nAo said:
A supposedly efficient unified architecture should show its strengths with far less effort..so I believe it should be the other way around
Ok, I will bite: What does unified shaders and the general strengths of the approach have to do with tiling hurdles or developers still locked in on SM2.0 (and below) shading models or the low amount of math to texturing typical of current game design?
Having more advanced featureset and better shader utilization or fillrate doesn't help you one bit if you are texture limited, are deploying unrefined middleware, fighting the CPU and multithreading, or just plain porting a game that is ill-conceived for the target platform given the time allocated for the conversion.
To be so blunt, if we swap what you are saying across the board we could say that Cell's supposedly processing edge should show its strength with less effort... so I believe it is the other way around. (Not that I do of course!)
But that doesn't really say much for the multitude of difficulties developers are facing in very finite time budgets. Of course you could probably lecture me on these points much more eloquently :smile:
On Xenos, if you are not shader limited in your software all the shading power in the world won't save you and give you more performance. I don't believe there is a very good arguement that RSX has a richer featureset or that RSX could not benefit in many ways by using eDRAM like Xenos (yet duly noting the reservations and limitations of the eDRAM in Xenos, but there is no denying the bandwidth it provides for its designated tasks).
Bobler said:
Megadrive said:
but I think as time goes on, developers will show Xenos' strengths while RSX will show its limitations, IMO
What an absurd thing to say.
What an absurd rebutal!
Ok, tongue in cheek comments out of the way, Megadrive would have been best probably giving some reasons why... although to his benefit he has stated them before and they are well known. Reading his mind I believe he probably had the following in view...
Xenos has a much broader and usable featureset, significantly faster dynamic branching in pixel shaders, much higher sustainable fillrate, easy to access features (like FP10) that give an immediate performance benefit over other hardware methods, and so forth.
Just looking at ATI's X1000 series and Xenos and where D3D10 is headed, and how RSX was implimented into the PS3, I can see a number of areas where RSX is going to show limitations in some future 3D game engines. RSX may be an SM3.0 GPU, but is has very poor dynamic branching in pixel shaders and its vertex texturing is very slow.
IMO, It is an SM2.0 GPU with SM3.0 features stapled on. It really doesn't do any core SM3.0 feature exceedingly well. Further, RSX has a split memory pool, is facing texturing penalties from XDR, has half the fillrate a typical GPU with its texturing and shading power possesses, has a significant performance penalty for FP16 blending and filtering, and while it has a paper flops advantage it has less effecient ALU utilization and also ties up shader ALUs when texturing.
Not that Xenos is perfect--not by a long shot. Tiling has been an issue (too small for 720p 4xMSAA, requires a renderer rewrite if there is not an early Z pass, predicated tiling support was not really resolved until Spring 2006, memexport requires a 3x load on the CPU and was not compatible with tiling, etc) and MS did not resolve it very quickly, MS did not get SM3.0 capable GPUs into developer hands until nearly the consoles launch, the dev kit situation was fubared and illconceived, Xenos is looking 2 years down the road and has a significant amount of transistor budget allocated in a way that doesn't help a lot of current game engines, has less peak texel fillrate, etc.
There are hurdles on both sides. But I agree with Megadrive: When developers begin developing SM3.0-based engines and begin exploring basic D3D10 features there is going to be decided overlap with the core focus of Xenos.
Of course I can see the arguement now: But the PS3 can offload work to Cell. Which is true (of both consoles), but the topic is the GPUs, not the systems.
spdistro said:
i believe when you have unified architecture with unused features like memexport and an API which is inbetween dx9 and dx10 still in its infancy compared to an already established GPU related API of Nvidia, it would be easier for devs to show with Nvidia thier strengths eaarly and the ones with ATI Xenos, theier strengths later :smile:
And as I mentioned earlier RSX is a much better understood -- and developed for -- architexture. The seminal aspects of RSX were introduced in the middle of 2004 via NV40. G70 is a mere refinement of the NV40 base (some bulking of the shaders, refinement of the workflow and features, and of course faster) and RSX is a direct derivative of G7x.
Sony developers have had NV40 SLI since early 2005. That is an obvious advantage to getting Xenos in July/Aug 2005 and going gold in October 2005. And MS dropped the ball on a number of issues, e.g. predicated tiling received a number of fixes in the API in March/April 2006. Likewise we have seen the problems associated with trying to shoehorn PC game engines into Xenos, which is not really unexpected as it is not a direct descendant of a PC GPU like RSX and NV2A were. Heck, you cannot even begin using tiling unless your engine does an early Z pass. Xenos has a lot of power, but you have to tailor your game to it. It is very much like the current CPU situation: They have a lot of power, but you MUST use multiple cores if you wish to extract that power.
Tossing a single threaded PC game engine on Cell/Xenon is NOT going to get the results you want. Ditto Xenos. RSX, to Sony's credit, is in a MUCH better situation to port existing 3D libraries and to hit the road running. It may be the least progressive part in all the consoles, but it is also the part most well understood and ready to be exploited--today.
Of course many of us stand on different sides of the fense on these issues, even developers do. Some developers have more resources and essentially have first party status and have the luxury of figuring this stuff out, others have to make do with their purchased 3rd party middleware and get the best out of it in 18 months and move on.
Oh well, we have all ping ponged these things back and forth a million times.
Some people feel the Xenon-PC approach to multi-core, some feel Cell is. Some feel Xenos is too future looking and has too many limitations on the eDRAM, others feel RSX is an outdated model that will show its age and Xenos is struggling due to system and last gen porting "first one out" issues like the PS2.
To degree most will be right and wrong.
Ps- Boy this thread went from ugly... to nasty! This is why I avoid posting here! eeeeek!
