Xenos - RSX - What was left out?

jvd said:
Did too!! It's on someone's signature!!
Na the quote your talking about clearly says cpus . It says the xbox 360 cpu and cell

Furthermore, and more importantly, the point was that the comment was in relation to one specific task (software rasterisation) and not the general case as was being suggested.
 
I think the quote is on Acert's sig, Deano mentioning the processors/consoles being within 80% performance of each other. I vaguely remember the discussion biut not the context; and obviously a sound-bite in a sig loses all context.

Which raises an interesting (if overly analysed!) consideration. I wonder if Deano and other forum members should 'trademark' their names so it's not used to promote or convey an opinion they don't necessarily hold? I mean, what's to stop me having in a sig like...

Shifty Geezer
"Though a formidable machine on paper, the PS3 is likely to achieve less than 15% of peak performance under normal loads." ERP
...

Hmmm, I see a new way to spread the dark forces of FUD. I'll quickly patent the idea (of abusing other people's names) and sell it to the console companies :devilish:
 
Shifty Geezer said:
I think the quote is on Acert's sig, Deano mentioning the processors/consoles being within 80% performance of each other. I vaguely remember the discussion biut not the context; and obviously a sound-bite in a sig loses all context.

The context was indeed software rasterisation, on the CPUs.

And yes, an awful lot of stuff from here gets taken out of context and/or blown out of proportion.
 
Titanio said:
Shifty Geezer said:
I think the quote is on Acert's sig, Deano mentioning the processors/consoles being within 80% performance of each other. I vaguely remember the discussion biut not the context; and obviously a sound-bite in a sig loses all context.

The context was indeed software rasterisation, on the CPUs.

And yes, an awful lot of stuff from here gets taken out of context and/or blown out of proportion.
http://www.beyond3d.com/forum/viewtopic.php?p=527112#527112

DeanoC said:
Shifty Geezer said:
Well the fact this scene was produced on Cell has already been mentioned on this forum. Regards whether it's possible on a tri-core Xenon CPU, I wouldn't have thought so based on real-world performance of similar cores. Cell is designed for this sort of data manipulation and works more akin to a GPU then a CPU. I'm kinda curious now. I haven't seen a software renderer on a PC in ages and don't know what they can manage.

XeCPU has 3 VMX128 cores at 3.2 Ghz, at worst it could get within 30% of Cell doing the same job, because the cores of Xenon are more flexible then SPU in all likelyhood it could achieve results much closer than that.

Cell has a FLOP advantage, XeCPU has a flexibilty advantage... I suspect a tuned advanced software engine for both would be within 70-80% of each other. I'm not even sure that if you have lots of vertex and texture data, that XeCPU would lose...
 
Sis said:
Titanio said:
Shifty Geezer said:
I think the quote is on Acert's sig, Deano mentioning the processors/consoles being within 80% performance of each other. I vaguely remember the discussion biut not the context; and obviously a sound-bite in a sig loses all context.

The context was indeed software rasterisation, on the CPUs.

And yes, an awful lot of stuff from here gets taken out of context and/or blown out of proportion.
http://www.beyond3d.com/forum/viewtopic.php?p=527112#527112

DeanoC said:
Shifty Geezer said:
Well the fact this scene was produced on Cell has already been mentioned on this forum. Regards whether it's possible on a tri-core Xenon CPU, I wouldn't have thought so based on real-world performance of similar cores. Cell is designed for this sort of data manipulation and works more akin to a GPU then a CPU. I'm kinda curious now. I haven't seen a software renderer on a PC in ages and don't know what they can manage.

XeCPU has 3 VMX128 cores at 3.2 Ghz, at worst it could get within 30% of Cell doing the same job, because the cores of Xenon are more flexible then SPU in all likelyhood it could achieve results much closer than that.

Cell has a FLOP advantage, XeCPU has a flexibilty advantage... I suspect a tuned advanced software engine for both would be within 70-80% of each other. I'm not even sure that if you have lots of vertex and texture data, that XeCPU would lose...

Right. He's clearly discussing one task alone, a software graphics engine on the CPU - IIRC it was all in the context of a discussion on The Getaway demo and how it might have been done on Cell without help from RSX.
 
Yes, seems so. Though if you look at Deano's exact wording it's quite conceivable that after talking about software renderers, he went on to say that regards any software engine the two are 70-80%

Cell has a FLOP advantage, XeCPU has a flexibilty advantage... I suspect a tuned advanced software engine for both would be within 70-80% of each other. I'm not even sure that if you have lots of vertex and texture data, that XeCPU would lose...
Did Deano continue the idea of just software renderers, or did he leap forwards with a more generalized statement of tuned optimised engines for any system, be it AI, physics or rendering? We can be sure of the former now, but it wasn't necessarily clear when first posted.

Replace 'engine' with 'renderer' and all ambiguity is cleared up (quick Deano - edit your post and deny everything! ;) )
 
I'm after all just human :) I often called software rasterisors - software engines.
Goes back to when PC games had two fairly seperate engines, one for fancy 3DFX cards and one for the rest of the world.
 
Deano i must ask you one question about the the team your working in and what you said earlier.
Are the HS engine at this stage using the SPE´s or are you soley doing the current work on the the PPE alone?
 
DeanoC said:
I'm after all just human :) I often called software rasterisors - software engines.
Goes back to when PC games had two fairly seperate engines, one for fancy 3DFX cards and one for the rest of the world.

Well why dont you just finish the doubts. Is overall Cell much more powerfull than Xenon, or are they pretty close in game PErformance?

I dont think you would be breaking any NDA.
 
Well why dont you just finish the doubts. Is overall Cell much more powerfull than Xenon, or are they pretty close in game PErformance?

I don't think there is a general answer to this question that will fit in all circumstances.

What does "game performance" mean? Deano cannot know the behaviour and performance characteristics of every specific game beyond his own. Perhaps he might be aware of "typical" behaviour and draw something from that, but..

It's a very complex thing. Different games offer different mixes of tasks and different priorities. Some of those might mesh better on Cell, some might mesh better on Xenon. Then there's the question of how much better things run, and whether it's significant. What tasks take most time. Whether one chip has as clear an advantage in its "strong areas" as the other. Straight up, you could write code that would run better on Xenon than on Cell. No doubt. And vice versa. But you need to think about what that code is doing and what you yourself place importance on. I guess a better place to start with questions like these is to ask about specific game features and how they might compare on Xenon and Cell and determine which you prefer and which are more important to you. And so then you can determine as far as your own priorities are concerned which might be the better chip for games. Basically try and build a profile of what games might look like on Cell or Xenon with either pushed to its full potential.

This isn't really a cop-out, and I'm not saying the chips are necessarily by any means equally powerful or that the whole thing is a wash. Saying they both have areas of strengths doesn't make them equal. Even last gen the CPUs in all the different systems had at least some advantages over the others. It's just ultimately, some advantages make a more dramatic difference than others, some advantages might be larger than those offered on other chips and more than make up for relative shortcomings elsewhere, and thus we may come to generally think of one chip as "overall" being more powerful. I think we will come to that conclusion, but it might take some time since there's little developer comment to ponder currently.

I more than welcome Deano's comments, of course. Questions kind of need to be asked on a number of levels IMO..it'd be wonderful if he and others could entertain them.
 
Shifty Geezer said:
Yes, seems so. Though if you look at Deano's exact wording it's quite conceivable that after talking about software renderers, he went on to say that regards any software engine the two are 70-80%

Cell has a FLOP advantage, XeCPU has a flexibilty advantage... I suspect a tuned advanced software engine for both would be within 70-80% of each other. I'm not even sure that if you have lots of vertex and texture data, that XeCPU would lose...
Did Deano continue the idea of just software renderers, or did he leap forwards with a more generalized statement of tuned optimised engines for any system, be it AI, physics or rendering? We can be sure of the former now, but it wasn't necessarily clear when first posted.

Replace 'engine' with 'renderer' and all ambiguity is cleared up (quick Deano - edit your post and deny everything! ;) )
To be honest, I'm disappointed. I've been using this simplification to think about the Xbox 360 and the PS3 as a whole, since it seems to hold up whenever anyone talks about the respective systems. To hear that I have misread the situation means that I can't easily abstract the two architectures into two simple buckets: flexible and powerful.

Ah well, just when I start to grok something, someone comes along and yanks the rug out from under me. ;)

.Sis
 
therealskywolf said:
Well why dont you just finish the doubts. Is overall Cell much more powerfull than Xenon, or are they pretty close in game PErformance?

I dont think you would be breaking any NDA.
Deano isn't working on XB360, so even if he were allowed to talk he won't have any insight. Plus as a developer it'd be unwise to talk about x system being better than y system as he doesn't know which of those systems is gonna be paying his wages for the next few years.
 
Tap In said:
Jaws said:
This is an interesting quote from that link that everyone seems to be ignoring,

DeanoC said:
...Now if we want to talk about procedural synthesis that creates procedural textures, than Cell is gonna whip XeCPU into touch in most cases...

http://www.beyond3d.com/forum/viewtopic.php?p=527210#527210

Interesting. ;)

not so fast... ;)


http://arstechnica.com/articles/paedia/cpu/xbox360-1.ars/1

Read that article. Posted in that thread. Procedurally generated graphics is nothing new in games and was touted even by the PS2 and it's Emotion Engine, so there's a sense of irony there... ;)
 
I read this whole thread and all I learned was that Deano Cleaver is stupid and Dean Calver is smart.
Was there anything else to learn from that thread, or that's all there is to be found here?

:LOL:
 
Jaws said:
Read that article. Posted in that thread. Procedurally generated graphics is nothing new in games and was touted even by the PS2 and it's Emotion Engine, so there's a sense of irony there... ;)

I was sure you had read it, but figured it might be okay to make another go around for those who had missed it since it was brought up. ;) :)


I'm no programmer but if I understand correctly, I'd like the idea for reducing bandwidth use that otherwise would be needed for moving textures. Is that right?

Kind of why I always liked the PowerVR and the Dreamcast (tile rendering) as anything that reduces the need for brute force is good in my book. :)

In one of the MS interviews a rep was talking about procedural synthesis as though it was going to be more widely used in the next gen. It will be interesting to see how it's used. If all of these BW saving techniques are employed, we could see some amazing achievements in games toward the last couple of years in these systems.
 
Shifty Geezer said:
therealskywolf said:
Well why dont you just finish the doubts. Is overall Cell much more powerfull than Xenon, or are they pretty close in game PErformance?

I dont think you would be breaking any NDA.
Deano isn't working on XB360, so even if he were allowed to talk he won't have any insight. Plus as a developer it'd be unwise to talk about x system being better than y system as he doesn't know which of those systems is gonna be paying his wages for the next few years.

Anyway, that is the reason why we came here.

But it will take a long time till we know the answer ...

If at least we knew the answer for AI and physics... :LOL: (just kiding)
 
Back
Top