Sony to cut semiconductor capital expenditure; 45nm Cell in FY08/09

Was there ever any proof to the affect that the PS3 GPU, the one sony was planning before RSX, was not cell based? I will admit I fell behind on things but from everything I saw a Visualizer GPU made most sense and from patent diagrams stemming from circa 01-02 it seems that's what Sony had in mind.

Proof is in the dev telling you.
 
Proof is in the dev telling you.

No wonder nAo is so far ahead of the other ps3 devs gpu technology. :oops: The creator knows his creation better than anyone.

:p

Seriously from the early patents I've seen it looked to be the case that cell would handle all graphics operations. They could have been tools of deception though to throw off the competition. From what I recall 4 cell's were to be used in ps3.
 
No wonder nAo is so far ahead of the other ps3 devs gpu technology. :oops: The creator knows his creation better than anyone.

:p

Seriously from the early patents I've seen it looked to be the case that cell would handle all graphics operations. They could have been tools of deception though to throw off the competition. From what I recall 4 cell's were to be used in ps3.

It could be neither - in engineering, its quite often that you have more than one solution for a problem. In today's world, you might as well patent as soon as possible to avoid the leeches that live on patents rather than implementation from getting there first.
 
It could be neither - in engineering, its quite often that you have more than one solution for a problem. In today's world, you might as well patent as soon as possible to avoid the leeches that live on patents rather than implementation from getting there first.

Good point.

Wasn't there a demo at e3 2005 that ran solely on cell (2 cells?)?
 
I'm not sure, but I believe the game called "The Getaway" was done purely on CELL, but only 1. Maybe you're talking about something else.



Other than that, there is the Terrain Rendering Engine that uses the CELL to power real-time Ray-casting at 1080p

Yeah I think it was the Getaway but I thought they said it was running on two cells. Anyway, doesn't this gpu-less demonstration go against the notion of ps3 conceptually designed with a gpu in mind?

Sure it also says "hey look how powerfull cell is. It can render this all by itself", but would it not be better to show what cell could be doing/improving in real ps3 situations?

(Such as physics, colission detection, etc.)
 
nAo, we have to let the truth out... the previous GPU was a cluster of 256 PSOne GPU's with an array of 1024x8-bits Z80's (yeah, Kutaragi wanted the second pair of registers ;)) with embedded DRAM to do z-buffering (we know that PSOne GPU does not support Z-buffering).

Finally, I had to get it off my chest...

;).
 
Yeah I think it was the Getaway but I thought they said it was running on two cells. Anyway, doesn't this gpu-less demonstration go against the notion of ps3 conceptually designed with a gpu in mind?

Sure it also says "hey look how powerfull cell is. It can render this all by itself", but would it not be better to show what cell could be doing/improving in real ps3 situations?

(Such as physics, colission detection, etc.)

I'm not a marketing person so I have no idea which is more impressive, or which has a bigger effect vs. the time needed to implement either demonstration.

All I know is that looking back at it now, it is still extremely impressive that the CELL can actually render this kind of thing. I managed to find a 60fps video from Kikizo here http://www.kikizo.com/viewer/vidvie...ps3_sep05/kikizo_hd_ps3_thegetaway3_sep05.wmv.
 
I'm not a marketing person so I have no idea which is more impressive, or which has a bigger effect vs. the time needed to implement either demonstration.

All I know is that looking back at it now, it is still extremely impressive that the CELL can actually render this kind of thing. I managed to find a 60fps video from Kikizo here http://www.kikizo.com/viewer/vidvie...ps3_sep05/kikizo_hd_ps3_thegetaway3_sep05.wmv.

Very impressive indeed.

But if rendering on cell was not in the cards to begin with, why demonstrate the ability? Why not show where the cpu would really outshine the competition and demo the same scene rendered on a gpu and have the cell work on crazy levels of interactive destruction(physics, collission) us gamers love so much.

Just seems "odd" to me that all these coincidences point to one thing, yet it is dismissed.
 
But if rendering on cell was not in the cards to begin with, why demonstrate the ability? Why not show where the cpu would really outshine the competition and demo the same scene rendered on a gpu and have the cell work on crazy levels of interactive destruction(physics, collission) us gamers love so much.

Just seems "odd" to me that all these coincidences point to one thing, yet it is dismissed.

That was the point I was highlighting earlier - they are likely to demonstrate something that wows more per time spent. Perhaps the graphics was easier to do perhaps not. Now we have Motorstorm and Havok 4.5 demonstrating that it is indeed a beast at physics. In 2005 they chose to show off CELL outshining the competition's CPU based on its graphical prowess. In the end, theres no magical formula to marketing and PR so what happens just happens. I think its best not to try to look too much into it.

Also, as demonstrated by recent trends, the relationship between graphics and physics is closer than you are giving credit. Just because CELL is good at graphics doesn't mean it was "built" for it and vice vera. There is a whole big area of computing that shares intrinsic links.
 
Very impressive indeed.

But if rendering on cell was not in the cards to begin with, why demonstrate the ability?
The purpose of the demo, which everyone at the time missed, was to illustrate the 'living world' created by Cell. Listen to Harrison's commentary.
Why not show where the cpu would really outshine the competition and demo the same scene rendered on a gpu and have the cell work on crazy levels of interactive destruction(physics, collission) us gamers love so much.
If they had no interface with the GPU, or were working on the graphics rendering during Cell software development for research purposes, there'd be reason to render on Cell. Furthermore the question of how much was rendered on Cell was discuessed on this board. Can't remember the outcome, but I think GPU assistance was there. You'd have to search.
Just seems "odd" to me that all these coincidences point to one thing, yet it is dismissed.
The coincidences are Cell rendering graphics, which you do when you want to show off the power of a device over a visual medium; and a patent, where patents are filed willy-nilly and filing of a patent does not mean you intend to build the product in that patent. The rest was speculation on forums. Oh, and the patents were for a Visualizer, and not rendering on Cell.

The idea of a CPU that would do both program and graphics is pretty cool, in that it gives a one-stop solution to everything, and is ultimately flexible. Where we're looking a t afuture with multicore CPUs and multicore GPUs and CPUs runnig graphics code and GPUs doing GP work, a single core solution that can turn its hand to everything woudl simplify things nicely. It's not happening this time though, and was never on the cards. Sony (or STI) were aware that a Cell cannot compete with a GPU in rendering pixels.
 
Very impressive indeed.

But if rendering on cell was not in the cards to begin with, why demonstrate the ability? Why not show where the cpu would really outshine the competition and demo the same scene rendered on a gpu and have the cell work on crazy levels of interactive destruction(physics, collission) us gamers love so much.

Just seems "odd" to me that all these coincidences point to one thing, yet it is dismissed.

They did various things to "show where Cell would really outshine the competition". Remember Toshiba's Magic Mirror, Mercury's MRI and "Eye of Judgement" ? The rendering demos may be an attempt to address the offline rendering needs (more scientific visualization), rather than using Cell as the PS3 GPU ?
 
http://www.beyond3d.com/forum/showpost.php?p=932089&postcount=28

Kutaragi said:
One of our ideas was to equip two Cell chips and to use one as a GPU, but we concluded that there were differences between the Cell to be used as a computer chip and as a shader, since a shader should be graphics-specific.

So this was obviously in the cards.

The original patents, the Toshiba "RS" modified cell (correct me if I'm wrong), the graphic demos on cell at e3, the late signing of partner nVidia, and nVidia's end product (RSX) seeming to be an off the shelf design, two different ram types/busses, all support the concept of a gpu-less ps3 design initially.

Question is, when did they switch gears?

Unless I'm way off base and someone can prove otherwise? :oops:
 
http://www.beyond3d.com/forum/showpost.php?p=932089&postcount=28



So this was obviously in the cards.

The original patents, the Toshiba "RS" modified cell (correct me if I'm wrong), the graphic demos on cell at e3, the late signing of partner nVidia, and nVidia's end product (RSX) seeming to be an off the shelf design, two different ram types/busses, all support the concept of a gpu-less ps3 design initially.

Read Lazy8s's post again, he said "while an alternate GPU design for PS3, the RS, did come from Toshiba", he did not say RS is based on Cell.

Also Kutaragi mentioned that he chucked the idea of a Cell GPU, he did not say he canned any Cell GPU prototypes. So there is no proof yet for a Cell based GPU.
 
All I know is that looking back at it now, it is still extremely impressive that the CELL can actually render this kind of thing. I managed to find a 60fps video from Kikizo here http://www.kikizo.com/viewer/vidvie...ps3_sep05/kikizo_hd_ps3_thegetaway3_sep05.wmv.
I still don't believe that was done on Cell. It really seems like a miscommunication to me from the interviews I've read. Cell was probably responsible for some HDR post-processing and that's it. A full-fledged software renderer optimized to run on Cell at high efficiency seems like such a huge waste of time.

I thought it was already confirmed that this was the case, but I guess not.
 
Good point.

Wasn't there a demo at e3 2005 that ran solely on cell (2 cells?)?

Maybe you were thinking about this (my favourite e3-05 Harrison interview):
The Doc Ock head - the Alfred Molina head - is actually more of a Cell demo than it is a graphics demo, because we're calculating hugely complicated light sources in real-time on the Cell, even to the point where we calculate the angle at which light enters the skin, the way that the light is then coloured by your blood, and the way that it is then reflected back out. It's something called transmission. Skin is hugely complicated - if I put my finger over a light, for example, you can see that the light is coming through my skin. We were simulating that - emulating, simulating, kind of a fine line - we were simulating that on the Doc Ock head demo.
Though I don't think they ever mentioned any 2 Cell setup.

@Mintmaster: According to this Harrison quote, it seems lke you are right that the Getaway demo was not entirely rendered by Cell.
You're right; obviously Cell allows you to do complex collisions, physics, dynamics, simulations, all of those things. Though, the Getaway demo was a good example of how you can have a living city brought to life as a result. Although it was pretty graphics, most of that power was actually Cell-based.
 
Back
Top