What is the true spec of PSX3 CELL???

You can run it in single precision.

Mainly because it is much more scalable. IMO all computationally significant tasks in a game engine can be parallelized well, which can not be said for SpecFP in it's entirety.
 
Re: ...

bbot said:
Someone's response:

Intels "Billion Transistor Processor" is supposed to be the dual core
Montesito to be released in 2005. All those transistors go for more
then 90% into SRAM cache memory and NOT into floating point processing
hardware. Montesino at 90 nm will be huge ~480 mm2, ~20 GigaFlops
(double precision) and 130 Watt or so TDP. They want it to be the
record holder for SPEC floating point performance.

Responce to nameless, one armed man:

http://www.eetimes.com/semi/news/OEG20021015S0036

But yes, my post clearly said that they intended to have 16MB of L2/3 cache on the die, thus the rest just follows.
 
bbot said:
How is this supposed to be a reply? You're simply reasserting what you've said before.

I was just providing a link if you wish to read the article. I don't need to respond to that anyways.
 
I would base my guess on Figure 9 @ 1GHz.
But there is a lot yet to learn. Such as I think the patents reflect 65nm.

I believe the PS3 will have these three things.
1. At least 50GB/sec off chip memory performance.
2. Start production in a 90nm process.
3. At least 16MB of eDRAM.

Here is a link to my most recent guesses.
I'm still undecided about 64bit processing.
All the other clues say yes. In fact they say MIPS64 LSI, not cell based.

Either way here's the link.
http://forum.pcvsconsole.com/viewthread.php?tid=6653
(I know a little something about a Blu-ray console in 2Q 2005 & 1Q 2006, PSX`2 OR PS3?)
 
Back
Top