Cell article: 40x faster than EE, Cell arch to last 10 yrs

[Megadrive1988]

*prototype Cell is at least 40 times faster / more powerful than
Emotion Engine.

*present generation Cell-architecture is intended to last a decade

*engineers of other ASICs might be better suited toward writing code for Cell, then engineers who write code for general purpose microprocessors

http://neasia.nikkeibp.com/neasia/001090

 

[Titanio]

Hehe, was about to post this. Some interesting stuff in there.

Presumably the above are "guesses"..

Also:

Entry into the development areas is strictly controlled, so very few engineers have actually seen Cell chips in operation. In these secret labs there are development boards the size of pillows, mounting twin Cell chips with little air-cooled heat sinks small enough to sit in the palm of your hand. Development is under way on 3D graphic draw libraries for gaming, HDTV demodulation software, and more.

Twin cell chips? Why?

Also, if those graphics libraries are being made at SCE, it's a good bet you'll be able to leverage the PS3 CPU for some graphics tasks.. (vertices? please?)

 

[one]

It's another excerpt from the Nikkei magazine

 

[McFly]

Twin cell chips? Why?

>500Gflops+GPU=1TFlops!

Fredi

 

[hey69]

that diagram is so idiotic.

they are comparing frequencies...

so in that math, a opteron 2Ghz is 2years behind ??

 

[Npl]

Twin cell chips? Why?

No dedicated Chip for GFX? Just because a dedicated GPU turns out to be better for a Game-Console doesnt mean a 2-Cell approach aint fitting better for other devices( a Media-Cell-Station would need beefy decoding abilities, not 3D-Gfx ).

 

[Qroach]

engineers of other ASICs might be better suited toward writing code for Cell, then engineers who write code for general purpose microprocessors

:?

 

[ndoogoo]

Divide whatever Sony says by 10 to get rid of the Sony's reality distortion bubble. Remember all the statements Sony made about the PS2.

 

[Titanio]

Divide whatever Sony says by 10 to get rid of the Sony's reality distortion bubble. Remember all the statements Sony made about the PS2.

Everything that has been disclosed thusfar re. Cell performance has been theoretical - we're all very well aware of that.

Also, what specific performance claims re. PS2 did you have a problem with? I'm not talking about "Toy Story" or "Missile guidance" or the like, I mean in terms of the numbers they gave.

It's best to clear this up now before we have a rash of such claims once PS3 is unveiled at E3..

 

[Kolgar]

Divide whatever Sony says by 10 to get rid of the Sony's reality distortion bubble. Remember all the statements Sony made about the PS2.

It may have taken a while (too long), but I think games like Metal Gear Solid 2 and 3, Gran Turismo 4, Tekken 4 and 5, God of War, and Virtua Fighter: EVO gave us pretty much what Sony promised with PS2.

You just need to put things in context - i.e., sure, maybe PS2 CAN do 66 million polygons, but with no effects and maybe not even in a real game situation.

Same probably holds true with Cell. Context. Luckily, we got some smart fellows on this here board who can help us sort out the shit from the Shinola.

 

[Phil]

Heh, I remember having tears in my eyes after seeing that 90MB Metal Gear Solid 2 trailer for the first time. It was mindblowing to even think that what they were showing was indeed realtime rendered. Another popular footage were the first screens of GT3 in motion, particularly the deep forrest track with the light shining through the forest and onto the cars at a brisk 60 frames per second.

Although the numbers and the hype were amazing - I don't feel let down with what we got from then until today. Despite the hype, it was also obvious that the learning curve of such hardware would be high and that it would take months/years for developers to get stunning results out of it. Despite that popular belief, I am still quite amazed at what developers are achieving today.

just my humble 2 cents.

 

[Shifty Geezer]

Divide whatever Sony says by 10 to get rid of the Sony's reality distortion bubble. Remember all the statements Sony made about the PS2.

So Cell is 4x power of EE?

 

[Guden Oden]

Divide whatever Sony says by 10 to get rid of the Sony's reality distortion bubble. Remember all the statements Sony made about the PS2.

Sony hasn't said anything about PS3 yet, and PS2 real-life performance is NOT 1/10th of max specs. Maybe it's you who's in possession of a reality distortion bubble...

MS has also been guilty of overinflating specs I might add (and let's not forget that LYING Abrash article of how xbox would be perfectly capable of maxing out rendering performance within the allotted bandwidth budget...)

 

[jvd]

Sony hasn't said anything about PS3 yet, and PS2 real-life performance is NOT 1/10th of max specs. Maybe it's you who's in possession of a reality distortion bubble...

Not true , they said it would have a bluray drive and they said the cell will be 1tflop Lets not forget what has been said .


Anyway lets cut the crap about what sony says and what actually is true and keep to the topic started in the thread

 

[Farid]

they said the cell will be 1tflop

Who did say that?

 

[BOOMEXPLODE]

Of course, everyone should realize that these kind of numbers are hokey. It's easy to see how they came up with the number though.

Cell= 256 gigaflops
EE = 6.4 gigaflops
256/6.4=40

So, Cell is 40x more powerful than EE! Woohoo bust out the confetti.

 

[Shrike_Priest]

IBM said that Cell systems would operate at 1Tflop back when they announced the project.

They never said the PS3 would have 1TFlop-performance, though.

 

[Shifty Geezer]

Wow! I thought I'd Google this to find the original source, and instead found MS FUD unparalled, which shows how 'the media' haven't got a clue when it comes to reporting factually...
Here's a site that reiterates the same info of several
http://www.nikush.com/index.php?option=com_content&task=view&id=761&Itemid=46

The version expected to power the Playstation 3 should run at over 4GHz (4.6GHz being the most likely frequency) and be capable of 256 billion calculations while it is claimed that it will include features which will make game copying impossible.

Microsoft has not been as forthcoming as Sony with the details of the processor behind the next Xbox console. We do know that the company hired IBM to develop an Xbox 2 exclusive processor based on Big Blue's Power architecture. Microsoft has also revealed that the design will be able to process more than a trillion calculations per second (1 teraflop) which is an impressive figure.

There's lots of talk of this XB360 having 1 teraflop, being powerful as a server etc. There's lots of other nonsense when you read back thorugh the history of speculation, including Cell not being in PS3 but unveiling in 2007/2010.

However I have failed to find even KK's talk of 1 teraflop thus far. I would have thought such a search would be easy.

Maybe we should hold a competition to find the original promises for Cell/PS3?

 

[Carl B]

IBM said that Cell systems would operate at 1Tflop back when they announced the project.

They never said the PS3 would have 1TFlop-performance, though.

Well I don't think it was even the Cell, but the 'Broadband Engine' that was going to be bringing the 1 TFlop performance. If it indeed was, then in fact the Cell has met, hypothetically at least, it's performance target 100%.

 

[jvd]

It was said by one of the sony people that cell would deliver 1tflop performance , i would have to search back like 30 pages or more in the forum to find it heh