Kutaragi Ken:Want a teraflop? You have to buy a rack from us

[rabidrabbit]

Re: ...

This "Teraflop" processor misunderstanding happened because Kutragi's definition of a "processor" is different from our accepted standard definition. To Kutaragi, a MCM rack module is still considered a processor, whereas the standard computer science terminology does not accept such definition.

CELL really is exactly the opposite of what we thought it was; it is not some 450 mm2 mega processor chip burning 200 W of power to attain 1 teraflop per chip, it is rather a collection of very low-power small processors small and cheap enough to use whole lot of them to construct computing modules. If you were to contruct a rack of CELL chips, then each CELL chips has to burn very little power, or else you cannot construct one due to power consumption and heat problems.

I must be more confused about this 'Cell' than ever, but wasn't Cell always thought of being a "collection of very low-power small processors small and cheap enough to use whole lot of them to construct computing modules", rather than "some 450 mm2 mega processor chip burning 200 W of power to attain 1 teraflop per chip" :?

As far as I've understood, a Cell chip or processor is constructed of multiple Cell cores. A PS3 Cell processor might have like some <64 Cell cores to achieve 1 TFlop (or less).

I'm really no computer architecture expert, so I don't know if it would be possible, but a PS3 with just 4 'Cells' just seems low in light of those numbers on that powerpoint presentation.

Anyway, why isn't there any more info anywhere about that meeting where KK held that presentation?

 

[_phil_]

.tiring times .too much prophets .

 

[passerby]

Actually, it's quite likely that the presentation slide in question was simplified to better explain scalability. Easier to use the unit "1" instead of "6", or "9", or "13.3" when you are trying to explain what they are trying to illustrate.

Anyway, I don't think it's interesting to spend too much time discussing about something which is likely due to over-simplication/poetic license.

Side note: Of course there is the oft-repeated assertion that Cell is a "broadband processor". So perhapes that's what it really is, and the PS3 core is really something else.

 

[London Geezer]

Re: ...

Anyway, why isn't there any more info anywhere about that meeting where KK held that presentation?

HAHAHA *imagines Deadmeat spending hours to prepare a fake presentation just to show it to us and say -i told u so!-* just kidding

 

[rabidrabbit]

Who is Seth Schuler?

 

[Grall]

Enough, already! Geez, time to toss away that rancid old dead meat, it's stinking up the whole joint!

I'm both surprised and NOT at the same time that people put such emphasis on that weird chart. I didn't think anyone would believe a next-gen product to have less than 1/6th the raw power of a current-gen product, but if you're blinded by fanatical hate against Sony and everything it stands for and is associated with (/me holds up picture of betamax player, sees Dead start to sputter with steam coming from the ears), I guess anything's possible.

I mean, common sense SHOULD tell you something's seriously f****d up there.

The weird chart that has Dead worked up into such an (undead?) frenzy is as far as I can tell the exact same used when illustrating Blue Gene (for f**k's sake it even says Blue Gene down near the bottom), and we know Cell is NOT the same as Blue Gene. Something's fishy here alright, that's for sure.

Like always, everyone should wait until official product announcements are made before crowing victory.


*G*

 

[akira888]

Is it just me or are any of you totally uninterested at this point (due to information saturation) what CPU goes in PS3 and more and more curious as to what sort of GPU will appear in the device? That, to me, seems to be a much more important factor at this point.

Having great physics and AI is terrific but console's graphics have always been their primary selling point. A 65816 @ 3.58Mhz is weak compared to a 7.16Mhz 68k yet nobody to this day claims that the Mega Drive was "more powerful" than the S-NES, which carried the day thanks to its far superior custom graphics and sound hardware.

 

[London Geezer]

Is it just me or are any of you totally uninterested at this point (due to information saturation) what CPU goes in PS3 and more and more curious as to what sort of GPU will appear in the device? That, to me, seems to be a much more important factor at this point.

Having great physics and AI is terrific but console's graphics have always been their primary selling point. A 65816 @ 3.58Mhz is weak compared to a 7.16Mhz 68k yet nobody to this day claims that the Mega Drive was "more powerful" than the S-NES, which carried the day thanks to its far superior custom graphics and sound hardware.

well... those were different times... no need to tell u...
and yes, there are people (ME) who are more concerned about physics and animation than "how many texture layers u can put on a polygon in one cycle".

 

[DeadmeatGA]

I'm both surprised and NOT at the same time that people put such emphasis on that weird chart.

It must have certain significance if Kutaragi kept showing it over and over at various presentations.

as far as I can tell the exact same used when illustrating Blue Gene and we know Cell is NOT the same as Blue Gene.

Maybe, Maybe not. We all know that Blue Gene/L had a great impact on the design of CELL architecture.

Something's fishy here alright, that's for sure.

Yap, we must been had as fools by SCEI marketting men for the past two years.

 

[akira888]

and yes, there are people (ME) who are more concerned about physics and animation than "how many texture layers u can put on a polygon in one cycle".

Well, these days "texture layers" are less important than the number of shader ops the chip can do per second. And while you might disagree with the prevailing sentiment concerning console design, that does not change the fact that nonetheless it is the prevailing sentiment. Nevertheless, whatever CPU Sony places in the PS3 will be more than fine enough for basic game engine/physics tasks, and animation is more and more a graphics function all the time.

It's the GPU I worry about.

 

[StefanS]

and yes, there are people (ME) who are more concerned about physics and animation than "how many texture layers u can put on a polygon in one cycle".

Well, these days "texture layers" are less important than the number of shader ops the chip can do per second. And while you might disagree with the prevailing sentiment concerning console design, that does not change the fact that nonetheless it is the prevailing sentiment. Nevertheless, whatever CPU Sony places in the PS3 will be more than fine enough for basic game engine/physics tasks, and animation is more and more a graphics function all the time.

It's the GPU I worry about.

I have to disagree here; game physics are light years away from being close to realistic: check http://www.racer.nl for a more realistic physics racing engine and that game is sloooow as hell. That game also uses very limited physics, limited shock absorber dynamics, etc. and still the engine is best I've seen at describing the swinging of a car's coachwork after turning. The site has also links to physics papers concerning car simulation... Physical calculation can be very heavy on a CPU it just depends on the intensity of realism one wants to create... A powerful CPU can greatly enhance the depth of a racing simulation, etc...

 

[Megadrive1988]

the GPU interests me more, as well.

there should be more discussion on Graphics Synthesizer 3

 

[wazoo]

Nevertheless, whatever CPU Sony places in the PS3 will be more than fine enough for basic game engine/physics tasks, and animation is more and more a graphics function all the time.

It's the GPU I worry about.

Hum, no. We want more than basic game engine/physics tasks, and if you do not know what to do with this computational power, other will surely find out.

 

[chaphack]

hey! If it all goes well, CELL PS3 will have a 5 year lead, over rivals, in physics and whatever CPU related stuffs.

 

[Phil]

Guys...

while I do love the speculation about CELL and PS3 at the moment, we got to remember that we still have 2003 and that a possible successor to PS2 is expected at the very earliest in 2005/2006.

We've recently found out more about Microsofts plans on Xbox2 than we even anticipated (the IBM/ATi deal). We can assume that both Nintendo and Microsoft are very hard at work in planing the new system and you can bet that they are watching Sony very carefully. They need to, after all, Sony is the dominant player this generation and they are the ones that determine how things will progress from now. Who knows, maybe these latest articles are the reality of PS3 (/-CELL) - or maybe it's just a publicity stunt to, once more, fool the competition into believing they're not up to it, when in fact they have it all figured out. It can be both ways, but knowing how big cooperations do their homework, I will assume the latter is true and that there is no way to determine Today, 2 years in advance, what the deal on CELL/PS3 really is.

Ask yourself one question: would you, as Sony, show all your cards, knowing that Microsoft & Co. are watching each and every step very carefully?

 

[chaphack]

or maybe it's just a publicity stunt to, once more, fool the competition into believing they're not up to it, when in fact they have it all figured out.

once more ? heheh they didnt exactly figured out much in the past though.

i dont think such a pubilicity stunt will deter MS from getting all known/needed info. everyone probably have bunch ol' moles in here and everywhere! nothing money cant buy or something!

 

[zidane1strife]

We can finally put the "Teraflop" chip non-sense behind us and discuss something productive instead...

That will have to wait till, most likely, early next year... but I agree it probably won't be 1Tflops

 

[akira888]

I'm sorry for not being more clear, but what I meant to say was that there has been several 10+ and a couple of 20+ page threads on PS3 CPU and almost nil discussion (or even raw speculation) into the GPU of the beast. That is my concern.

 

[chaphack]

Maybe this might prove an insight to KK design choice,

"In regards to what the PlayStation 2 development team has in mind for the system, Kutaragi commented, "Our target is a crazy level in bandwidth, pixel rate, design rules and transistor count. It's a very, very cutting edge chip."

Sounds like he is more of a brute power CPU guy...?
I say the PS3 GPU aint going to be 1 up on whatever ATi has to offer.

 

[Paul]

This is hilarious, both Chap and Deadmeat going crazy as usual.

Hey Deadmeat, does this look famililar?

Oh yea! IT'S BLUE GENE.

The Cell project is totally different than the Blue Gene project, both in moneys spent and performance they will attain.

So please Deadmeat, get a grip. A four core chip is going to attain 4GFLOPS when a Emotion Engine done in 98 can attain 6? Get real.