Nerve-Damage said:A link directly to the video would be a lot of help.....
Sure:
http://www.gametrailers.com/gttvmedia.php
Its the third video down.
Could be more RSX info...
- Thread starter leechan25
- Start date
Sure:
http://www.gametrailers.com/gttvmedia.php
Its the third video down.
Thats quite a contradictary post. First you say Derek Perez didnt say RSX had the equivalent performance to a Quad SLI. Then you say PS3 will have similar graphical quality to a top end pc, which right now is a pc equiped with Quad SLI.
And your right it is absurd to think PS3 would contain 4 GPUs and no sane person here are making such wild claims. I'm saying like you said yourself the performance will be similar.
Actually much better when put it into context of a closed system with huge bandwidth benefits like the PS3.
Joe DeFuria
Legend
Xen said:
Um, Quad SLI is not a top end pc. It's an (as of yet unavailable I believe), niche of a niche of a niche PC.
Seriously folks, anyone who thinks a single chip solution in a console can pysically be even comparable to 4 top of the line GPUs with in effect 1024 bit wide memory bus....is just....naive to put it nicely.
This is not to say the PS3 GPU can't or won't be great....
"Nerve-Damage"
Regular
pjbliverpool said:
Click Video
I don’t think the dispute or argument was ever about the PS3 having Quad NVIDIA based cores (because you’re talking well over a billion some transistors alone with that design). The argument for a hypothetical RSX design revolves around one NVIDIA based IP core and four smaller cores based around Sony/NVIDIA IP previous workings (RSX around 400+ million transistors).
Something along these lines..........
RSX Specs (IMO):
* NVIDIA G80 derivative based “Main Setup Engine” @ 600MHz around 300+ Million transistors.
* Quad (Sony/Toshiba) proprietary sub-engines (83+ Million transistors combine)
* 4MB of Cache
* FlexIO Interface
* 60-87 °C (Idle/Gaming)
* 84.3W power consumption (GPU alone)
Note: One of the sub-engines is more specifically geared towards PS1/PS2 capability; however it still can be used towards PS3 based functions as well.
Edit: And for those who believe the Sony/Toshiba Visualizer was just paper based and that nothing ever materialized into a working mockup…think again!!
Last edited by a moderator:
Xen said:
No, its not contradictary, you just misunderstood it. He wasn't saying the PS3 is capable of outputting the same graphical quality as the quad sli rig, he's saying that the quad sli rig at CES was demonstrating graphics similar to PS3 level. Do you think any of the games running on that rig were stressing it at all?
And no, the PS3 won't contain 4 GPU's. Nor will it contain some magical single GPU that has the equivilent power of 4 top of the line GPU's which only six months before RSX mass production are no-where to be found in retail. Nvidia doubles the power of its GPU's per year, it doesn't quadruple it in 6 months.
p.s. what "huge bandwidth benefits"? Yes the CPU and GPU communicate a lot faster than in a PC but thats required to make up for the lack of local memory bandiwdth to the RSX. And even with the Cell and RSX memory bandwidth combined the PS3 falls behind a single GTX512, let alone 4 of them which offer similar total bandwidth to the X360's edram to a full 512MB!
Okay i understand it may not be available yet, but the point still stands in that im confident the RSX will have similar if not better performance. And Derek Perez seems to back me up.
I dont know, call me naive if you like but Ken Kuturagi has never stuck to convention when technology is concerned he's also wanted to push the envelope beyond what is currently possible to match his almost impossible desires and goals. Take the EE in playstation 2 and now Cell in playstation 3, its pretty obvious he dreams up these wild ideas and pursues the right people/companies to make it happen. I mean following current standards the cpu in ps3 would have resembled xenon or other chip maker cpu's with multiple general cores on a die. Cell breaks that mold entirely to produce unpresidented levels of performance for a single chip.
I feel the PR silence from both sony and Nvidia is just the calm before a very big storm.
I dont know, call me naive if you like but Ken Kuturagi has never stuck to convention when technology is concerned he's also wanted to push the envelope beyond what is currently possible to match his almost impossible desires and goals. Take the EE in playstation 2 and now Cell in playstation 3, its pretty obvious he dreams up these wild ideas and pursues the right people/companies to make it happen. I mean following current standards the cpu in ps3 would have resembled xenon or other chip maker cpu's with multiple general cores on a die. Cell breaks that mold entirely to produce unpresidented levels of performance for a single chip.
I feel the PR silence from both sony and Nvidia is just the calm before a very big storm.
Joe DeFuria
Legend
Xen said:
Which does not stand out as any better / worse than any other architecture of the same time frame...
Which we have no real idea how it stacks up either.
Unprecedented very specific type of performance.
Personally, I think you're just setting yourself up with unrealistic expectations.
Nerve-Damage said:
sorry, the console would eat your electric bill alive. People that go off on these wild fantasies about 300-400M transistors at high clocks sound rediculous.
They develop consoles, NOT super computers with limitless specs. They must follow a general users acceptances and guidlines while finding other ways rather then trashing the size of the console, heat output or power draw. I would hope Sony would have the sense to design all components of the PS3 to be relatively low in power draw and heat output so they dont have the consoles burning out before the heavily invested in BD drives become useful.
I'd be shocked to see the RSX break 300M to be honost. Remember its a general consumer item, thats who Sony wants to cater to. Not something for an elitist. That means heat/power/size must all be in check.
Last edited by a moderator:
Perhaps what is meant to further clarify the situation is that if a game was designed for a Quad SLI type performance bracket then you would get the same kind of visuals on PS3 because PS3 is a closed nbox environment etc. etc.
Anyway.. it is all PR rubbish and certainly some people are going to be disappointed with the un"presidented" levels of performance.
Anyway.. it is all PR rubbish and certainly some people are going to be disappointed with the un"presidented" levels of performance.
I'm not saying Nvidia can quadruple its performance every 6 months (they only wish!) but wouldnt it be sensible for a company that may have a new technology that isnt part of their current range of cards to milk the sales of the current cards. I mean why rush, they have the great performing cards available for pc's and it may present an even better boost to both Sony and Nvidia to introduce the cards within a similar time window. Increasing the profile of each.
In regards to bandwidth yeah the theorectical bandwidth of the Quad SLI is much much higher. I was really just commenting on the connection between CPU and GPU. I think qraphics of the future will not only be rated on looks but how they react. ie tree's reacting to wind, the interaction between water and objects, better more flowing hair and cloth simulation, realistic physics calculations etc. And of course with Cells ability to also perform like a GPU the relationship between the two should create some amazing images in realtime.
Xen said:
So its your belief that nvidia are capable of designing and mass producing a GPU with 4 times the power of the just launched GTX512 within 6 months? Not to metnion that this new magic GPU needs to run cool enough to go inside a console.
I have alrady explained Dereks meaning, you are merely interpreting him incorrectly to feed your unrealistic speculations.
I don't understand why Toshiba/Sony IP would end up on the GPU?
If they were going to do something custom, I think it would be more reasonable to guess on removing vertex shaders and sticking more pixel shaders. Since geometry shaders fit in between vertex and pixel shading, a GPU of just geometry shaders and pixels shaders is another possibility.
The most extreme situation I can see is
CELL = Vertex Shader
RSX = Geometry Shader + Pixel Shader
XDR2 128 bit memory pool [multithreading provides 5x the performance over GDDR-3(Plenty of bandwidth for HDR+A.A.)]
If they were going to do something custom, I think it would be more reasonable to guess on removing vertex shaders and sticking more pixel shaders. Since geometry shaders fit in between vertex and pixel shading, a GPU of just geometry shaders and pixels shaders is another possibility.
The most extreme situation I can see is
CELL = Vertex Shader
RSX = Geometry Shader + Pixel Shader
XDR2 128 bit memory pool [multithreading provides 5x the performance over GDDR-3(Plenty of bandwidth for HDR+A.A.)]
I'm not trying to get re-involved in this whole RSX "what is it?" thing, but SugarCoat I think you're putting too much emphasis on transistor counts. Obviously a number of things factor into what the power draw of a certain chip will be, and 300 million transistors is no big deal assuming low-enough clocks, low voltage, and well... whatever else. In this case we know RSX will be SOI and on 90nm to begin, so really even were it just a 'ported' G71 (which will be a large chip), I imagine the power draw would be significantly less than it's PC cousin (not to mention it'll be running at a much lower clock seemingly than the PC version).
The only logical conclusion why Sony and Toshiba went with RSX and NVIDIA is because they were not happy with their performance and feature levels. Or perhaps they wanted an environment better suited to developers (anyone who is anyone knows G70).
As to NVIDIA resting on their laurels and then KAPOW! surprising everyone with a 4x G70 design for PS3 - that is the stuff of fantasy. You rest on your laurels in the GPU IHV industry and you end up like 3dfx.
As to NVIDIA resting on their laurels and then KAPOW! surprising everyone with a 4x G70 design for PS3 - that is the stuff of fantasy. You rest on your laurels in the GPU IHV industry and you end up like 3dfx.
Tahir2 said:
If a closed box environment could produce those kinds of results then we would be seeing 9800pro+ performance from the original xbox.
pjbliverpool said:
Yes exactly and if you look at games like Ninja Gaiden an others you will see that you do get a lot more performance out of the GF4 class GPU then you would ever on an equivalent PC.
I believe John Carmack made similar remarks about XBOX being more efficient in utilising its theoretical power due to being a closed environment.
Tahir2 said:
Im not seeing anything in Ninja Gaiden or any other xbox 1 game that looks better than say HL2, Doom 3 and Farcry on a GF4Ti powered PC at playable settings.
And from the other side, im not seeing any xbox -> PC ports which are not perfectly playable on a GF4Ti at 480p with the same detail settings as the xbox.
There's no doubt xbox is more efficient because of its closed box environment but its not going to give you 400% more power!
Joe DeFuria said:
I think, had the EE not have to calculate T/L for the graphic sythesizer and just concentrate on CPU related tasks it would be pretty good for the time.
I take it you haven't kept up with Cells performance details and the released benchmarks from IBM which throw a number of tasks at Cell and it perfoms very well as far as i'm concerned. And i should imagine the "very specific performance" of cell suits PS3's needs very well.
Maybe i am setting myself up for a big fall to reality, maybe i should be more like you. Aim low and be more surprised when the truth is revealed.
G71?xbdestroya said:
