Don't forget the <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> at the end of that post, though.
I don't even want to think about how hot one of those puppies would run, let alone two. They would be venturing up into P4 and Athlon territory.
Also, apparently the leap from .18μ to .13μ is a big one technically - or from .15μ to .13μ - and the subsequent shrink to .09μ is comparatively easy. So what I'm talking about is essentially a limited run of the 970 (the 969?) specifically tweaked for the .18μ -or, for a larger than .13μ - process, concurrent with the announced 970.
Given: that it would require its own engineering effort; that it would be physically large, that it would have a very short shelf life, that it would be hard pressed to clock much higher than the current G4; that adapting the design to be fabbed on .13μ would be a nontrivial task (doable, though, especially if they'd prepared for it); I consider this fairly unlikely. Large size + limited quantity + high heat = high cost, especially from the most expensive foundry in the world.
Don't forget the <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> at the end of that post, though.
I don't even want to think about how hot one of those puppies would run, let alone two. They would be venturing up into P4 and Athlon territory.
Also, apparently the leap from .18μ to .13μ is a big one technically - or from .15μ to .13μ - and the subsequent shrink to .09μ is comparatively easy. So what I'm talking about is essentially a limited run of the 970 (the 969?) specifically tweaked for the .18μ -or, for a larger than .13μ - process, concurrent with the announced 970.
Given: that it would require its own engineering effort; that it would be physically large, that it would have a very short shelf life, that it would be hard pressed to clock much higher than the current G4; that adapting the design to be fabbed on .13μ would be a nontrivial task (doable, though, especially if they'd prepared for it); I consider this fairly unlikely. Large size + limited quantity + high heat = high cost, especially from the most expensive foundry in the world.
</strong><hr></blockquote>
By applying Occam's Razor to this situation it would seem that the most logical conclusion that we will see a 0.13μ 970 in the August-October timeframe. Is everybody so desperately bored that they have to sit around speculating on this over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over, ad infinitum ad nauseum...?
Lets at least speculate about interesting things while we wait.
<strong>Have a little faith. We are no longer dealing with the likes of Moto. </strong><hr></blockquote>
As of now we're still bound to them. But as soon as the complete line of Apple's computers have made the (real) "switch" to IBM's 970 we may define ourselves as being grown up. It's a painful slow bandwagen we're on right now heading west, and I do hope things will change in this regard.
Given Apple needs to release a 970 based PowerMac as soon as possible, but not repeat the mistakes of the G4 introduction (lack of chips, down grading of processor speeds etc), I guess the Autumn.
I can't see them being released before that, given Steve Jobs announced a short time ago it's the year of the notebook. An Autumn introduction would have quantities in place for the year of the desktop, next year.
Motorola's road map includes a G5 built on 0.13-micron technology, using the relatively new RapidIO bus architecture for connecting CPUs or other I/O components on the motherboard. Both 32-bit and 64-bit versions will operate at 1.2 to 2 GHz. The estimated release date is June or July.
Interestingly, it is rumored that Apple may use the recently disclosed IBM PowerPC 970 architecture, a scaled-down version of IBM's 64-bit Power4 architecture. The 970 will be built on a 0.13-micron process, use SOI technology, run both 32-bit and 64-bit code, and start at around 1.4 GHz with a 512K L2 cache. The chip will ship in quantity in the second half of 2003. Apple is not saying whether it will use IBM's or Motorola's technologies.
For the most part it's good reading, also found a processor roadmap.
Only hope, whatever Apple chooses to do, they do it quick.
heh, looking at that roadmap, Apple's not going to find themselves in much better of a situation come next year -- they'll merely be "on par" with the other Wintel machines it would seem (if the roadmap's any good). But I don't see, well, I hope that Steve doesn't accept second best again. I'm sick and tired of having to settle...Macs should have the hardware to glorify the already brilliant software (that's another myth out there from Wintelon people --- "there's no games for Mac..." blah, blah, blah -- if you're that idle then you don't need the Mac anyway).
To answer the thing about this being "boring" speculation - yep. But oh, well. We need something soon - and hearing that the next desktop may not appear til this time NEXT year is a bit disappointing. <img src="embarrassed.gif" border="0">
By applying Occam's Razor to this situation it would seem that the most logical conclusion that we will see a 0.13μ 970 in the August-October timeframe.</strong><hr></blockquote>
I really hate having to explain jokes, and I really hate it when I have to explain the explanation.
But yes, that was my point in a nutshell.
[quote]<strong>Lets at least speculate about interesting things while we wait.</strong><hr></blockquote>
<strong>heh, looking at that roadmap, Apple's not going to find themselves in much better of a situation come next year -- they'll merely be "on par" with the other Wintel machines it would seem (if the roadmap's any good).</strong><hr></blockquote>
Realistically "on par" is about as good as we can really expect. Apple doesn't have some source of magic pixie dust that lets them build supercomputers. "on par" is a slippery term, however, and frankly if on paper the Apple machines are rated the same speed as the contemporary PCs then its a pretty safe bet that in practice they'll be faster. And certainly much nicer to use. And given the current relative performance, "on par" is going to be a tremendous improvement.
Another thing to note about the roadmap is that Intel and AMD publish theirs, Motorola and IBM don't (at least not in any formal sense). The 2004/2005 columns are blank for the PowerPC guys, but that doesn't mean there aren't products there. We know the POWER5 and POWER6 are coming, so its a pretty good bet that we will see the 980 and 990, and they will be at least "on par" with Intel's deliveries.
Realistically "on par" is about as good as we can really expect. </strong><hr></blockquote>
Certainly in the desktop market that has become use to a single fast processor that might be the case and I wouldn't be surprised if the 970 is a touch behind the PIV of the time. In fact I'd be amazed if it wasn't considering Intel will have moved to 90 nm production.
Where I'd expect the PPC 970 to be a very nice chip is in 2 way, 4 way or 8 way solutions against the likes of Xeons or Opterons. I rather expect IBM is planning it that way too it is just a question of cost now. Personally I think that would be the best argument for why Apple should finally release some much higher end (and higher margin) workstations or servers using 4 or even 8 processors.
<strong>And given the current relative performance, "on par" is going to be a tremendous improvement.</strong><hr></blockquote>
I'm still obsessing over the floating point improvements. A 1GHz G4 scores SpecFP=187, a 1.8 GHz ppc 970 is conservatively estimated at 1051. I don't see that as 'on par' inprovement for the year
<strong>I'm still obsessing over the floating point improvements. A 1GHz G4 scores SpecFP=187, a 1.8 GHz ppc 970 is conservatively estimated at 1051. I don't see that as 'on par' inprovement for the year </strong><hr></blockquote>
The 2.8 GHz P4 posts just under 1000 SPECfp, and it will be > 3 GHz by the time the 970 arrives. That seems "on par" to me. It will kick the snot out of the G4 though.
sorry I ever uttered the dreaded term...I don't even realistically need the 970. There, I said it. I still run a G3/400 laptop for god's sake; I don't know why I really care that much about it -- getting a TiBook since the prices will be dropping will be a good upgrade for me --'ve never used a G4 in full practice with apps that are written to take advantage of it. Apple must move ahead, agreed. They're choosing to move ahead in the area that's their strength right now - notebooks, and they're listening for our input (or so it would seem with the Al-Book's improvements).
Perhaps next year at this time we will have seen the release of the second revision of 970-based Powermacs. My thinking is that the PowerBook may stay G4-based like *my* PowerBook was for quite some time until the Tis came around in Jan. '01 (was almost two years with no G4 laptop).
[quote]fred_lj: wouldn't releasing the 970 in February undermine Jobs' whole marketing strategy for the year (or at least what seems his marketing strategy -- "year of the notebook")?<hr></blockquote>
I think the marketing strategy has devolved more to "don't get slaughtered like arthritic, myopic old mutton."
They'll put out faster machines as soon as they can possibly do so.
<strong>That seems "on par" to me. It will kick the snot out of the G4 though.</strong><hr></blockquote>
Right, the _improvement_ will be a lot better than "on par", although the chip is expected to end up "on par" with its contemporaries. That's plenty for me. Well, that and low power dissipation + designed for SMP
Comments
<strong>
Great observation.</strong><hr></blockquote>
Don't forget the <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> at the end of that post, though.
I don't even want to think about how hot one of those puppies would run, let alone two. They would be venturing up into P4 and Athlon territory.
Also, apparently the leap from .18μ to .13μ is a big one technically - or from .15μ to .13μ - and the subsequent shrink to .09μ is comparatively easy. So what I'm talking about is essentially a limited run of the 970 (the 969?) specifically tweaked for the .18μ -or, for a larger than .13μ - process, concurrent with the announced 970.
Given: that it would require its own engineering effort; that it would be physically large, that it would have a very short shelf life, that it would be hard pressed to clock much higher than the current G4; that adapting the design to be fabbed on .13μ would be a nontrivial task (doable, though, especially if they'd prepared for it); I consider this fairly unlikely. Large size + limited quantity + high heat = high cost, especially from the most expensive foundry in the world.
[ 01-24-2003: Message edited by: Amorph ]</p>
<strong>
Don't forget the <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> at the end of that post, though.
I don't even want to think about how hot one of those puppies would run, let alone two. They would be venturing up into P4 and Athlon territory.
Also, apparently the leap from .18μ to .13μ is a big one technically - or from .15μ to .13μ - and the subsequent shrink to .09μ is comparatively easy. So what I'm talking about is essentially a limited run of the 970 (the 969?) specifically tweaked for the .18μ -or, for a larger than .13μ - process, concurrent with the announced 970.
Given: that it would require its own engineering effort; that it would be physically large, that it would have a very short shelf life, that it would be hard pressed to clock much higher than the current G4; that adapting the design to be fabbed on .13μ would be a nontrivial task (doable, though, especially if they'd prepared for it); I consider this fairly unlikely. Large size + limited quantity + high heat = high cost, especially from the most expensive foundry in the world.
</strong><hr></blockquote>
By applying Occam's Razor to this situation it would seem that the most logical conclusion that we will see a 0.13μ 970 in the August-October timeframe. Is everybody so desperately bored that they have to sit around speculating on this over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over and over, ad infinitum ad nauseum...?
Lets at least speculate about interesting things while we wait.
<strong>Have a little faith. We are no longer dealing with the likes of Moto. </strong><hr></blockquote>
As of now we're still bound to them. But as soon as the complete line of Apple's computers have made the (real) "switch" to IBM's 970 we may define ourselves as being grown up. It's a painful slow bandwagen we're on right now heading west, and I do hope things will change in this regard.
I can't see them being released before that, given Steve Jobs announced a short time ago it's the year of the notebook. An Autumn introduction would have quantities in place for the year of the desktop, next year.
[ 01-25-2003: Message edited by: RodUK ]</p>
Just found this <a href="http://www.pcmag.com/article2/0,4149,806465,00.asp" target="_blank">PC Mag</a>
Motorola's road map includes a G5 built on 0.13-micron technology, using the relatively new RapidIO bus architecture for connecting CPUs or other I/O components on the motherboard. Both 32-bit and 64-bit versions will operate at 1.2 to 2 GHz. The estimated release date is June or July.
Interestingly, it is rumored that Apple may use the recently disclosed IBM PowerPC 970 architecture, a scaled-down version of IBM's 64-bit Power4 architecture. The 970 will be built on a 0.13-micron process, use SOI technology, run both 32-bit and 64-bit code, and start at around 1.4 GHz with a 512K L2 cache. The chip will ship in quantity in the second half of 2003. Apple is not saying whether it will use IBM's or Motorola's technologies.
For the most part it's good reading, also found a processor roadmap.
Only hope, whatever Apple chooses to do, they do it quick.
[ 01-30-2003: Message edited by: iCom ]</p>
To answer the thing about this being "boring" speculation - yep. But oh, well. We need something soon - and hearing that the next desktop may not appear til this time NEXT year is a bit disappointing. <img src="embarrassed.gif" border="0">
<strong>
Don't forget the <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> at the end of that post, though.
</strong><hr></blockquote>
I saw it. I guess I'm getting desperate for any piece of information that could point to ppc970 powermacs being released in a few weeks.
<strong>
By applying Occam's Razor to this situation it would seem that the most logical conclusion that we will see a 0.13μ 970 in the August-October timeframe.</strong><hr></blockquote>
I really hate having to explain jokes, and I really hate it when I have to explain the explanation.
But yes, that was my point in a nutshell.
[quote]<strong>Lets at least speculate about interesting things while we wait.</strong><hr></blockquote>
So, uhhh, come to Future Hardware often?
<strong>heh, looking at that roadmap, Apple's not going to find themselves in much better of a situation come next year -- they'll merely be "on par" with the other Wintel machines it would seem (if the roadmap's any good).</strong><hr></blockquote>
Realistically "on par" is about as good as we can really expect. Apple doesn't have some source of magic pixie dust that lets them build supercomputers. "on par" is a slippery term, however, and frankly if on paper the Apple machines are rated the same speed as the contemporary PCs then its a pretty safe bet that in practice they'll be faster. And certainly much nicer to use. And given the current relative performance, "on par" is going to be a tremendous improvement.
Another thing to note about the roadmap is that Intel and AMD publish theirs, Motorola and IBM don't (at least not in any formal sense). The 2004/2005 columns are blank for the PowerPC guys, but that doesn't mean there aren't products there. We know the POWER5 and POWER6 are coming, so its a pretty good bet that we will see the 980 and 990, and they will be at least "on par" with Intel's deliveries.
<strong>
I really hate having to explain jokes, and I really hate it when I have to explain the explanation.
But yes, that was my point in a nutshell.
</strong><hr></blockquote>
Sorry, I just clicked on your message at random in order to inject my comment. I think you and I are more or less in agreement.
[quote]<strong>
So, uhhh, come to Future Hardware often?
</strong><hr></blockquote>
Far too often, I just don't feel like actually doing any work right now...
<strong>
Realistically "on par" is about as good as we can really expect. </strong><hr></blockquote>
Certainly in the desktop market that has become use to a single fast processor that might be the case and I wouldn't be surprised if the 970 is a touch behind the PIV of the time. In fact I'd be amazed if it wasn't considering Intel will have moved to 90 nm production.
Where I'd expect the PPC 970 to be a very nice chip is in 2 way, 4 way or 8 way solutions against the likes of Xeons or Opterons. I rather expect IBM is planning it that way too it is just a question of cost now. Personally I think that would be the best argument for why Apple should finally release some much higher end (and higher margin) workstations or servers using 4 or even 8 processors.
[ 01-26-2003: Message edited by: Telomar ]</p>
<strong>And given the current relative performance, "on par" is going to be a tremendous improvement.</strong><hr></blockquote>
I'm still obsessing over the floating point improvements. A 1GHz G4 scores SpecFP=187, a 1.8 GHz ppc 970 is conservatively estimated at 1051. I don't see that as 'on par' inprovement for the year
<strong>I'm still obsessing over the floating point improvements. A 1GHz G4 scores SpecFP=187, a 1.8 GHz ppc 970 is conservatively estimated at 1051. I don't see that as 'on par' inprovement for the year </strong><hr></blockquote>
The 2.8 GHz P4 posts just under 1000 SPECfp, and it will be > 3 GHz by the time the 970 arrives. That seems "on par" to me. It will kick the snot out of the G4 though.
Perhaps next year at this time we will have seen the release of the second revision of 970-based Powermacs. My thinking is that the PowerBook may stay G4-based like *my* PowerBook was for quite some time until the Tis came around in Jan. '01 (was almost two years with no G4 laptop).
I think the marketing strategy has devolved more to "don't get slaughtered like arthritic, myopic old mutton."
They'll put out faster machines as soon as they can possibly do so.
<strong>That seems "on par" to me. It will kick the snot out of the G4 though.</strong><hr></blockquote>
Right, the _improvement_ will be a lot better than "on par", although the chip is expected to end up "on par" with its contemporaries. That's plenty for me. Well, that and low power dissipation + designed for SMP