PDA

View Full Version : STI Cell unveiling today


Renderman_XSI
02-07-2005, 09:39 PM
"Information about IBM and Sony's Cell processor is due to be revealed later today at the International Solid State Circuits Conference in San Francisco, with the unveiling of PlayStation 3 now expected to follow next month.

IBM will be outlining the architecture of the chip, which is aimed both at consumer applications ranging from televisions and media centres to games consoles and at high end "supercomputer" applications, in a presentation at the conference.

Cell is, of course, set to be used in the PlayStation 3 console, which gives the chip an automatic market for tens of millions of units - and it's thought that the PS3 will feature four or even eight of the processors, which are designed to work efficiently together in parallel....."


http://www.gamesindustry.biz/content_page.php?aid=6677


http://news.com.com/PlayStation+3+chip+has+split+personality/2100-1043_3-5566340.html

http://www.electronicsweekly.co.uk/ImageLibrary/GetImage.asp?liAssetID=1006 (picture)


http://home.businesswire.com/portal/site/home/index.jsp?epi-content=PUBLIC_IMAGE_VIEW&newsId=20050207005340&newsLang=en&contentItemId=1216538


http://img226.exs.cx/img226/4315/cell3jb.jpg
http://img204.exs.cx/img204/1457/cell21iw.jpg
http://img204.exs.cx/img204/7944/cell9vw.jpg

dmonk
02-07-2005, 09:43 PM
Very exciting! I hope they can be produced cheap.

richcz3
02-07-2005, 11:10 PM
It looks impressive on paper but so was the pre RISC hype when IBM, Motorola, and Apple teamed up in the 90's. With RISC archtecture the x86 days were numbered. Well The x86 architecture is stronger than ever. Right now these Cell PR Machines are throwing allot of unsubstantiated specs around. They say multiple OS support but don't mention any OS support beyond Linux.

Some very creative PR spinning here
"It also will not carry the same technical baggage that has made most of today's computers compatible with older PCs. That architectural divergence will challenge the current dominant paradigm of computing that Microsoft Corp. and Intel Corp. have fostered."

In other words scrap backward compatability and associated apps. I don't know of any company, software or hardware willing to put SONY at the helm of their computings future.

Article Here (http://story.news.yahoo.com/news?tmpl=story&cid=528&ncid=528&e=1&u=/ap/20050207/ap_on_hi_te/cell_processor)

.

allseeingi
02-07-2005, 11:23 PM
Mmm...chocolate drops.

Ghlughlughlugh...

- allseeingi

Kabab
02-08-2005, 12:36 AM
It looks impressive on paper but so was the pre RISC hype when IBM, Motorola, and Apple teamed up in the 90's. With RISC archtecture the x86 days were numbered. Well The x86 architecture is stronger than ever. Right now these Cell PR Machines are throwing allot of unsubstantiated specs around. They say multiple OS support but don't mention any OS support beyond Linux.


Thoes machines where very succesful in the engineering/automotive/aerospace sectors.

richcz3
02-08-2005, 02:20 AM
Thoes machines where very succesful in the engineering/automotive/aerospace sectors.
"With RISC archtecture the x86 days were numbered"
There is no doubt that every chip finds its market. My point is the grandiose claims of replacing the x86 architecture that have never come to pass.

SOPLAND
02-08-2005, 03:28 AM
"With RISC archtecture the x86 days were numbered"
There is no doubt that every chip finds its market. My point is the grandiose claims of replacing the x86 architecture that have never come to pass.

Modern x86 processors borrow quite a bit from RISC designs, and if MS and Intel didn't have so much invested in x86, we'd likely all be using RISC processors now. The overwhelming majority of mass produced processors nowadays are RISC. The only market where CISC based processors are used is in any meanigful quantity is in desktop computers, specificly Windows powered computers. RISC dominates the server market, consoles, thin clients, and embedded systems by a pretty wide margin. Your Gameboy, Gamecube, PS2, phone, pda, DVD player, cable/satellite box, car, camera, etc are all probably using a RISC processor of some sort. If I were to ask you which company sells the most cpus per year, you likely say Intel. It's actually ARM. ARM makes embedded RISC processors and they sell about 3 times as many each year as Intel and AMD combined. That's just one company, there are many more making RISC processors for all markets.

Don't have any Illusions about x86 being a better architecture, it all comes down to money; MS and Intel's money. The desktop PC is not long for this world, and when it goes the way of the dinosaur, it will likely be the last gasp for non-RISC processors.

heavyness
02-08-2005, 03:38 AM
cool and exciting. only time will tell if it hits it market.




but, why does the chip look like a dresser from Ikea??

http://img226.exs.cx/img226/4315/cell3jb.jpg

Beamtracer
02-08-2005, 03:50 AM
Some very creative PR spinning here[/color][/b]
[b][i][font=arial][size=-1] "It also will not carry the same technical baggage that has made most of today's computers compatible with older PCs. That architectural divergence will challenge the current dominant paradigm of computing that Microsoft Corp. and Intel Corp. have fostered."

IBM's Cell architecture is a major major threat to Intel. Yes there's a lot of hype, but it's still the biggest jump in performance we've ever seen.

Why worry about whether it's Windows compatible? It's time to switch to a UNIX derived operating system like Linux.

Also, how is Microsoft going to sell any xbox2 consoles when the PlayStation3 with cell processor is just around the corner?

richcz3
02-08-2005, 03:52 AM
....Don't have any Illusions about x86 being a better architecture, it all comes down to money; MS and Intel's money. The desktop PC is not long for this world, and when it goes the way of the dinosaur, it will likely be the last gasp for non-RISC processors.

I'll paste my statement for the 2nd time.
"With RISC archtecture the x86 days were numbered"
Please don't read, assume, or have illusions about something I am not stating. The best counter to anything I've written is to support the purported demise of the x86 by the RISC architecture. Or better that "With the Cell architecture the x86 days are numbered"

Thank You,

SOPLAND
02-08-2005, 04:28 AM
How's this then... You said.

"Well The x86 architecture is stronger than ever."

Actually x86 is weaker than ever. It's in danger of being a fading memory in the next couple of years. Your first post seemed to be saying that RISC is dead and x86 is dominating the market. The opposite is actually true. x86 has lost all but one market, is struggling to hold on to this single market, and losing ground. Look at the state of the industry. They are struggling to get every last ounce of power out of x86 designs, so much so that they are resorting to multiple cores. A lot of people seem to think this is some sort of improvement, but I think it's Intel and AMD admitting defeat. They can't make these processors much faster so they are just going to throw two of them in a box whether your software can take advantage of them or not. If engineers are going to have to start writing parallel code, and consumers are going to have their arms twisted into completely replacing their current computers to run Windows 64, a lot of them are going to start looking around to see what else there is. This is Sony, Apple, and IBM's chance to step in a grab some market share. It's not really a question of whether these companies can topple Intel and MS, it's looking more like the question is how quickly can they do it. I don't think Sony is going to need to build a cell based PC to destroy Microsoft's market, they'll probably do it with the PS3.

cha0t1c1
02-08-2005, 04:40 AM
I don't think that us, the tiny unimportant people we are, have both the capability nor the understanding of what the future holds for either type of the architecture. if x86 is dying, why come up with dual core. you just can't generalize. both architecture types have their applications. if x86 wasn't competetive enough, or satisfactory, it would have lost its position long ago. that's my opinion anyway :shrug:

richcz3
02-08-2005, 05:11 AM
I realize there are sensitiveties to my post but I never stated that RISC was dead.
Nor did I even hint at it.

In the early mid 90's when IBM, Motorola, and Apple joined forces to produce their version of RISC chip Apple had a very respectable market share. Some state higher than 20% Do some searches to see how many businesses used Apple desktops as well as PC's. It was pretty respectable number.
Again, I am stating the demise (means death) of the x86 architecture due to this trios RISC chip. Well history has recorded Apples market share decline steadily since 2000. Motorolas failure to produce a speedy chip is partly to blame.

Apples US market share has dropped from 3.6% from 3.2% Read this link (http://www.macobserver.com/article/2004/10/29.6.shtml)
Companies that once had art departments fully stocked with Apple computers have either made the switch to PC's or have co-opted PC's and Macs.

People, lets not muddy the waters
I am not talking server systems, computers used at Los Alamos, or Earth modeling super computers that use arrays of RISC chips either. I am refering to computers that CGTalk artists use day in and day out. The x86 is doing better than ever. Show me the company(s) that has handed this purported x86 death knell reducing its market to less than 3.2% and I will gladly read that article with much enthusiasm. :)

PC Market Share 1975-2004 (http://www.pegasus3d.com/total_share.html)


Thanks

Beamtracer
02-08-2005, 05:30 AM
Apples US market share has dropped from 3.6% from 3.2% Read this link (http://www.macobserver.com/article/2004/10/29.6.shtml)
Companies that once had art departments fully stocked with Apple computers have either made the switch to PC's or have co-opted PC's and Macs.
Your Apple death-knell comment is factually incorrect. Apple is a niche player, and its niche is content creation, and it's doing very well in this niche.

Apple is increasing its share in content creation, as is Linux. Recent surveys are showing dramatic rises, with over a third of content creation companies and film/broadcast production houses using Macs as their main machines.

It's Microsoft Windows that is losing market share in this area, as post production houses prefer to migrate to UNIX derived operating systems (ie OS X and Linux).

Yes, there are more Windows boxes on the planet, but many of these are lying idle at night in insurance company buildings and accounting firms. Who cares what they use?

The only platforms we know are going Cell is Playstation3 and Linux. It would be a reasonable guess that Apple will follow.

Para
02-08-2005, 07:41 AM
IBM's Cell architecture is a major major threat to Intel. Yes there's a lot of hype, but it's still the biggest jump in performance we've ever seen.

Or biggest jump for a while anyway. What about AMD? ;)

Also, how is Microsoft going to sell any xbox2 consoles when the PlayStation3 with cell processor is just around the corner?

By putting consoles to stores and pack a bunch of nice games with 'em? The technology doesn't really matter anymore, there's more than enough power as is in systems nowadays to do stuff that's seems essentially impossible. Technology doesn't matter and it never did, it's all about what you can do with it. Besides Sony has a bad track record with PR, for example anyone remember Emotion Engine for PS2? It was supposed to be some sort of hardware-level special chip that makes characters in games automatically lifelike and unique to make players bond with them and cry when they die etc. That was the biggest thing in PS2 Sony's PR department was shouting for and when PS2 finally landed everyone just forgot the whole thing - Sony being among the first ones to forget.

Beamtracer
02-08-2005, 08:18 AM
What about AMD? ;)

It's possible Cell could also be a threat to AMD, however unlike Intel, AMD has a technology sharing arrangement with IBM. Who knows, that may allow AMD to get into Cell technology.

As for Intel, they've become a big monopoly anyway. Something that cuts them down to size will be a good thing for everyone.

Coliba
02-08-2005, 12:54 PM
This is very exciting news! The more I read about the Cell, the more I think this is just what us 3D freaks have been looking for!

But is it realistic to believe we will someday have access to this type of processor and be able to build a system with it like we do today? The processor is already built, what obstacles will take the longest time to overcome so it can become useful to us?

richcz3
02-08-2005, 03:06 PM
"Any new technology like this has two components," Kleynhans said. "It has the vision of what it could be because you need the big vision to sell it. Then there's the reality of how it's really going to be used, which generally several levels down the chain from there."
See LINK (http://apnews.excite.com/article/20050208/D8844QS00.html)

For now it's all PR Hype for the Cell. There's nothing wrong with PR hype, but blind faith has its problems though. Wait until it hits the market channels in masse and see which markets adopt it. Then the market will evaluate its threat potential.
If someone could post some real world benchmarks or demonstrations on the Cell please post links.

Does anyone remember Bitboys and their revolutionary 3D chip that was supposed to lay waste to the PC 3D card market in the late 90's. All PR hype of course. They have a niche market now in graphics for mobile phones and embedded devices. See link (http://www.bitboys.fi/index.php)
Not bad but nVidia and ATI are fairing quite well.

Beamtracer
There is no death knell for Apple. Who stated that? :shrug:
That trio that produced the RISC processor for Apple overstated its impact on the x86 market. Simple as that. It doesn't mean that the x86 is better it just means that the marketing hype in the 90's didn't pan out. We all know Motorola dropped the ball on RISC development and IBM picked it up. :)

Heck I use AMD64's and plan on getting WINXP64 when its released for all my systems. I am not fanatical or religious about the systems I use, I just use what has the most support. :thumbsup:



cheers

mustique
02-08-2005, 04:10 PM
It's possible Cell could also be a threat to AMD, however unlike Intel, AMD has a technology sharing arrangement with IBM. Who knows, that may allow AMD to get into Cell technology.

As for Intel, they've become a big monopoly anyway. Something that cuts them down to size will be a good thing for everyone.

Agree.

I only wish that the developments won't create new monopolies.
Hope the fresh competition will lead to getting Linux and MacOSX
embraced by the masses. (and we will do our renderings in real-time. :)

mummey
02-08-2005, 04:20 PM
Wow! All this processor trolling, and Thalaxis hasn't even posted yet! ;)

Para
02-08-2005, 04:29 PM
Does anyone remember Bitboys and their revolutionary 3D chip that was supposed to lay waste to the PC 3D card market in the late 90's. All PR hype of course. They have a niche market now in graphics for mobile phones and embedded devices. See link (http://www.bitboys.fi/index.php)
Not bad but nVidia and ATI are fairing quite well.

Wasn't hype. Those of us who have Pyramid3D/AXE reference cards are very happy of 'em. The main reason for the failure was Infinium Labs (Phantom, anyone?) and I already explained this to you last year :)

richcz3
02-08-2005, 05:48 PM
Wow! All this processor trolling, and Thalaxis hasn't even posted yet! ;)
LOL, and where is AMDMAX I mean MADMAX. Surely he has some x86 loving to support. :)

Here's my take on sites i've read so far.
IBM is selling its PC business to China. There is little chance that IBM will produce a Personal Computer for the masses based on the Cell. IBM will make powerful, scalable, affordable servers based on the Cell that run many\all flavors of LINUX. Big $$$ for IBM. That will hurt Intel and AMD server sales. That is the threat. Sorry No Cell PC, it will be a highly scalable server running various flavors of Linux.
There is that media component to the Cell. Hmmm.

The PS3 has had a price of $400-$500 floated about. Internet ready media centers is what the next gen consoles are being positioned for. (side note) Some people are unaware, but IBM is making XBOX 2 Processor as well. Go figure.

The selling point of the PS3 is its faster than a powerful PC Workstation at 1/3 the cost. But you won't be able to buy a Cell PC. That would conflict with PS3 sales. sorry.
It's genious marketing and I think they wil pull it off.

chadtheartist
02-08-2005, 06:09 PM
Sony isn't the only company working on the Cell. So is Toshiba, and IBM. I wouldn't be surprised if "someone" doesn't take the cell into other markets than the PS3 right off the bat.

It would be interesting to see what Apple decides to do if the Cell becomes a "desktop" processor. IBM already has had tons of problems with the G5, and I wonder how long they will keep at it to get it right? Especially considering that they have this new "Cell" technology sitting right in their labs.

rendermania
02-08-2005, 09:03 PM
I doubt Sony, IBM and Toshiba would pour billions of dollars into building the Cell processor and just use it for a goddamn game console. My guess is that Cell will power many Sony electronics products that need multimedia capabilities - TV's that can browse the web and access things like on-demand streaming video. I would also bet that the Cell will come in handy for HD based video products, because manipulating HD video in any way takes some horsepower.

For IBM the appeal is probably high-end servers and industrial graphics workstations. For Toshiba it may be multimedia laptops, portable devices etc etc.

If its fast and cheap to manufacture, don't underestimate how far this platform can go. Microsoft, AMD and Intel have brough very little innovation to the consumer market for some time. The 'Cell' branded CPU may be just the thing needed to spark demand for nextgen consumer electronics products.

Renderman_XSI
02-08-2005, 09:05 PM
The PS3 has had a price of $400-$500 floated about. Internet ready media centers is what the next gen consoles are being positioned for. (side note) Some people are unaware, but IBM is making XBOX 2 Processor as well. Go figure.

The selling point of the PS3 is its faster than a powerful PC Workstation at 1/3 the cost. But you won't be able to buy a Cell PC. That would conflict with PS3 sales. sorry.
It's genious marketing and I think they wil pull it off.

400-500 USD? thats new to me, even as a rumour. ;)(but the cost of living is always going up). Indeed IBM is working on the ex-BOX2 cpu and also on the new N5 cpu, but if your excepting Cell architecture in all three console..you'll be disaapointed. They got different teams of engineering working on them seperately.

"But you won't be able to buy a Cell PC. That would conflict with PS3 sales. sorry. " I have to totally disagree. for one, Sony has there own VAIO desktop/workstation line, theres no reason they couldnt introduce a serie of cell base PC in the future(given the "fact" that Cell can run any OS). Secondly, a cell base workstation wont be a conflict with the PS3 at all, more likely it would give you the best of both worlds, gaming(PC+PS3+PS2+PS1) and working( Maya,Modo, XSI,etc) at a higher price point than the average PS3 with no HDD.

So how is that a conflict of sales or interested? It would be a conflict of sales if the cell workstation costed more than the PS3 and did more things, but thats not how reality works. Also with the workstation there no "limited" by cost, so they can have as many Cell cpu on a motherboard as they see fit,of course at a higher workstation price point(thats a given).

BTW PS3, will be inveil next month before E3.

another article you should read: http://arstechnica.com/articles/paedia/cpu/cell-1.ars
if you got question, best to ask in their forum they got alot of cpu nerds there :).

http://pc.watch.impress.co.jp/docs/2005/0208/kaigaip016.jpg

http://pc.watch.impress.co.jp/docs/2005/0208/kaigaip028.jpg

http://pc.watch.impress.co.jp/docs/2005/0208/kaigaip039.jpg

http://pc.watch.impress.co.jp/docs/2005/0208/kaigaip051.jpg

allseeingi
02-08-2005, 11:31 PM
So if I've got this right, the Cell is scalable. So technically, where traditionally a console would have the same amount of 'horsepower' throughout it's lifespan even if there were cosmetic changes (e.g. PS2 --> Slim PS2), this time round Sony could release an upgraded 'faster' PS3 year on year as costs decrease. So a game could be released that works better on the newer ones and not so great on the older ones, or even doesn't work at all. Anyone think this is a possibility?

- allseeingi

Beamtracer
02-09-2005, 01:23 AM
IBM's Power processors generally have a different byte order to Intel processors. This could prevent Windows from easily being ported to it.

cha0t1c1
02-09-2005, 03:01 AM
I don't think that the predictions are correct. I've been following the project with cynicism, but you have to mind me because of the PR bull. http://www.ps3portal.com/?view=article&article=83
I don't think that they have proven anything. whenever Intel and AMD reveal, they show benchmarks. I'm not saying do such, but show practical and touchable proof.:wise::twisted:

Apoclypse
02-09-2005, 05:27 AM
IBM's Power processors generally have a different byte order to Intel processors. This could prevent Windows from easily being ported to it.

Your point being.

In fact this is exaclty what IBM may ahve been looking for. A way to supplant the x86 spec with their own, and thus getting back at MS for burning them like they did back in the day. Also alot of the patent issues currently present int terms of MS would totally disappear, so this just might gibe linux that chance they've been waitung for where they are the druvung force behind a market instead of playing catchup like they are doing now with the x86 market due to hradware support. (it would be interesting to see M$ try to release an OS om this platform. They would probably try to buy some unix patents and take out a unix kernel based os. I could see it now, MS Windex now you can have windows with a unix kernel that is as reliable as XP - get familiar with the phrase kernel panic)

MacRonin
02-09-2005, 07:20 AM
The IBM chips being used by Microsoft for the XBox2 are some PowerPC chips...

The boxes being sent out to developers for XBox2 are Apple PowerMac G5s...

Apple will continue to use IBM PowerPCs for their hardware lineups, eventually switching all products over to various G5s; some single CPU, some dual CPU; eventually all dual core...

But I see a future product utilizing an Apple specific variation of Cell, designed for hardcore DCC work...

Imagine piping optimized Maya/Shake/Final Cut Pro/RenderMan/etc. code thru an array of Cell units...

js33
02-09-2005, 08:08 AM
The Cell (Power 5) is what the G5 should have been. Maybe it will be in the future. Doesn't IBM make an 8 core Power 5 processor? Why not put that in a PC/Mac for the high end? Sure it will be expensive and require some power but who cares if you can have an 8 core processor on your desktop!

It's high time we had a high end Silicon graphics style machine for the desktop. I'm making the comparison of many years ago when SGI ruled the world of graphics/3D and all a Mac or PC could do was desktop publishing or spreadsheets.

Cheers,
JS

Beamtracer
02-09-2005, 08:28 AM
What about Xbox? Microsoft is about to release Xbox2 but it will only have a standard non-cell processor in it. Not long afterwards, Sony will release the Playstation 3 featuring the new cell processor.

The Playstation 3 will have a lot more processing power than Xbox 2. Why would anyone buy an Xbox2? The thing is already obsolete, and it hasn't even been released yet.

I can only assume that Sony's partnership with IBM and Toshiba allows them to lock Microsoft out of the Cell technology.

Thalaxis
02-09-2005, 02:30 PM
Modern x86 processors borrow quite a bit from RISC designs, and if MS and Intel didn't have so much invested in x86, we'd likely all be using RISC processors now.


It's not their investment that matters, they're just catering to the audience... it's all those billions of
dollars of other software that matters.


The overwhelming majority of mass produced processors nowadays are RISC. The only market where CISC based processors are used is in any meanigful quantity is in desktop computers, specificly Windows powered computers. RISC dominates the server market, consoles, thin clients, and embedded systems by a pretty wide margin


Except servers (x86, specifically Xeon right now, accounts for 80% of that market by unit), yes. The
only reason that the other 20% is so desirable is that it supports the $4000 price tags of the high
end processors -- try selling an Opteron 854 at $2000+ to your hardcore gamer friends, and see
how they react ;)


If I were to ask you which company sells the most cpus per year, you likely say Intel. It's actually ARM. ARM makes embedded RISC processors and they sell about 3 times as many each year as Intel and AMD combined. That's just one company, there are many more making RISC processors for all markets.


Ironically, ARM doesn't actually sell processors at all. They don't make them, they get their revenue
from licensing. Intel is one of the bigger vendors of ARM processors (XScale is an ARM).

And BTW, you're not even close with the 3x estimate... you need to find a considerably larger
number than that.


Don't have any Illusions about x86 being a better architecture, it all comes down to money; MS and Intel's money. The desktop PC is not long for this world, and when it goes the way of the dinosaur, it will likely be the last gasp for non-RISC processors.

What's relevant isn't the PC, it's the requirement for x86 compatibility.

Thalaxis
02-09-2005, 02:31 PM
The opposite is actually true. x86 has lost all but one market, is struggling to hold on to this single market, and losing ground.


Are you BLIND?

Thalaxis
02-09-2005, 02:34 PM
Wow! All this processor trolling, and Thalaxis hasn't even posted yet! ;)


Ah, the children showed up early, I see...

Thalaxis
02-09-2005, 02:36 PM
IBM already has had tons of problems with the G5, and I wonder how long they will keep at it to get it right?

IBM won't be the only one producing Cell processors, so that shouldn't be a worry. Toshiba does
volume a lot better than IBM.

Thalaxis
02-09-2005, 02:43 PM
The Cell (Power 5) is what the G5 should have been. Maybe it will be in the future. Doesn't IBM make an 8 core Power 5 processor? Why not put that in a PC/Mac for the high end? Sure it will be expensive and require some power but who cares if you can have an 8 core processor on your desktop!


That 8 core Power5 costs something like $40,000... and the system uses a 100 TERABYTE per second
interconnect and a 144 MB eDRAM L3... so that 8 core MCM + a system will set you back more quite
a bit more than a 2048 Athlon Beowulf cluster :)


It's high time we had a high end Silicon graphics style machine for the desktop. I'm making the comparison of many years ago when SGI ruled the world of graphics/3D and all a Mac or PC could do was desktop publishing or spreadsheets.


SGI might yet come through, they've hinted at workstation-ish products in the pipeline, the Prism being
the first. Not that I'd consider a 512 processor box a workstation, exactly :)

chadtheartist
02-09-2005, 02:46 PM
I wasn't saying that IBM's manufacturing was an issue. I was saying the design of the chip was an issue, which caused the manufacturing problems. I think the G5 has had problems from the start, and I'm curious as to how long it is going to take IBM to get it worked out.

Thalaxis
02-09-2005, 03:04 PM
I wasn't saying that IBM's manufacturing was an issue. I was saying the design of the chip was an issue, which caused the manufacturing problems. I think the G5 has had problems from the start, and I'm curious as to how long it is going to take IBM to get it worked out.

IBM's been blaming poor yields from their process for their semiconductor losses for a long time, and
both Xilinx and nVidia cited poor yields as their reason for taking several large fab contracts
elsewhere, so it's hard to say whether or not there was a flaw in the G5's design. My impression
was that the "flaw" was the system (POWER is designed VERY closely with its systems, which
are IBM's specialty), though it's possible that there is a speedpath issue with their altivec
implementation. I've seen speculation to that effect, but based on the thermal data that IBM
published for the 970FX, I think it's equally likely that they just ran into the same problems with
leakage as Intel, but haven't thrown as much money into dealing with them, because they're
too focussed on POWER5 to spare the resources.

However, my point was that since Toshiba and Sony are not strangers to high-volume semiconductor
manufacturing like IBM is, the CELL should be relatively immune to IBM's comparative lack of
experience in that regard.

That's what I'm hoping, at any rate.

chadtheartist
02-09-2005, 03:09 PM
I agree with that. I'm just hoping that Apple doesn't remain stubborn and stick with something that may not work. Especially if the Cell processor is as good as it is cracked up to be.

mummey
02-09-2005, 03:11 PM
Glad to see you finally arrived Thalaxis... ;)

imashination
02-09-2005, 03:17 PM
How's this then... You said.

"Well The x86 architecture is stronger than ever."

Actually x86 is weaker than ever. It's in danger of being a fading memory in the next couple of years. Your first post seemed to be saying that RISC is dead and x86 is dominating the market. The opposite is actually true. x86 has lost all but one market, is struggling to hold on to this single market, and losing ground. Look at the state of the industry. They are struggling to get every last ounce of power out of x86 designs,

Several years ago I was reading how x86 had reached the end of the line and that it was impossible to increase speeds due to physics problem x. I read this same story every year about how we've all reached the end and that a new 10x faster incompatible chip is the only way forward. Every year it turns out to be completely untrue.

You say they're 'resorting' to using multiple cores, well so what? the entire design of the cell system is to link together many cpu cores. You make it sound like sony doing it is a good thing but if intel/amd do it then its a terrible last resort of a dying technology.

Thalaxis
02-09-2005, 03:24 PM
I agree with that. I'm just hoping that Apple doesn't remain stubborn and stick with something that may not work. Especially if the Cell processor is as good as it is cracked up to be.

If they got together with Sony and added OSX to the mix, they could win big. I just hope that MS doesn't
make significant inroads into the consumer electronics market first. Since they've decided to go after
it, there's really no single entitiy in existence that can stop them, but if for example Apple were to
team with, say, Shuttle and/or Sony...

I hope also that this time, Sony's attempts to license the Cell work out better than with the Emotion
Engine. If they make enough of a market for the Cell that way, as a media engine for PC's, it
would undermine the relevance of the host processor in the PC. The graphics hardware is already
gradually beginning to do exactly that, but the current generation isn't general enough to make
the host irrelevant just yet.

BigJay
02-09-2005, 03:41 PM
cool and exciting. only time will tell if it hits it market.




but, why does the chip look like a dresser from Ikea??

http://img226.exs.cx/img226/4315/cell3jb.jpg

LOL

The one thing that the cell has going for it that risc didn't in the pc world is that it will run a game machine so it already has a market that won't complain about software not working. Now if Sony expands what you can do with the PS3 to include email, web, word editing etc then it could eventually push pcs out of the way.

Personally I don't see it happening as AMD et al are taking about making multi-core chips anyway so again the x86 will borrow the cell features to continue to dominate.

mummey
02-09-2005, 03:46 PM
LOL

The one thing that the cell has going for it that risc didn't in the pc world is that it will run a game machine so it already has a market that won't complain about software not working. Now if Sony expands what you can do with the PS3 to include email, web, word editing etc then it could eventually push pcs out of the way.

Personally I don't see it happening as AMD et al are taking about making multi-core chips anyway so again the x86 will borrow the cell features to continue to dominate.

Would do you people think that x86 stills runs on CISC?

P4's are RISC processors that are able to decode CISC instructions into sets of RISC instructions. The processors themselves are pretty much RISC and have been for a while now.

Thalaxis
02-09-2005, 05:05 PM
Would do you people think that x86 stills runs on CISC?


Hm... reality, for one thing.


P4's are RISC processors that are able to decode CISC instructions into sets of RISC instructions. The processors themselves are pretty much RISC and have been for a while now.

That's never actually been true. If you oversimplify what the P4 and K8 do far enough, that starts to
look reasonable, but you have to oversimplify it a LOT to get to that point.

mummey
02-09-2005, 05:17 PM
The P4 uses a CISC instruction set, but calling it a 'true' CISC processor I would say is a bastardization of what RISC and CISC processors are.

I'll throw this one out there for perspective:
http://www.embedded.com/story/OEG20030529S0047

Thalaxis
02-09-2005, 05:32 PM
The P4 uses a CISC instruction set, but calling it a 'true' CISC processor I would say is a bastardization of what RISC and CISC processors are.


It would a lot more accurate to say that modern CISC processors have a lot in common with modern
RISC processors... but there is a lot of baggage associated with the CISC instruction set that the
x86 processors implement that they have to deal with.

Renderman_XSI
02-09-2005, 07:26 PM
I agree with that. I'm just hoping that Apple doesn't remain stubborn and stick with something that may not work. Especially if the Cell processor is as good as it is cracked up to be.

"3. Apple is going to use this processor in their new machine.
Doubtful. The problem is that though the main CPU is PowerPC-based like current Apple chips, it is stripped down, and the Altivec support will be much lower than in current G5s. Unoptomized, Apple code would run like a G4 on this hardware. They would have to commit to a lot of R&D for their OS to use the additional 8 processors on the chip, and redesign all their tweaked Altivec code. It would not be a simple port. A couple of years to complete, at least. "

what is Altivec, and how does it affect DCC apps on the mac(performance wise).?

http://games.slashdot.org/comments.pl?sid=138810&cid=11616545

BTW you guys shouldnt compare a dual core Intel/AMD chip to a multi-core cell chip, one is being builded from the ground up. This is only first generation cell, they'll improve to 65-45 nm technology later on. Theres also a host of other cell related technology; FlexIO, rambus, redwood,etc. So its just not cell that appealing but the interconnect. And the unveiling at ISSCC isnt meant as hype/PR spin..its just a technicail inveiling(clock speed, flops,memory,packing,etc. If you want hype you'll have to wait until e3 :)

richcz3
02-09-2005, 08:53 PM
Several years ago I was reading how x86 had reached the end of the line and that it was impossible to increase speeds due to physics problem x. I read this same story every year about how we've all reached the end and that a new 10x faster incompatible chip is the only way forward. Every year it turns out to be completely untrue.

You say they're 'resorting' to using multiple cores, well so what? the entire design of the cell system is to link together many cpu cores. You make it sound like sony doing it is a good thing but if intel/amd do it then its a terrible last resort of a dying technology.

Yes, It has been like this since the early to mid 90's. XXXX was supposed to harald in the decline of the x86 archtecture but somehow engineers continue to push the x86 further. Yes it may be like building more stories onto an unstable house of cards, but the fact the vaunted alternate CPU's failed to dethrone the x86 PC is a testament to deep pockets, engineuity, and a huge market base.

Isn't the Cell more like a decentralized CPU. I think with the Cell people have to rethink the way CPUs control functions. Toshiba plans on using Cells in their high end TVs. :shrug:

CGTalk Moderation
02-09-2006, 09:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.