In terms of rendering power what would Cell be equal to?

[Titanio]

Emulation is the only way you're going to be able to compare performance of CELL to X rasterizer, unless you have your own custom software rasterizer that does the same things as X rasterizer. Since there is no software rasterizer running on a single cell, it all speculation.

There is MESA, it's just not optimised (fully) yet. That work seems to be underway though. IIRC, it could run Quake3 to one degree or another, using just the PPE (don't know if it was 30fps, but I guess it was reasonably 'realtime' judging from reporting).

As for the argument..I can see where you're coming from, but it's sort of like using multiplatform titles to compare system performance. The argument tends to be made here that you have to really look at the bespoke titles on each platform to do that, and I tend to subscribe to that argument myself. It also opens the question of having a GPU emulate a Cell renderer rather than the other way around and compare that performance too..

 

[Neb]

If you want a more immediate comparison with somewhat similar techniques, there was that julia fractals ray tracing demo (on a single cell) that beat an implementation on an overclocked 7800 GT.

I take benchmarks done by them at gametomorrow with a spoon of salt.

 

[Shifty Geezer]

That's just my opinion. Do YOU think a single CELL can fully emulate a PowerVR DC chip at full speed in software?

No, I don't think it can emulate PowerVR DC at full speed. However, that's not the question. Could Cell (single CBE chip) render a scene faster than PowerVR DC in a Cell-optimized scanline renderer (or other technique)? I don't know, but I think, gut feeling, it'd be a close call, especially depending on the scene. At 640x480 resolution (demos for Cell are all HD) and other comparable technical limits, I haven't seen any screenshots from DC that I think impossible to render on Cell.

 

[Titanio]

I take benchmarks done by them at gametomorrow with a spoon of salt.

Well, anyone can check the GPU numbers themselves using the original author's code. And the Cell version is included in the SDK so anyone can check that out too. How much more transparent do you want them to be?

 

[almighty]

If we're talking about mapping a GPU pipeline to Cell and wondering what kind of GPU it could match, that is one thing. I think that's what the OP was asking.

Thats it, maybe my wording was off and i did'nt explain myself properly. My old PC had a Geforce FX5200 AGP in it. If a game was written for the FX5200 and the same assets were programed to run on Cell, how would Cell compare performace wise? I supose what im trying to figure is just EXACTLY how much Cell can do and then use that to estimate just how much Cell can assist RSX.

 

[Titanio]

I supose what im trying to figure is just EXACTLY how much Cell can do and then use that to estimate just how much Cell can assist RSX.

On the latter, asking how well Cell could emulate a GPU's pipeline doesn't really tell you much about how Cell can help RSX. Cell helping RSX isn't typically going to involve emulation of GPU pipelines. You would play to each chip's strengths. Look at the Edge tools, they utilise Cell for what it's good at and RSX for what it is good at.

I don't think asking how Cell can emulate a particular GPU will really enlighten you much in that context. Nor would it enlighten you as to Cell's 'rendering power' either.

 

[Shifty Geezer]

The only interest in Cell's power in rendering graphics itself comes in Linux homebrew, as to what we could hope for from the open community. It has zip relevance to PS3's games or software. And of course in most cases homebrew won't have the art assets to utilize Cell fully in a Linux game anyway. Even if Cell can render a game as well as DC's Resident Evil, who'd actually create a game as complete as that? The sorts of games to look for are Geometry Wars and maybe a LocoRoco clone, not MGS2 and FF!

 

[_phil_]

Thanks Titanio for bringing this (IMO allmost stupid) thread back to earth.

 

[almighty]

So it would be relevent with Linux then PS3 itself?

 

[Neb]

Would be interesting to see a benchmark for a single Cell and also se how much the RAM limitations impact (256MB limit in linux?). Best would be to wait until someone does a single Cell benchmark.

 

[Neb]

It could sustain 7 million polys/sec doing ray casting which is similar to ray tracing.

Are there any benchmarks or test to show this, becouse if there is then it's really impressive.

 

[Ill.Skillz]

But what about Cell doing Post-effects such as motion blur or bloom effect in HDR rendering????

 

[Fafalada]

it could sustain 7 million polys/sec doing ray casting which is similar to ray tracing.

That's like saying rasterizing solid color polygons(which Cell would coincidentally be very fast at) is the same thing as rasterizing with textures and shading. They are "similar" after all.

And actually I seem to remember PVR2 HSR was not akin to raycasting, but with tile-depth buffer, might be confusing it with some other chip though.

 

[Rainbow Man]

Never get involved in politics I say.

Dreamcast fans can be deadly! Nevermind anything I said.
Peace.

 

[pjbliverpool]

As a helper to RSX I think Cell brings a fair bit to the table in terms of shader power, both vertex and pixel, perhaps an additional 6800Ultra's worth of shader power (or a lot more if used purely for vertex shading). It also adds some capabilities that are beyond a traditional DX9c GPU.

However I read the question as asking, "if Sony had used a second Cell in PS3 instead of RSX, what GPU would that have given the same results as". Or perhaps an easier to answer question, what if Cell had been used as the GPU in the original xbox? Would it have been better or worse than NV2a? How about R300 in the same scenario?

 

[archie4oz]

I almost wish they would have... It would make the system more interesting (and hey we need to keep those clipper and scanline converter skills fresh!), and then people would shut up about that myth since that's what it is...

Of course that would just engender a whole new class of stupid comparisons...

 

[sunscar]

That's like saying rasterizing solid color polygons(which Cell would coincidentally be very fast at) is the same thing as rasterizing with textures and shading. They are "similar" after all.

And actually I seem to remember PVR2 HSR was not akin to raycasting, but with tile-depth buffer, might be confusing it with some other chip though.

I think he's just really very confused - That seven million polys per second (Which I believe should actually be six) is the absolute maximum draw rate.

 

[zed]

However I read the question as asking, "if Sony had used a second Cell in PS3 instead of RSX, what GPU would that have given the same results as". Or perhaps an easier to answer question, what if Cell had been used as the GPU in the original xbox? Would it have been better or worse than NV2a? How about R300 in the same scenario?

it would be a very strange gpu, in many respects the cell is more powerful than any graphics card eg the r600 + g80 (as its more programable) but for the meat + potatoes of graphics eg drawing textured polygons in speed its not even close, prolly more similar as a geforce 1 or 2 (or even worse).
we'll see when mesa is running properly on it, + thus we can benchmark it + see how it compares to pc cards (though of course its gonna be a bit handycapped)

 

[pjbliverpool]

it would be a very strange gpu, in many respects the cell is more powerful than any graphics card eg the r600 + g80 (as its more programable)

Not quite sure how being more programmable equals being more powerful. Otherwise one could claim a 100Mhz Pentium is more powerful than R600. Its certainly more flexible, but would that flexibility allow it to render (in what ever way you like) graphics better than NV2a, or R300 etc...

I guess im interested in that question because of the early rumors that Sony actually considered using another Cell or 2 as the GPU for PS3. I think that may have been debunked though, not sure.

 

[almighty]

Not quite sure how being more programmable equals being more powerful. Otherwise one could claim a 100Mhz Pentium is more powerful than R600. Its certainly more flexible, but would that flexibility allow it to render (in what ever way you like) graphics better than NV2a, or R300 etc...

I guess im interested in that question because of the early rumors that Sony actually considered using another Cell or 2 as the GPU for PS3. I think that may have been debunked though, not sure.

From what i remeber the original plan was to include 4 Cell's in PS3.