Playstation 3 CPU Almost Finished?

dnxthx writes "According to this ZDNet article the design of the Playstation 3 chip is nearly complete. The PS3 chip will have near "supercomputer capabilities" --- including 1 TFLOP. Reportedly, this chip is being engineered with Linux in mind."

Comments

Linux in mind?

[slutdot]

Cell's designers are engineering the chip to work with a wide range of operating systems, including Linux.

I don't see how that sentence translates to the statement by the submitter that the chip is designed with Linux in mind. Besides, shouldn't the OS adapt to the chip, not the reverse?

Hrm..

[qurob]

The PS3 chip will have near supercomputer capabilities --- including 1 TFLOP.

Wasn't the old PS2 a supercomputer, and there were export rules on it?

Saddam was rumored to buy some to control missles or something?

export controls?

[Jucius Maximus]

Does this mean that Japan will add export controls to this like they did with the PS2?

TheJapanese government realised that the computers in the PS2s were very powerful for the time and could be networked to create a crude missile guidance system.

Re:export controls?

[Lord_Slepnir]

create a crude missile guidance system.

Actually, the atari 2600 had this. It was called missile commander i do believe.

Must be a good thing

[JohnCC]

A large international company trusted by millions can only be a good thing for the linux community...

Learn from the last Sony hype-fest.

[Performer Guy]

Come on, we've heard the hype from Sony before with their PS2, which was a nice system but not all it way hyped to be. OK PS3 will be an interesting piece of cheap hardware but do we have to see a round of flawed comparrisons that measure a single metrics as Sony try to promote themselves to an audience only too eagre to lap it all up. Take it all with a pinch of salt.

Slashdot in mind

[ucblockhead]

Reportedly, this chip is being engineered with Linux in mind."

Translation: the marketing guys mention Linux to get slashdot coverage.

Re:Slashdot in mind

[Daniel Dvorkin]

You know, at this point, I wouldn't be terribly surprised if Slashdot coverage actually does register on marketing radar. Sure, the number of Slashdotters is pretty small compared to the total target market of the PS3 (or any other major piece of geekware) but we're early adopters, a big enough crowd to provide a spike in early sales figures; we're also, more importantly, the sorts of people others come to for advice on what geekware to buy.

A terraflop?

[wilburdg]

The terraflop statistic is a little hard for me to swallow.

The NERSC IBM SP RS/600 (the fifth most powerful computer in the world, according to top500.org) located in Berkeley consists of 2,944 processors. The processors are distributed among 184 compute nodes with 16 processors per node. Each node has a common pool of between 16 and 64 GBytes of memory.

This machine is a 3 terraflop system. Although, I guess three PS3's could do the same...

Sony is pretty funny

[quantax]

Wow, same song, different year. Last time Sony acted like the PS2 chip was 'God-on-a-PCB'. They even claimed that they could make highend 3D dev systems that could blow the machines of that time away with super realtime rendering, etc. And now, they say they have a supercomputer-like chip. Maybe for the PS4 they can tell us about the NASA beowulf-cluster-like chip which can predict the stock market's picks up to 1 year in advance. Oh, and also create a 1:1 model of the universe, complete with infinity. Seriously, I understand that these chips are powerful, but Sony hypes this crap like its god-in-a-can. Lets not buy into it.

Re:1 TFLOP?

[unicron]

It's not about the fact that the PS3 can push 1 TFLOP, it's about the fact that the PS3 can push 1 TFLOP while being small and relatively affordable that makes it bleeding edge technology.

I for one am getting kind of tired of all these technology pushes in gaming consoles while the games continue to go down hill in terms of enjoyability. Now, it may just be my age at the time, but when I remember back to being a kid and playing Nintendo, I remember more than half the games I ever played were REALLY, REALLY fun to play. I'm 23 years old and I can talk forever about old school Nintendo with friends that can remember the days. Too often these days we judge games based on their technological feats, giving a game credit for crap like "volumetric fog" and "real time shadows", etc. but we hardly ever just say "That game is just plain fun".

I think it may be time to pick up a Gamecube, especially with 3 old school classics getting a revamp(Metroid, Zelda, Starfox). Maybe then I can relive that joy from childhood.

Re:Late 2004?

[slyfox]

The article states that they have "taped-out" the design. However, when I visited IBM-Austin last December (I gave a presentation on my research) they were still in the high-level idea phase. There is no way they could have decided on the design and completed it so quickly. My guess is this is a "test chip", like the one they did for Power4. Power4's test chip tested some of the critical circuits and such, but it was not the final design.

That said, it seemed like they were considering some pretty wild ideas. However, I remember hearing about plans for the Playstation 2 chip a couple of years before it shipped; at the time it was hard to fathom, but when it arrived it wasn't as big a leap as I thought it was going to be. (Though still quite impressive.)

I expect the Playstation 3 will be just as impressive, but not earth-shattering. They key will be how easy it is to write programs that take advantage of the raw computational power.

Great ...

[UnknownSoldier]

From the article:
"It's like a beehive -- cell components can also be ganged together," he said.

Just when I thought programming the PS3 couldn't be any *worse* the then PS2 (lots of fun debugging the EE, VU0, VU1, GS, SPU, IOP all running simulatenously on the PS2 :), along comes 'linked' cpus. Sure parallelization rocks for performance, but it's a head ache for game design & implementation. This is one thing the X-Box got right - port your PC game over in days, not months. Ok, enuf k'vitching.

How long do we have to wait for Gran Turismo to show-case the PS3 ? ;-)

Re:1 TFLOP?

[Skyshadow]

Of course, you were probably more easily amused as a kid. I seem to remember that a lot of games were always fooled by perfecting a single trick or strategy, then repeating it over and over.

To me, Crash Bandicoot is every bit as fun as Super Mario (not to mention that it has great attitude), Morrowind kicks Phantasy Star's ass and Grand Theft Auto III... well, there's nothing that really compares.

So, basically, I completely disagree with the idea that games aren't as good as they used to be. *Some* games are worthless tech showcases (I call these "Jurrasic Park games"), but then those were always around, weren't they?

Re:Thats it?

[Lord_Slepnir]

Have you played the new Unreal 2003 demo yet? People already have toasters in their computers, so computers in their toasters isn't that far off.

You saw it here first: The FIRST TOASTER VIRUS

if(Toast_Present == 1)
{
Turn_On_Coils();
While(Toast.OnFire() = 0)
{
if(Lever.Manual_Eject() == 1)Lever.Jam();
if(Power.Unplug() == 1)Power.Source = reserve_battery;
}
Eject_Flaming_Toast_At_User();
}

Several things

[Sycraft-fu]

First yes, basically all current Intel and AMD chips can pull a Gflop. More or less and P3 or Athlon chip above 850mhz can do 1 Gflop in real world tests (specifically according to SiSoft Sandra).

Second the classification of a G4 as a "wepaon" or a "supercomputer" is not correct. The way that is done is based off of theortical operations per seconds (be they interger or floating point). In 1998 that was 2,000 MTOPS (million theoritical operations per second) or 2 Gflops if you want to look at it that way. That has since changed and currently the US can export up to 190,000 MTOPS computers to "Tier-3" countries (countries judged unsafe in terms of non-proliferation of mass destruction weapons) which are places like China, Russia, and most of the Middle-East.

Finally, Sony probably is telling the truth about Tflop perofrmance.... Sort of. I'm betting that the chip wiill have a theoritical max of 1 Tflop, which is not unheard of, provided we are talking about speical DSP operations for graphics type stuff. The GeForce 4 4600 gets about 1.23 Trillion ops per second according to nVidia. Thing is, the GeForce 4 is a graphics DSP, all it does is push pixels. It's subunits do things very fast, but can do only that one thing (ie vertex shaders ONLY do vertex transforms, not general work). A P4/G4, on the other hand, can do anything. It can do all the same kinds of calculations a GeForce 4 can, but can also do all the calculations any other DSP or system can, given enough time.

For a long time we've had the ability to design specialised chips that ar much faster, but more limited, than general purpose CPUs. That's the whole reason for ahaving a 3d accelerator. You just can't make a CPU that fast yet, it would take hundreds of CPUs working together to equal the power of a GPU, HOWEVER that GPU is good only for graphics. You still need a CPU for general purpose calculation.

In a video game console, the lines often become a bit more blurred. One chip may do many different things. Some of the functions traditonally on the GPU in computers might be on the same chip that happens to do CPU work as well.