970's introduction so that it doesn't affect Apple's sales as much. From a marketing perspective, it would help Apple's Powermac sales more if potential buyers thought this IBM CPU was a long ways off. After all, fall 03 is only a prediction, and if IBM and Apple could get their sh!t together sooner then all the better.
If reports of OS X running on PPC970 prototypes in March are correct, then I fail to see why it would take another full year to get this CPU up and running. Apple can get a mobo designed around the prototype CPUs, and IBM can ramp up to full production by the end of this year (the damn plant opened in mid-summer!). So Apple could announce PPC970 Towers at MWSF03, and ship in Feb or March.
I know I'm setting myself up for a disappointment, but it's difficult to see how Apple could keep their Powermac sales afloat without the leap in performance that the PPC970 promises.
[quote] Quote Dude, why the frick are you getting your panties up in a knot because the thing only runs at 1.8 GHz. I don't care what speed the CPUs Apple uses run at as long as they are truly faster than x86 based machines. The POWER4 chip runs at what, top speed 1.3 GHz or something? Might even be lower than the G4 for all I know, but who cares? The thing is a moster, and having a PowerPC processor that goes into Mac systems based on the same technology is killer. I think I'm gonna pee my pants come Tuesday.
LETS GO BIG BLUE<hr></blockquote>
From what source do we know, and what stats that we have leads us to believe the power4 is rip roaring fast @1.3GHz?
I have never seen any power4 benchmarks vs. intel, or anything for that matter. Does anyone have a link?
Also, on the first page of the thread there was a link to a <a href="http://www.wired.com/news/mac/0,2125,55722,00.html" target="_blank">Wired</a> Article that said the new IBM processors did 8 instructions at once rather than the 3 that the current G4's do now. Is this true? I didn't notice that information anywhere other than the wired article. :cool:
Now I don't care about x86 in this list. What I want to know is to see the new P4PC (my new acronym ) chip on that list. Because Power4 has dual core + high L3, and the P4PC has higher clock speed.
Being that Power4 is dual cored does that mean that some instructions are doubled? Will "P4PC" being a single core Processor reduce performance greatly?
Now I don't care about x86 in this list. What I want to know is to see the new P4PC (my new acronym ) chip on that list. Because Power4 has dual core + high L3, and the P4PC has higher clock speed.
If you are talking about 'just spec benchmarks' then yes I could see IBM being able to do this.
The GPUL could bench out better in the SPEC test since this new CPU is a single core chip (I was kinda surprised about that). In real world usage a DUAL or QUAD core Power4 would KILL this chip.
[ everyone should know that all the standard quoted SPEC numbers are for SINGLE CORE versions of the Power4 (not the dual or quads) so when someone says 'the intel/amd are close to beating the power4 you can feel free to throw that back in their face' ]
I'm not sure I have this totally right but....
IBM started the Power4 as a DUAL CORE cpu and then as the process went on they found a way to make use of 'bad' two core chips. When the 'bad' part of the CPU was only affecting a single core IBM started to market single core versions of the Power4.
One would hope that as IBM had more time to look at the building process they should have been able to produce more 'good' (dual or quad core) CPUs and less and less single core (bad) chips.
If IBM is at that stage where they don't really have all that many 'bad' (single core) CPUs comming off the line they need to do one of two things.
CRIPPLE a GOOD dual core CPU. (not too smart)
Have something else to use IN PLACE OF a single core Power4.
Now they have that 'something else' in the form of the GPUL. Faster than a 'single core' power4 but still keeps the dual-quad core Power4s on the top of the list for big iron power servers.
This is just me thinking aloud so take it for what it's worth but it seems to make sense.
At certain tasks most likely. For example in common tasks like web browsing, word processing, excel, etc. But for highly optimized multiprocessor apps, not.
1) dual cores v. dual chips: Dual cores will make for lower yields thus increasing costs. Dual core also locks the consumer into making a decision at buy time, instead of just buying a board and putting in another cpu if you want later. Well, I suppose apple is already like this, but in the pc world where you build your own, it's a nice choice to have.
2) "If reports of OS X running on PPC970 prototypes in March are correct, then I fail to see why it would take another full year to get this CPU up and running. "
And Itanium has been in development for ten years, hammer for a few years (with initial silicon shown much more than a year before it will launch)... These things take time. Also, they have a prototype working. yippee. Now they need to take that prototype, scale it, while maintaining yields. As amd is finding out, it's not a trivial task.
3) as for the move to .09 being beneficial... it may be, but it may also reveal some hotspots on the die that were spaced out more on .13, so they were ok, that become problematic on .09. Thus the processor needs a re-layout, and be tweaked as necessary, before it can be mass produced at speeds. The thoroughbred A to tbred B transition is a stellar example of this.
4 "If one 1.8GHz PPC970s isn't enough to clobber Intel and AMD, two should do the job."
As if multi-processor xeons, athlon mps, or hammers wouldn't turn that right back around.
I too am excited as anyone, however I have 2 questions...
1. Will the introduction of these chips into PowerMacs (the obvious first choice) make the iMac, eMac, iBook and PowerBook look like absolute dogs and put people of purchasing them until an upgrade is made to them? Would this situation be avoided by the introduction of a workstation class machine or would it just be pointless...
2. Do you think that IBM and Apple may have an agreement that the 970 goes on release for general sale in Q3 2003 but that Apple get 1st dibs and can use it in Q1 maybe Q2?......
[quote] IBM plans to make the PowerPC 970 chip available next year.<hr></blockquote>
Now I understand that this doesn't mean that we will get it in January, but I think that (as per G4 introduction) it may be a little earlier than many people are giving IBM credit for.
Hm.. According to the specFp2000 site posted above, the power4 is about 1.33 times faster at 1300 MHz that a 2,8 GHz P4. Allthough that isnt bad at all, the PPC970 need to be able to scale fast enough to keep up with P4 and Itaniums and Hammers when they come out.
Allthough I have confidence in IBM being able to do just that.
However, I guess 1.3 times speed increase + altivec + faster bus speed (900Mhz - WOW), means a 2-3 times better performance that the "fastest pentium in the world", in Apple marketing
I think we have some Photoshop/Shake demos ahead of us at MacWorlds
Anyways, Im too excited about this news! I just hope this will be announces in january, and avaliable in march. (Yes, I just love to be able to be dissapointed in 5 months time
Well, well.. Enough rambling
FAST EDIT: the Power4 at 1100 MHz only seems to be 1.05 times faster than the 2.8 GHz P4. Still good for being 2.5 times "slower" MHz wise...
What is Apple going to sell in the mean time. How many people are going to buy PM's until this machine arrives? I should think Apple are worried and estatic at the same time.
<strong>Anybody else disappointed this chip isn't dual-cored?</strong><hr></blockquote>
I am a little but it does give System makers more flexibility in configurations. And a future 980 or 990 may have dual cores and even quad once moved to 90nm or lower.
I'm still down about the timeline for it's introduction. MWSF04 is just too damn long, . . . I'm very hopeful that IBM is giving a pessimistic timeline for the PPC970's introduction so that it doesn't affect Apple's sales as much. . .
</strong><hr></blockquote>
There are a few ways to hope for something sooner, but don't count on it and get discouraged.
First, no one knows better than Apple how desperately they need this IBM 970. I believe Apple will have the PowerMacs sitting in the manufacturing plant, just waiting for chips to come from IBM.
Second, Apple may introduce the new PowerMacs with two low-end G4 models, and two high-end G5 (IBM 970) models. The highest model may be a new workstation class PowerMac. In this way, those who do not need the highest performance will buy one of the low-end G4 models and keep up sales. Those who wish the G5 will wait, as others did for the 1.25 GHz models.
Third, if you read very carefully, IBM does not say we have to wait that long. Even if the IBM 970 ships to Apple manufacturing in March, the 970 will still 'be available' in the second half of 2003. IBM never said that the 970 will 'not be available until' the second half of 2003. I am sure IBM would love to state that everything went so smoothly that they shipped early.
(Edit add) Fourth, in the official IBM news release, they say available next year. The second half of 2003 was from the Forbes story.
Comments
Can't wait to do some protein folding on that thing!
How much faster than my 600mhz ibookG3? Mmmm...
[quote]As the first in a new family of high-end PowerPC processors, the chip is designed for initial speeds of up to 1.8 gigahertz, manipulating data in larger, 64-bit chunks and accelerating compute-intensive workloads like multimedia and graphics through specialized circuitry known as a single instruction multiple data (SIMD) unit.<hr></blockquote>
[quote]IBM plans to pack performance and new features into the chip using ultra-thin 0.13-micron circuitry (nearly 800 times thinner than a human hair), constructed of copper wiring and about 52 million transistors based on IBM?s efficient silicon-on-insulator (SOI) technology.<hr></blockquote>
[quote]IBM plans to make the PowerPC 970 chip available next year.<hr></blockquote>
[quote]Running at a speed of up to 900 megahertz, the bus can deliver information to the processor at up to 6.4 gigabytes per second, to help ensure that the high-performance processor is fed data at sufficient speeds.<hr></blockquote>
The PPC 970 will begin on a 130 nm process. I remember that this chip foundry it's being fabbed at will be starting at 130 nm, but will soon migrate to 90 nm. If the PPC 970 is included in the migration to 90 nm, then the speed should scale VERY nicely, and we may even see this chip in Powerbooks.
I'm still down about the timeline for it's introduction. MWSF04 is just too damn long, I can't imagine how Apple will keep their Powermac sales up for much longer with the ever-increasing performance gap. Something must give sooner or later.
I'm very hopeful that IBM is giving a pessimistic timeline for the PPC970's introduction so that it doesn't affect Apple's sales as much. From a marketing perspective, it would help Apple's Powermac sales more if potential buyers thought this IBM CPU was a long ways off. After all, fall 03 is only a prediction, and if IBM and Apple could get their sh!t together sooner then all the better.
If reports of OS X running on PPC970 prototypes in March are correct, then I fail to see why it would take another full year to get this CPU up and running. Apple can get a mobo designed around the prototype CPUs, and IBM can ramp up to full production by the end of this year (the damn plant opened in mid-summer!). So Apple could announce PPC970 Towers at MWSF03, and ship in Feb or March.
I know I'm setting myself up for a disappointment, but it's difficult to see how Apple could keep their Powermac sales afloat without the leap in performance that the PPC970 promises.
LETS GO BIG BLUE<hr></blockquote>
From what source do we know, and what stats that we have leads us to believe the power4 is rip roaring fast @1.3GHz?
I have never seen any power4 benchmarks vs. intel, or anything for that matter. Does anyone have a link?
Also, on the first page of the thread there was a link to a <a href="http://www.wired.com/news/mac/0,2125,55722,00.html" target="_blank">Wired</a> Article that said the new IBM processors did 8 instructions at once rather than the 3 that the current G4's do now. Is this true? I didn't notice that information anywhere other than the wired article. :cool:
1. Alpha 21264C at 1250MHz
2. Itanium2 at 1000MHz
3. POWER4 at 1300MHz
4. SPARC64 V at 1350MHz
5. POWER4 at 1100MHz
6. Alpha 21264C at 1224MHz
7. Alpha 21264C at 1000MHz
8. Pentium 4 at 2.8 GHz
9. Pentium 4 at 2.66 GHz
10. Pentium 4 at 2.53 GHz
Now I don't care about x86 in this list. What I want to know is to see the new P4PC (my new acronym ) chip on that list. Because Power4 has dual core + high L3, and the P4PC has higher clock speed.
These are also FP calcs:
<a href="http://www.ideasinternational.com/benchmark/spec/specfp2000.html" target="_blank">http://www.ideasinternational.com/benchmark/spec/specfp2000.html</a>
HTH,
Being that Power4 is dual cored does that mean that some instructions are doubled? Will "P4PC" being a single core Processor reduce performance greatly?
<strong>From slashdot.org:
1. Alpha 21264C at 1250MHz
2. Itanium2 at 1000MHz
3. POWER4 at 1300MHz
4. SPARC64 V at 1350MHz
5. POWER4 at 1100MHz
6. Alpha 21264C at 1224MHz
7. Alpha 21264C at 1000MHz
8. Pentium 4 at 2.8 GHz
9. Pentium 4 at 2.66 GHz
10. Pentium 4 at 2.53 GHz
Now I don't care about x86 in this list. What I want to know is to see the new P4PC (my new acronym ) chip on that list. Because Power4 has dual core + high L3, and the P4PC has higher clock speed.
These are also FP calcs:
<a href="http://www.ideasinternational.com/benchmark/spec/specfp2000.html" target="_blank">http://www.ideasinternational.com/benchmark/spec/specfp2000.html</a>
HTH,</strong><hr></blockquote>
All the SPEC stuff on the Power4 is with a SINGLE CORE version of the Power4. Most people forget to mention that and/or don't realize that.
Dave
<strong>So will the Power4 "Lite" aka 970 @1800mhz be faster than the Power4 @1300mhz?</strong><hr></blockquote>
Total speculation....
If you are talking about 'just spec benchmarks' then yes I could see IBM being able to do this.
The GPUL could bench out better in the SPEC test since this new CPU is a single core chip (I was kinda surprised about that). In real world usage a DUAL or QUAD core Power4 would KILL this chip.
[ everyone should know that all the standard quoted SPEC numbers are for SINGLE CORE versions of the Power4 (not the dual or quads) so when someone says 'the intel/amd are close to beating the power4 you can feel free to throw that back in their face' ]
I'm not sure I have this totally right but....
IBM started the Power4 as a DUAL CORE cpu and then as the process went on they found a way to make use of 'bad' two core chips. When the 'bad' part of the CPU was only affecting a single core IBM started to market single core versions of the Power4.
One would hope that as IBM had more time to look at the building process they should have been able to produce more 'good' (dual or quad core) CPUs and less and less single core (bad) chips.
If IBM is at that stage where they don't really have all that many 'bad' (single core) CPUs comming off the line they need to do one of two things.
CRIPPLE a GOOD dual core CPU. (not too smart)
Have something else to use IN PLACE OF a single core Power4.
Now they have that 'something else' in the form of the GPUL. Faster than a 'single core' power4 but still keeps the dual-quad core Power4s on the top of the list for big iron power servers.
This is just me thinking aloud so take it for what it's worth but it seems to make sense.
Dave
[ 10-14-2002: Message edited by: DaveGee ]</p>
1) dual cores v. dual chips: Dual cores will make for lower yields thus increasing costs. Dual core also locks the consumer into making a decision at buy time, instead of just buying a board and putting in another cpu if you want later. Well, I suppose apple is already like this, but in the pc world where you build your own, it's a nice choice to have.
2) "If reports of OS X running on PPC970 prototypes in March are correct, then I fail to see why it would take another full year to get this CPU up and running. "
And Itanium has been in development for ten years, hammer for a few years (with initial silicon shown much more than a year before it will launch)... These things take time. Also, they have a prototype working. yippee. Now they need to take that prototype, scale it, while maintaining yields. As amd is finding out, it's not a trivial task.
3) as for the move to .09 being beneficial... it may be, but it may also reveal some hotspots on the die that were spaced out more on .13, so they were ok, that become problematic on .09. Thus the processor needs a re-layout, and be tweaked as necessary, before it can be mass produced at speeds. The thoroughbred A to tbred B transition is a stellar example of this.
4 "If one 1.8GHz PPC970s isn't enough to clobber Intel and AMD, two should do the job."
As if multi-processor xeons, athlon mps, or hammers wouldn't turn that right back around.
neye
1. Will the introduction of these chips into PowerMacs (the obvious first choice) make the iMac, eMac, iBook and PowerBook look like absolute dogs and put people of purchasing them until an upgrade is made to them? Would this situation be avoided by the introduction of a workstation class machine or would it just be pointless...
2. Do you think that IBM and Apple may have an agreement that the 970 goes on release for general sale in Q3 2003 but that Apple get 1st dibs and can use it in Q1 maybe Q2?......
[quote] IBM plans to make the PowerPC 970 chip available next year.<hr></blockquote>
Now I understand that this doesn't mean that we will get it in January, but I think that (as per G4 introduction) it may be a little earlier than many people are giving IBM credit for.
Allthough I have confidence in IBM being able to do just that.
However, I guess 1.3 times speed increase + altivec + faster bus speed (900Mhz - WOW), means a 2-3 times better performance that the "fastest pentium in the world", in Apple marketing
I think we have some Photoshop/Shake demos ahead of us at MacWorlds
Anyways, Im too excited about this news! I just hope this will be announces in january, and avaliable in march. (Yes, I just love to be able to be dissapointed in 5 months time
Well, well.. Enough rambling
FAST EDIT: the Power4 at 1100 MHz only seems to be 1.05 times faster than the 2.8 GHz P4. Still good for being 2.5 times "slower" MHz wise...
.: BoeManE :.
[ 10-14-2002: Message edited by: BoeManE ]</p>
What is Apple going to sell in the mean time. How many people are going to buy PM's until this machine arrives? I should think Apple are worried and estatic at the same time.
<strong>Anybody else disappointed this chip isn't dual-cored?</strong><hr></blockquote>
I am a little but it does give System makers more flexibility in configurations. And a future 980 or 990 may have dual cores and even quad once moved to 90nm or lower.
edit - nevermind, I see lots of replies were posted between when I started reading and when i made this post.
[ 10-14-2002: Message edited by: rogue27 ]</p>
<strong>
I'm still down about the timeline for it's introduction. MWSF04 is just too damn long, . . . I'm very hopeful that IBM is giving a pessimistic timeline for the PPC970's introduction so that it doesn't affect Apple's sales as much. . .
</strong><hr></blockquote>
There are a few ways to hope for something sooner, but don't count on it and get discouraged.
First, no one knows better than Apple how desperately they need this IBM 970. I believe Apple will have the PowerMacs sitting in the manufacturing plant, just waiting for chips to come from IBM.
Second, Apple may introduce the new PowerMacs with two low-end G4 models, and two high-end G5 (IBM 970) models. The highest model may be a new workstation class PowerMac. In this way, those who do not need the highest performance will buy one of the low-end G4 models and keep up sales. Those who wish the G5 will wait, as others did for the 1.25 GHz models.
Third, if you read very carefully, IBM does not say we have to wait that long. Even if the IBM 970 ships to Apple manufacturing in March, the 970 will still 'be available' in the second half of 2003. IBM never said that the 970 will 'not be available until' the second half of 2003. I am sure IBM would love to state that everything went so smoothly that they shipped early.
(Edit add) Fourth, in the official IBM news release, they say available next year. The second half of 2003 was from the Forbes story.
[ 10-14-2002: Message edited by: snoopy ]</p>