They are making the Cell processor in Fishkill, people!
This is the one chip that everyone keeps overlooking: the IST (IBM-Sony-Toshiba) or Cell proc, the chip that is most likely to end up in the PlayStation 3, among other things.
I suspect that Apple will stick with Motorola's 74xx and 75xx family for another 12-24 months, then switch over to whatever IBM cooks up between POWER, PPC and Cell. Just a hunch; I have no real evidence.
[quote]The result will be consumer devices that are more powerful than IBM?s Deep Blue supercomputer, operate at low power and access the broadband Internet at ultra high speeds. Cell will be designed to deliver "teraflops" of processing power.<hr></blockquote>
[quote]This mix will allow IBM to create higher-performance chips for use in servers or communications equipment. Or the techniques can be used to greatly reduce a chip's power consumption for use in consumer electronics that run on batteries.<hr></blockquote>
If we wade through the vague marketing double-speak, we see a family of chips that can function in high-powered servers, low-powered consumer products, and ultra-blast capacity (I guess).
Can you think of any fruit-flavored company who might be interested?
This family of chips has it all: high-power, low power consumption, flexibility, PPC, latest tech, and a company that isn't hemorhaging money like certain cell-phone manufacturers I could mention). It's got a pretty damn large market (once you factor in PS3 and Apple and other consumer electronics, not to mention IBM themsleves in their budget server business) to boot.
So, this thread started with the question of what IBM is going to make in their new foundry. And I think I have the answser. Ladies and Gentlemen: I present to you the next generation of Mac processors: The Cell.
<strong>If we wade through the vague marketing double-speak, we see a family of chips that can function in high-powered servers, low-powered consumer products, and ultra-blast capacity (I guess).
Can you think of any fruit-flavored company who might be interested?
This family of chips has it all: high-power, low power consumption, flexibility, PPC, latest tech, and a company that isn't hemorhaging money like certain cell-phone manufacturers I could mention). It's got a pretty damn large market (once you factor in PS3 and Apple and other consumer electronics, not to mention IBM themsleves in their budget server business) to boot.</strong><hr></blockquote>
There's a very simple rule in life and that is if something sounds too good to be true (like a processor with the potential to fill any niche amazingly perfectly) it probably is.
There is a very simple rule in engineering and that is usually to succeed in one niche you often have to make trade offs in others. This has particularly been the case in the embedded and desktop markets.
Not to say "Cell" won't be a great design, as I'm sure it will be, but right now I would be waiting to see the fine print and white papers before getting your hopes up. I see a lot of marketing hype and very little substance so far.
One important quote to note:
[quote] The chip could end up inside the PlayStation 3, and elements of its design will be seen in future server chips from IBM. <hr></blockquote>
They don't say Cell is destined for servers only elements of it.
Rereading it I would say it sounds more like a notebook chip. Low-power with large computational abilities.
They don't say Cell is destined for servers only elements of it.
Rereading it I would say it sounds more like a notebook chip. Low-power with large computational abilities.
</strong><hr></blockquote>
You seem to be equating "Cell" to a chip, and I don't think that is correct. I think it is a technology which will be used to build a family of chips, and that multiple cores is inherent in the architecture. Many, many cores.
Is...the 'Cell' going to be a 'mass market' version of it?
It it's ready for 2004.
That's only a year ahead of the supposed 'G5'.
Will Apple use beefed up G4s until the 'Cell' arrives and use this?
The higher up the price bracket...the more 'cells' you can bung into your 'beast'?
I'm kinda wondering what implications the 'Cell' project has for Apple in light of the current difficulties with PPC...the analysts comments on Apple go Intel (in four years time I don't think so...)
Is 'Cell' going to be a new 'AIM' alliance against the Wintel monopoly?
M$ under attack from Linux/X?
Intel from Apple/IBM/Toshiba/Sony?
Is it true that Intel are pinning their hopes on Itanic 2 and don't have anywhere to go if OEMs don't want to pick it up?
Is...the 'Cell' going to be a 'mass market' version of it?
It it's ready for 2004.
That's only a year ahead of the supposed 'G5'.
Will Apple use beefed up G4s until the 'Cell' arrives and use this?
The higher up the price bracket...the more 'cells' you can bung into your 'beast'?
I'm kinda wondering what implications the 'Cell' project has for Apple in light of the current difficulties with PPC...the analysts comments on Apple go Intel (in four years time I don't think so...)
Is 'Cell' going to be a new 'AIM' alliance against the Wintel monopoly?
M$ under attack from Linux/X?
Intel from Apple/IBM/Toshiba/Sony?
Is it true that Intel are pinning their hopes on Itanic 2 and don't have anywhere to go if OEMs don't want to pick it up?
Anybody care to speculate...?
Lemon Bon Bon</strong><hr></blockquote>
Read the article. Most of these questions are answered in the article.
"The new multimedia processor, touted as a "supercomputer on a chip," is well on the way to completion, IBM says. The chip could end up inside the PlayStation 3 game console, and elements of its design are expected in future server chips from IBM. "
This might be the next CPU-architecture inside a mac..
Yeah, I read the article. My 'aussie' question style was inviting comment ie what do guys on these boards think it means for the Mac..?
I can barely imagine a hundred fold fpu performance. Real time 3D previewing? Real time radiosity previews?
Gurgle.
It seems to back up the IBM vision of the future...dual, multicore cpus and bandwidth over Intel's Mhz is all approach.
Intel could be in for a rough ride...
Maybe IBM's recent history of cpu development will see all their box of tricks in the 'Cell'. But instead of being stuck in a lab' and us lot going 'ooh...clever...but...' it will be commercially available to a mass market.
They're designing an OS from the ground up for this. Shame Apple can't get involved and license a special version of 'X' for it!
<strong>Shame Apple can't get involved and license a special version of 'X' for it!</strong><hr></blockquote>
I have to wonder why you say this? We all know how hush-hush Apple, IBM, and Moto are about the future of Mac processors. If this thing is PowerPC based (and the various articles I have read seem to imply this) then Apple should be in a position to take advantage of it. Imagine being able to buy 32, 64, 128, and 256 processor Macs? Heh...
Would we even need 256 cpu Macs? Consider what just 8 1Ghz PPC CPUs could do, especially linked with similarly powerful gfx hardware. Actually it would be more interesting to know what they couldn't do.
<strong>Would we even need 256 cpu Macs? Consider what just 8 1Ghz PPC CPUs could do, especially linked with similarly powerful gfx hardware. Actually it would be more interesting to know what they couldn't do.</strong><hr></blockquote>
<strong>Would we even need 256 cpu Macs? Consider what just 8 1Ghz PPC CPUs could do, especially linked with similarly powerful gfx hardware. Actually it would be more interesting to know what they couldn't do.</strong><hr></blockquote>
Im running genetic algorithms on a fairly small cluster ( 48 * 1.4 Ghz AMD). For complex problems it could take days&weeks to finish calculations on a cluster like that(!!). With more powerfull hardware people are just going to solve even harder GA-problems. There isnt going to be a "enough calculation-power" in a field like that, and thats true for other areas too. Like some problems ( Admitedly not something Average Joe cares about) in bioinformatic, data mining, artificial intelligence..
I'm thinking about the penetration a version of 'X' on the PS 3 could have for Apple.
Sure. It would be a PS3 games machine...but it would also be a Mac...could be a powerful combination.
It would sure boost the 'mindshare' argument for Macs.
People they would not otherwise reach with Macs. And as with Moki's x86 argument...or getting 'X' licensed for IBM's servers...just thinking about how Apple can increase their 'mindshare'/market presence...
Lemon Bon Bon
I wish people would talk about 'not needing power'.
Yeesh. I need more power. Lots of it. I want real time 3D previews. Real time volumetrics...real time radiosity. I want it at the highest resolutions. I may want to build a scene with millions of hi-res textured models...
I may want to animate it and have it rendered in real time...with lots of people milling about in it...
Power? Can't have too much of it if you're into 3D.
<strong>I wish people would talk about 'not needing power'.
Yeesh. I need more power. Lots of it. I want real time 3D previews. Real time volumetrics...real time radiosity. I want it at the highest resolutions. I may want to build a scene with millions of hi-res textured models...
</strong><hr></blockquote>
That's true for a chunk of people. The question is, how big a chunk of people? And how many of those people already have better boxes anyway?
If Apple did make a '256 CPU' box, it'd sell for $50,000+ and be for the Avid-defectors. Or a Shake behemoth. (Or both). But it probably wouldn't sell enough to cover cost at that price tag - too far into competition with some pretty capable workstations of all sorts.
What's a more reasonable size for Apple to actually release as the new flagship of the 'Pro' line?
8x pretty much seems like the absolute max there. Plenty to run the benchmarks and taunt Dell. Plenty to rationalize bundling Shake (a $10,000 app) into the $12,000 pricetag. Maybe 'Shake Lite' [iShake? Ow. Ow. Ow.] with a rebate on a full Shake.
That'd leave quads and/or duals down in the 'mere mortals can reach this high' arena. Fine by me
That's true for a chunk of people. The question is, how big a chunk of people? And how many of those people already have better boxes anyway?
If Apple did make a '256 CPU' box, it'd sell for $50,000+ and be for the Avid-defectors. Or a Shake behemoth. (Or both). But it probably wouldn't sell enough to cover cost at that price tag - too far into competition with some pretty capable workstations of all sorts.
What's a more reasonable size for Apple to actually release as the new flagship of the 'Pro' line?
8x pretty much seems like the absolute max there. Plenty to run the benchmarks and taunt Dell. Plenty to rationalize bundling Shake (a $10,000 app) into the $12,000 pricetag. Maybe 'Shake Lite' [iShake? Ow. Ow. Ow.] with a rebate on a full Shake.
That'd leave quads and/or duals down in the 'mere mortals can reach this high' arena. Fine by me </strong><hr></blockquote>
I think you're missing the point. The point is that IBM is talking about building game consoles with 32+ processors -- that means they are less than US$500. A $2000 Apple machine could have 256 processors. And this is in the 2005-6 time frame.
As for the "average Joe" not needing such computing power... hogwash. The main thing is what "killer apps" does this enable for the consumer? Sure the things you're doing nowadays don't require this kind of computing power, but if this kind of computing horsepower was commonly available what sort of software would come out to use it? I don't have an answer right now, but only because I haven't started thinking about it. There are all sorts of algorithms that we currently wouldn't even consider using, and these massively parallel machines could run them continuously in the background without the user noticing.
There is no such thing as "too much computing power", and this kind of power is going to arrive on your desk in the not-too-distant future (at least if IBM can deliver on the promise of its Cell architecture). Its going to take longer for the software developers to figure out what to do with it.
So I can get a 256-way PowerMac which is NOT the size of a refrigerator, doesn't require 3-phase power or a 2 ton AC unit...?!?
And for FAR less than the original US$50,000.00!?!
Great! Sign me up for a bakers dozen!
One to work on, the rest to render...!</strong><hr></blockquote>
Why limit yourself to working on just one?
Obviously this isn't going to show up in the next PowerMac, but it seems like a plausible direction for the future. And with the massive transistor counts that will be possible in tomorrow's chips (for no greater cost than today's chips) the wildly multi-core approach makes a fair bit of sense. Time-to-market with many less powerful cores will be better than with fewer more powerful cores. The IBM article quoted above (or was it in another thread?) also talks about working around failed cores and other redundancy features which effectively improve the production yield.
Comments
This is the one chip that everyone keeps overlooking: the IST (IBM-Sony-Toshiba) or Cell proc, the chip that is most likely to end up in the PlayStation 3, among other things.
I suspect that Apple will stick with Motorola's 74xx and 75xx family for another 12-24 months, then switch over to whatever IBM cooks up between POWER, PPC and Cell. Just a hunch; I have no real evidence.
However.....
13 months ago: <a href="http://www.ibm.com/news/2001/03/12.phtml"; target="_blank">http://www.ibm.com/news/2001/03/12.phtml</a>;
IBM to build new fab in Fishkill to make Cell.
[quote]The result will be consumer devices that are more powerful than IBM?s Deep Blue supercomputer, operate at low power and access the broadband Internet at ultra high speeds. Cell will be designed to deliver "teraflops" of processing power.<hr></blockquote>
This week: <a href="http://news.com.com/2100-1001-947358.html"; target="_blank">http://news.com.com/2100-1001-947358.html</a>;
IBM opens new fab in Fishkill.
[quote]This mix will allow IBM to create higher-performance chips for use in servers or communications equipment. Or the techniques can be used to greatly reduce a chip's power consumption for use in consumer electronics that run on batteries.<hr></blockquote>
If we wade through the vague marketing double-speak, we see a family of chips that can function in high-powered servers, low-powered consumer products, and ultra-blast capacity (I guess).
Can you think of any fruit-flavored company who might be interested?
This family of chips has it all: high-power, low power consumption, flexibility, PPC, latest tech, and a company that isn't hemorhaging money like certain cell-phone manufacturers I could mention). It's got a pretty damn large market (once you factor in PS3 and Apple and other consumer electronics, not to mention IBM themsleves in their budget server business) to boot.
So, this thread started with the question of what IBM is going to make in their new foundry. And I think I have the answser. Ladies and Gentlemen: I present to you the next generation of Mac processors: The Cell.
Jet
[ 08-06-2002: Message edited by: Jet Powers ]</p>
I like the sound of that...
Lemon Bon Bon
<strong>If we wade through the vague marketing double-speak, we see a family of chips that can function in high-powered servers, low-powered consumer products, and ultra-blast capacity (I guess).
Can you think of any fruit-flavored company who might be interested?
This family of chips has it all: high-power, low power consumption, flexibility, PPC, latest tech, and a company that isn't hemorhaging money like certain cell-phone manufacturers I could mention). It's got a pretty damn large market (once you factor in PS3 and Apple and other consumer electronics, not to mention IBM themsleves in their budget server business) to boot.</strong><hr></blockquote>
There's a very simple rule in life and that is if something sounds too good to be true (like a processor with the potential to fill any niche amazingly perfectly) it probably is.
There is a very simple rule in engineering and that is usually to succeed in one niche you often have to make trade offs in others. This has particularly been the case in the embedded and desktop markets.
Not to say "Cell" won't be a great design, as I'm sure it will be, but right now I would be waiting to see the fine print and white papers before getting your hopes up. I see a lot of marketing hype and very little substance so far.
One important quote to note:
[quote] The chip could end up inside the PlayStation 3, and elements of its design will be seen in future server chips from IBM. <hr></blockquote>
They don't say Cell is destined for servers only elements of it.
Rereading it I would say it sounds more like a notebook chip. Low-power with large computational abilities.
[ 08-06-2002: Message edited by: Telomar ]</p>
<strong>
They don't say Cell is destined for servers only elements of it.
Rereading it I would say it sounds more like a notebook chip. Low-power with large computational abilities.
</strong><hr></blockquote>
You seem to be equating "Cell" to a chip, and I don't think that is correct. I think it is a technology which will be used to build a family of chips, and that multiple cores is inherent in the architecture. Many, many cores.
IBM have a 'Cell' already? 'Power4'?
Is...the 'Cell' going to be a 'mass market' version of it?
It it's ready for 2004.
That's only a year ahead of the supposed 'G5'.
Will Apple use beefed up G4s until the 'Cell' arrives and use this?
The higher up the price bracket...the more 'cells' you can bung into your 'beast'?
I'm kinda wondering what implications the 'Cell' project has for Apple in light of the current difficulties with PPC...the analysts comments on Apple go Intel (in four years time I don't think so...)
Is 'Cell' going to be a new 'AIM' alliance against the Wintel monopoly?
M$ under attack from Linux/X?
Intel from Apple/IBM/Toshiba/Sony?
Is it true that Intel are pinning their hopes on Itanic 2 and don't have anywhere to go if OEMs don't want to pick it up?
Anybody care to speculate...?
Lemon Bon Bon
<strong>IS..(!) the 'Cell' PPC based?
IBM have a 'Cell' already? 'Power4'?
Is...the 'Cell' going to be a 'mass market' version of it?
It it's ready for 2004.
That's only a year ahead of the supposed 'G5'.
Will Apple use beefed up G4s until the 'Cell' arrives and use this?
The higher up the price bracket...the more 'cells' you can bung into your 'beast'?
I'm kinda wondering what implications the 'Cell' project has for Apple in light of the current difficulties with PPC...the analysts comments on Apple go Intel (in four years time I don't think so...)
Is 'Cell' going to be a new 'AIM' alliance against the Wintel monopoly?
M$ under attack from Linux/X?
Intel from Apple/IBM/Toshiba/Sony?
Is it true that Intel are pinning their hopes on Itanic 2 and don't have anywhere to go if OEMs don't want to pick it up?
Anybody care to speculate...?
Lemon Bon Bon</strong><hr></blockquote>
Read the article. Most of these questions are answered in the article.
<a href="http://news.com.com/2100-1001-948493.html?tag=fd_lede"; target="_blank">http://news.com.com/2100-1001-948493.html?tag=fd_lede</a>;
"The new multimedia processor, touted as a "supercomputer on a chip," is well on the way to completion, IBM says. The chip could end up inside the PlayStation 3 game console, and elements of its design are expected in future server chips from IBM. "
This might be the next CPU-architecture inside a mac..
I can barely imagine a hundred fold fpu performance. Real time 3D previewing? Real time radiosity previews?
Gurgle.
It seems to back up the IBM vision of the future...dual, multicore cpus and bandwidth over Intel's Mhz is all approach.
Intel could be in for a rough ride...
Maybe IBM's recent history of cpu development will see all their box of tricks in the 'Cell'. But instead of being stuck in a lab' and us lot going 'ooh...clever...but...' it will be commercially available to a mass market.
They're designing an OS from the ground up for this. Shame Apple can't get involved and license a special version of 'X' for it!
Lemon Bon Bon
[ 08-06-2002: Message edited by: Lemon Bon Bon ]</p>
<strong>Shame Apple can't get involved and license a special version of 'X' for it!</strong><hr></blockquote>
I have to wonder why you say this? We all know how hush-hush Apple, IBM, and Moto are about the future of Mac processors. If this thing is PowerPC based (and the various articles I have read seem to imply this) then Apple should be in a position to take advantage of it. Imagine being able to buy 32, 64, 128, and 256 processor Macs? Heh...
<strong>Would we even need 256 cpu Macs? Consider what just 8 1Ghz PPC CPUs could do, especially linked with similarly powerful gfx hardware. Actually it would be more interesting to know what they couldn't do.</strong><hr></blockquote>
Would we ever need more than 640K of RAM?
I can see Governments not wanting 256 way Cell systems to exist!
<strong>Would we even need 256 cpu Macs? Consider what just 8 1Ghz PPC CPUs could do, especially linked with similarly powerful gfx hardware. Actually it would be more interesting to know what they couldn't do.</strong><hr></blockquote>
Im running genetic algorithms on a fairly small cluster ( 48 * 1.4 Ghz AMD). For complex problems it could take days&weeks to finish calculations on a cluster like that(!!). With more powerfull hardware people are just going to solve even harder GA-problems. There isnt going to be a "enough calculation-power" in a field like that, and thats true for other areas too. Like some problems ( Admitedly not something Average Joe cares about) in bioinformatic, data mining, artificial intelligence..
[ 08-06-2002: Message edited by: blabla ]</p>
Sure. It would be a PS3 games machine...but it would also be a Mac...could be a powerful combination.
It would sure boost the 'mindshare' argument for Macs.
People they would not otherwise reach with Macs. And as with Moki's x86 argument...or getting 'X' licensed for IBM's servers...just thinking about how Apple can increase their 'mindshare'/market presence...
Lemon Bon Bon
I wish people would talk about 'not needing power'.
Yeesh. I need more power. Lots of it. I want real time 3D previews. Real time volumetrics...real time radiosity. I want it at the highest resolutions. I may want to build a scene with millions of hi-res textured models...
I may want to animate it and have it rendered in real time...with lots of people milling about in it...
Power? Can't have too much of it if you're into 3D.
[ 08-06-2002: Message edited by: Lemon Bon Bon ]</p>
<strong>I wish people would talk about 'not needing power'.
Yeesh. I need more power. Lots of it. I want real time 3D previews. Real time volumetrics...real time radiosity. I want it at the highest resolutions. I may want to build a scene with millions of hi-res textured models...
</strong><hr></blockquote>
That's true for a chunk of people. The question is, how big a chunk of people? And how many of those people already have better boxes anyway?
If Apple did make a '256 CPU' box, it'd sell for $50,000+ and be for the Avid-defectors. Or a Shake behemoth. (Or both). But it probably wouldn't sell enough to cover cost at that price tag - too far into competition with some pretty capable workstations of all sorts.
What's a more reasonable size for Apple to actually release as the new flagship of the 'Pro' line?
8x pretty much seems like the absolute max there. Plenty to run the benchmarks and taunt Dell. Plenty to rationalize bundling Shake (a $10,000 app) into the $12,000 pricetag. Maybe 'Shake Lite' [iShake? Ow. Ow. Ow.] with a rebate on a full Shake.
That'd leave quads and/or duals down in the 'mere mortals can reach this high' arena. Fine by me
I would get one, one way or another...
Log in and work from a 'dumb terminal', like when working on the Onyx...
And render while you work...!
<strong>
That's true for a chunk of people. The question is, how big a chunk of people? And how many of those people already have better boxes anyway?
If Apple did make a '256 CPU' box, it'd sell for $50,000+ and be for the Avid-defectors. Or a Shake behemoth. (Or both). But it probably wouldn't sell enough to cover cost at that price tag - too far into competition with some pretty capable workstations of all sorts.
What's a more reasonable size for Apple to actually release as the new flagship of the 'Pro' line?
8x pretty much seems like the absolute max there. Plenty to run the benchmarks and taunt Dell. Plenty to rationalize bundling Shake (a $10,000 app) into the $12,000 pricetag. Maybe 'Shake Lite' [iShake? Ow. Ow. Ow.] with a rebate on a full Shake.
That'd leave quads and/or duals down in the 'mere mortals can reach this high' arena. Fine by me
I think you're missing the point. The point is that IBM is talking about building game consoles with 32+ processors -- that means they are less than US$500. A $2000 Apple machine could have 256 processors. And this is in the 2005-6 time frame.
As for the "average Joe" not needing such computing power... hogwash. The main thing is what "killer apps" does this enable for the consumer? Sure the things you're doing nowadays don't require this kind of computing power, but if this kind of computing horsepower was commonly available what sort of software would come out to use it? I don't have an answer right now, but only because I haven't started thinking about it. There are all sorts of algorithms that we currently wouldn't even consider using, and these massively parallel machines could run them continuously in the background without the user noticing.
There is no such thing as "too much computing power", and this kind of power is going to arrive on your desk in the not-too-distant future (at least if IBM can deliver on the promise of its Cell architecture). Its going to take longer for the software developers to figure out what to do with it.
So I can get a 256-way PowerMac which is NOT the size of a refrigerator, doesn't require 3-phase power or a 2 ton AC unit...?!?
And for FAR less than the original US$50,000.00!?!
Great! Sign me up for a bakers dozen!
One to work on, the rest to render...!
<strong>WOW!!!
So I can get a 256-way PowerMac which is NOT the size of a refrigerator, doesn't require 3-phase power or a 2 ton AC unit...?!?
And for FAR less than the original US$50,000.00!?!
Great! Sign me up for a bakers dozen!
One to work on, the rest to render...!</strong><hr></blockquote>
Why limit yourself to working on just one?
Obviously this isn't going to show up in the next PowerMac, but it seems like a plausible direction for the future. And with the massive transistor counts that will be possible in tomorrow's chips (for no greater cost than today's chips) the wildly multi-core approach makes a fair bit of sense. Time-to-market with many less powerful cores will be better than with fewer more powerful cores. The IBM article quoted above (or was it in another thread?) also talks about working around failed cores and other redundancy features which effectively improve the production yield.