Indeed. the Cell seems like its gonna be a crazy piece of technology and if all goes according to plan....*shivers*

btw heres imho a better article : http://ps2.ign.com/articles/585/585865p1.html

 

Slashdot has several articles referenced

http://games.slashdot.org/article.pl?sid=05/02/07/2310237&from=rss

 

i have one Q' tho, when its used in computers, how much will the processor cost? cuz i might save up for it, and continue using my crap pc (900mhz) until its released, or buy a 3.4ghz in May for like 300usd

 

I can't answer your question, but it sounds REALLy expensive.

edit: Oh btw, I don't think it's going to run Windows would it? I read it's going to support Linux though.

 

Very nice, we shall how the computer industry will handle this change. Hopefully we can finally be rid of this ancient x86 architecture and go forward.

 

Cell Processor Breakdown

Ok so finally the cell processor is unveiled by Sony, IBM and Toshiba. It will not be running at 4.6 Ghz, and even during internal testing the speeds were in the vicinity of 4Ghz.

Most probably to gain stability the final clock speed is going to be lower according to me :thumb:. Glad to know that the bulk of work on the processor is done. Full story here. Ok so what about the graphics part?

On a side note PS3 will be unveiled BEFORE E3....check here. W00t cant wait so looks like if they put their back into it PS3 might make it by Christmas. :wub:

 

and i hope that the ps3 have alot of blue ray disc errors cause sony sucks and i bet there will be a problem runing games on it as well.

 

It is great everyone is posting about the Cell processor, although I do think it is a little bit too much. There is already a post in the Hardware discussion about it here:

http://www.ngemu.com/forums/showthread.php?t=59967

I suggest a merge of threads, since it is pretty much the same topic. It is about the specs and capabilities of the cell processor, which includes its use in the PS3.

 

Yeah it would be good to replace the existing hardware. But Cell seems to good to be true. If they are gonna release cell they can't have more than two since no one will code programs to run all of them. Also like most sony consoles the cell will be hard to write games for.

 

Yeah it would be good to replace the existing hardware. But Cell seems to good to be true. If they are gonna release cell they can't have more than two since no one will code programs to run all of them. Also like most sony consoles the cell will be hard to write games for.

that's not a isue because sony already said that their main concern is to make it more easy to delope programs for the console, lets hope that tey keep their promise.

 

I can't answer your question, but it sounds REALLy expensive.

edit: Oh btw, I don't think it's going to run Windows would it? I read it's going to support Linux though.

I read (I think at theregister.com) that Sony says that Cell will be "OS-neutral", so I'm not sure if that means Windows support or not. My guess is if M$ cant get it easily working under xp, they'll wait till longhorn.

 

Well I'd imagine that if the Cell makes big enough waves, if MS doesn't give it support, MS are the ones that'll end up screwed, but then again: can Cell make waves without MS behind it?

>I read (I think at theregister.com) that Sony says that Cell will be "OS-neutral",

The OS-neutral is refering to the Cell requiring low-level programming (ASM essentialy) and an app for the Cell would essentialy bypass the OS' hardware abstraction layer.

 

The OS-neutral is refering to the Cell requiring low-level programming (ASM essentialy) and an app for the Cell would essentialy bypass the OS' hardware abstraction layer.

Would this mean that our current set of programming language is going to be reduced back to using ASM and current programming language will be out the door?

 

It just makes me wounder, how much will it cost to build these cells... it sounds expensive, but if they're to be in a console, it can't be that expensive HMM, very interesting indeed.

 

I have no idea how much it would be to produce the cell processer, but considering at least the PS3 one will be mass-produced, it should be relativly cheaper to produce than a normal desktop processer.

 

>Would this mean that our current set of programming language is going to be reduced back to using ASM and current programming language will be out the door?

That would be the plan I'd imagine. That's how the best games on the PS2 are made and Sony encourages developers to work like that IIRC.

 

You aslo have to take into account that the cell processor in the PS3 will most likely be a weaker version than the most powerfull ones, because then it would probably be too expensive, but then again who really knows. We will have to wait and see.

 

>
"Would this mean that our current set of programming language is going to be reduced back to using ASM and current programming language will be out the door?

That would be the plan I'd imagine. That's how the best games on the PS2 are made and Sony encourages developers to work like that IIRC.
>

LOL, I can assure you that no developer have ever made a commercial game for the PS2 in ASM, good sarcasm Kane, not sure if this guy got it though.

 

PS2 games are made by programming the bulk of the game code in C or C++, and the most hardcore graphics routines and the transform code for the vector units in assembler. I'd assume that the Cell will be programmed like this too: C/C++ for the main PowerPC core, and specialized assembly, or perhaps something looking like HLSL with luck, for the SPE:s or APU:s or what today's name for the 8 vector thingies is.

 

I was given to understand that Sony wanted people programming 'directly into the metal' and thus didn't even provide C libraries for developers at the PS2's release. Obviously I've been mislead

 

Lets stay serious

The fact that they're suposing that the cell have been tested on Windows is funny,I don't think it's DEC Alpha nor X86 architecture compatible if we consider the poor performance of the X87 FPU and I can't imagine that M$ could have developed even a kernel for this CPU. :rotflmao:
This CPU is quite remarkable offering 8+1 cores, but there's a need of a seriously optimized compiler to manage the fact that it may be able to FETCH,DECODE,READ,EXECUTE,WRITE 10 instructions silmutaneously cause it needs that the data of an instruction is not manipulated by the previous ones,so I emit the hypothesis that part of Cell will be used to compute transformations of vertices...etc...
But we can imagine than both the cell and the GPU may work together when a lot of graphic processing power is required so it's that both of them compute transformation etc, so they may share part of the cache memory in this case that both of them works the fillrate have to be awesome, or this solution is rather inefficient.
The cell may be used also for post processing effects.
In all cases I think the graphics will not rely uniquely ont the GPU
The number of transistor is not very high though if we consider the 2.5 MB caches (2500000*8*7 =140*10^6 transitors for the cache) There is only few transitors remaining for the cores and the PPC alike "manager"
So i think that it's a fake information to destabilize people
I consider also that the 32MB memory is a real lie, a single texture of 2048*2048*32B (uncompressed) could not fit in this case.
Maybe I'm wrong but still I doubt it's true.

 

Well since they are four cells on the ps3 its gonna be fast. Also the Cell chips are gonna be in IBM high end servers so its gonna be awhile till they make them for desktops. Also since IBM is getting out of the personal pc market we gonna have to see who is gonna release them on the pc.

 

Well since they are four cells on the ps3 its gonna be fast. Also the Cell chips are gonna be in IBM high end servers so its gonna be awhile till they make them for desktops. Also since IBM is getting out of the personal pc market we gonna have to see who is gonna release them on the pc.

Would it be too hard for IBM to just release the chip like Intel and AMD does? Couldn’t this lead to a new, more competitive, personal PC chip race? Then we could build our own cell processor systems. Hell, It would be awesome to buy a motherboard with say, 10, cell slot (those things are tiny!) and when we need more power, add another cell

About Operating Systems, IIRC the reason that the cells are so powerful is that they have "super ultra uber" powerful vector processing, and that windows is not vector based. Also IIRC isn’t Apple's OSX vector based? This would give Apple a good reason to allow their operating system to be sold and put on personal PCs. I really like the way OSX flows on my buddy's comp.

 

I would really like to see Cells on Apple computers. Since they use processors used from IBM servers. But it all comes down to IBM and other companies associated to Cell making money. Also they would have better chance using these Cell processors at Server market instead of desktops. I dont really see microsoft supporting Cell processor on their OS since they are competing agaisnt each other.


OH yeah IBM is selling its PC Division to Lenovo. Lenovo is computer based in China. Lenvo will be selling under IBM's name for next 5 yrs. Just let you all know that IBM wont be releasing Cell processors for their PC's because they wont have PC division.

 

The coolest concept is the ps3 on a PCI-E card.