So whats the buzz on CELL workstations?

Are Cell workstations different from PS3 development kits at this point? If so, how common are Cell based PS3 dev kits?

PS3 and its dev kits probably will be derived from this workstation. We'll find out soon enough, several more weeks now :)
 
Megadrive1988 said:
I haven't heard a peep about Cell workstations actually being in the hands of any developers. then again i dont have ties to any developers or anyone in any industry that might have access to one. my bet is, by spring (March-June) we'll hear of developers having early Cell-based workstations.
As the schedule suggested in the last July, TOOL ver. 1 for the Next System has been slated for April 2005, so except for the SCEI close allies such as Namco, Koei, EA etc. under NDA, no one won't see TOOL ver. 0.xx...

psm01.jpg
 
one said:
As the schedule suggested in the last July, TOOL ver. 1 for the Next System has been slated for April 2005, so except for the SCEI close allies such as Namco, Koei, EA etc. under NDA, no one won't see TOOL ver. 0.xx...

A lot has changed since then, the change in graphics design has likely delayed things.
 
DeanoC said:
one said:
As the schedule suggested in the last July, TOOL ver. 1 for the Next System has been slated for April 2005, so except for the SCEI close allies such as Namco, Koei, EA etc. under NDA, no one won't see TOOL ver. 0.xx...

A lot has changed since then, the change in graphics design has likely delayed things.

What change? :rolleyes: The nVIDIA partnership announcement was in Dec 2004, and the diagram is from PlayStation Meeting in July 2004.
 
DeanoC said:
http://www.cmpevents.com/GD05/a.asp?option=C&V=11&SessID=4777

Description: This presentation gives an overview of the architecture of a CELL Processor, which is a multi-core chip consisting of a 64-bit Power architecture processor, multiple streaming processors, a flexible IO interface, and a memory interface controller. Additional detail is given on the design of a 4-way SIMD streaming data processor in which software controls data movement and instruction flow in order to improve data bandwidth and pipeline utilization.

This description is a carbon copy of the ISSCC 2005 Preview.
 
What change? The nVIDIA partnership announcement was in Dec 2004, and the diagram is from PlayStation Meeting in July 2004.

He's trying to tell you something. That is, if you want to listen.;) For me it doesn't matter, so I'll say my thoughts out loud. The deal with Nvidia was completed just before they annouced it. They didn't collaborate for two years prior on any Nvidia designs. That was all PR. They are working on it with sony now.
 
Qroach said:
What change? The nVIDIA partnership announcement was in Dec 2004, and the diagram is from PlayStation Meeting in July 2004.

He's trying to tell you something. That is, if you want to listen.;) For me it doesn't matter, so I'll say my thoughts out loud. The deal with Nvidia was completed just before they annouced it. They didn't collaborate for two years prior on any Nvidia designs. That was all PR. They are working on it with sony now.

Then, can I ask you a question? Why was the announcement of the nVIDIA partnership at that timing, just a week after the official Cell announcement?
Coincidence, coincidence :D Though they've not unveiled PS3 yet, obviously Sony was trying to tell me something. :LOL:
 
one said:
Qroach said:
What change? The nVIDIA partnership announcement was in Dec 2004, and the diagram is from PlayStation Meeting in July 2004.

He's trying to tell you something. That is, if you want to listen.;) For me it doesn't matter, so I'll say my thoughts out loud. The deal with Nvidia was completed just before they annouced it. They didn't collaborate for two years prior on any Nvidia designs. That was all PR. They are working on it with sony now.

Then, can I ask you a question? Why was the announcement of the nVIDIA partnership at that timing, just a week after the official Cell announcement?
Coincidence, coincidence :D Though they've not unveiled PS3 yet, obviously Sony was trying to tell me something. :LOL:

There was high level talk and nVIDIA was following very closely PlayStation 3's R&D without receiving money from SCE or having a real contract.

nVIDIA for the last two years had its best engineers working hard on NV5X and while they were developing it they kept as one of their R&D goals the scalability of the architecture from the high-end (desktop PC GPUs, possible console GPU, Render-farms, etc...) to the low-end (PDAs, Cell Phones, low cost Desktop and Notebook GPUs, etc...) of the CE market.

They wted to go for PlayStation 3, but it was probably a mix of the "we have the best technology" + "we have the best tools" + "we have the best 3rd party ISV Developer Relations teams" speach from nVIDIA with more and more aggressive pricing that won them the contract.

Believe what you will, but as stated before on this forum... the GPU was not supposed to be nVIDIA's, someone else HAD the contract till about two weeks or so before the announcement was made that SCE had selected nVIDIA a partner in the GPU for PlayStation 3.
 
DeanoC said:
http://www.cmpevents.com/GD05/a.asp?option=C&V=11&SessID=4777

Eh!!! So is the 'main CPU' really going to be 64bit, or is that just some cut-and paste without thinking. Also note the wording Additional detail is given on the design of a 4-way SIMD streaming data processor. Where data processor is implied as not equivalent to Cell processor.

OK, so CP is made up of DP(s) - that one's easy.

Also note the words 4-way. Not sure how they are using the terminology here, but it is universally accepted that if we say n-way box it means we are refering to a system with n processors within.

So 1 DP is made up of 4... something. Guess.

Wonder if that one lecture will be overcrowded.
 
It can't be that hard for Sony and Nvidia engineers to get a NV40 up and running on a CELL workstation, until NV50 is ready later in the year.
 
passerby said:
DeanoC said:
http://www.cmpevents.com/GD05/a.asp?option=C&V=11&SessID=4777

Eh!!! So is the 'main CPU' really going to be 64bit, or is that just some cut-and paste without thinking.

I do not think it is a typo... it is the same thing they wrote in the ISSCC preview papers and had in the famous Peter Hofstee's video talk on the CELL processor.

I guess we all underestimated the role of the PU in regards to high performance implementations like PlayStation 3.

Also note the words 4-way. Not sure how they are using the terminology here, but it is universally accepted that if we say n-way box it means we are refering to a system with n processors within.[/quote]

Read this patent:

http://makeashorterlink.com/?S23512C3A

4-way SIMD instructions can be something like this: we have 4 FMACs and they all perform a FP MADD on one field of three 128 bits registers writing the result in one field of a 128 btis destination register.
 
Brimstone said:
It can't be that hard for Sony and Nvidia engineers to get a NV40 up and running on a CELL workstation, until NV50 is ready later in the year.

Who sai that CELL WorkStations were planned to have standard PCI-Express and AGP busses ? Maybe they have them, maybe they do not.

Maybe they want to talk to developers about the GPU when they have the NV5X-PS3 almost completed.

The CPU will be the main focus at GDC for developers, they can experiment with advanced DirectX 9 features on the PC side while they wait for the final shipping WorkStation.
 
Gubbi said:
IMO, the four way SIMD streaming processors refers to four APUs.
IMHO, that's not the case. That quote is just referring to a SIMD processor that is able to manipulate 4 component vectors.

ciao,
Marco
 
7.4 A Streaming Processing Unit for a CELL Processor
3:15 PM
B. Flachs1, S. Asano2, S. Dhong1, P. Hofstee1, G. Gervais1, R. Kim1 , T. Le1 ,
P. Liu1, J. Leenstra3, J. Liberty1, B. Michael, S. Mueller3, O. Takahashi1 ,
Y.Watanabe2 , A. Hatakeyama4,H. Oh1, N.Yano2
1IBM, Austin, TX
2Toshiba, Austin, TX
3IBM, Boeblingen, Germany
4Sony, Austin, TX
The design of a 4-way SIMD streaming data processor emphasizes
achievable performance in area and power. Software controls data
movement and instruction flow, and improves data bandwidth and pipeline
utilization. The micro-architecture minimizes instruction latency and
provides fine-grain clock control to reduce power.
 
nAo said:
Gubbi said:
IMO, the four way SIMD streaming processors refers to four APUs.
IMHO, that's not the case. That quote is just referring to a SIMD processor that is able to manipulate 4 component vectors.

ciao,
Marco

Exactly, as it is basically the same kind of terminology they use when they talk about the VUs in the Emotion Engine.
 
Panajev2001a said:
Believe what you will, but as stated before on this forum... the GPU was not supposed to be nVIDIA's, someone else HAD the contract till about two weeks or so before the announcement was made that SCE had selected nVIDIA a partner in the GPU for PlayStation 3.
Hmm? Do you have any concrete info on exactly when it's confirmed that nVIDIA won the contract? :?:

In my estimation, it's before July 2004 when the schedule diagram was presented. How can they present the plan of the unveiling event without determing who is the main partner for Media Processor?

In DeanoC's and Brimstone's view, it's in December or November 2004. But assuming SCEI does all fabbing, especially with eDRAM, it's very unlikely that they can proceed to the mass production in Q4 2005 if they reached the agreement in Nov/Dec 2004. It's still a speculation, but I believe it has no logical error.

Also, change to nVIDIA can speed up the development of the PS3 rather than slowdown DeanoC suggests if the nVIDIA solution is superior than the alternative, as nVIDIA must be developing something which could at least meet the requirement by Sony.
 
Back
Top