exactly! (though is not always possible/feasible...
)
Are PS3 devs using the two mem pools for textures?
- Thread starter Love_In_Rio
- Start date
)From your comments above, it seems MS/ATI made some smart moves using the Unified Shaders architect with eDRAM in reducing the transistors and gained some (almost) freebies.
That said, there are always trade-offs and this is one side I've not heard/read much of (or found using search/google). All I can recall is for the inital target resolution 720p, Xenos is a great choice in a closed setup. Moving beyond 720p, predicated tiling is required thus gaming engines needs to be built from the ground up for it (more work for the coders).
Other then the re-think for the game engine designers (covered many times over), from a h/w GPU design point of view what were some of the sacrifices or limitations of Xenos in a broader sense (dont want any of your guys breaking any NDAs
) comparable to the many GPUs out in the wild. In other words what are some of the 'potential' shortcomings of Xenos at this given time. I say potential as it might not be a limitation in the future who knows.Yes, very good point that Unified Architecture reduces redundancy, but there will still be overhead in both performance and transistor counts in the scheduling and/or load-balancing department.
Now in a closed box like PS3 if RSX is'nt being fully utilised whos fault is that?
Its nobody's fault, its just that you don't need to vector shade stuff at all times for example.
As soon as you don't use it, those pipes are just idling.
archangelmorph
Veteran
vector shade?
But you DO need to transform many many vertices especially if you want to do alot of animation or generally alot of 3D transforms..
At the end of the day the more you can plop into your vertex shaders then the more CPU cycles you can free up for other/non-vertex processing tasks..
Exact
The trade-off is always easy of use/program vs power! Having the hardward do all the heavy lifting for you will not challenge the talented coders.
Also the good ones (apply to all professions) will find ways to get the most from the environment or platform they work on.
And if there idleing its the devs fault, clever devs ALWAYS make sure every bit of the hardware is always used. If i were making a game and i had idleing vertex shaders i would find somewere to use them.
I guess there aren't any clever devs then ...
Thats a rather silly statment imo, I don't know a great deal about game dev. However I do know that its not just skill, and tech that need to be used there are other resources like time and budget that need to be met. At the end of the day Devs' need to keep the business in mind as well.
You could also say that Dev's should be using all the Cell's potential to, however when you consider the time it would take to do the research and the money to fund that research it's just not practical to push any hardware to the limit especally at day one.
Lastly I am a musician. I have a pro level keyboard and I have had it for about 9 months now. I know that I really am bearly scrating the surface of it. However that doesn't mean that I need to know or use all its functions to get a great piece of music out of it. Same with devs and hardware.
In exactly the same way every time I make up more music using the keyboard I might learn more about it while completing a set task for myself (a single song for example) this knowledge bas can then be stored for the next track. This again is very similar to the way that Devs would work with their skills and hardware to.
game = hardware/software(including 3rd party software engines etc/in house engines/art packages etc)+talent(including artistry/programing/creativity)+resourcse(including amount of time/budget/amount of people etc.).
Then going by that, there is no clever devs, in the entire world.
Just using vertex shaders for the sake of using it, is silly. There are places where you don't need it, and if you don't need it, you don't use it.
What your saying would kinda be like driving a car always close to the redline, because that utilizes more of the engine. And there is little point in doing that, seing as you just use more gas, and make more noise. A better result, would be if you drove it normally, and floor the throttle whenever you need to accelerate really fast etc.
If you don't need you simply don't use it as I don't think ppl code shaders full of nops just for sake of having VS doing something.
The interesting thing is that sometime you think you don't need it.. but maybe you do, as you can implement some feature at small or no extra cost, or you can shift predictable workload from PS to VS.
I agree that an unified approach is better (well, I had been advocating it for so many years..) but not having it is not an excuse to not try to use all the resources available, even when it seems impossible to use them.
IMO you need to keep in mind that although Bizarre Creations have done some amazing looking racing games, they have been doing it on really strong hardware for their times, Dreamcast PVR2, XBox Nv2A and now X360.
Polyphony Digital have two GTs on PS1, two more on PS2 and now the PS3 gives them so much more hardware spec than what they were able to use at the time with the Cell, RSX and Blue Ray
PD always improved their GT games on the previous hardware so now that they have all this inside the PS3 to work with and given some time for them to implement it its highly likely that GT5 and later will look amazing in the most technical of terms.
I don't see how this relates at all.
In fact, PD is at an advantage having worked with the weaker, and wierder systems (xbox 1 vs PS2 architecture). They have more experience exploiting systems that required a lot of self learning and tweaking.
If the PS3 was just a super tuned up PS2, polyphony would know pretty much every trick there is to know in an instant. Having worked with a more powerful system before doesn't mean anything.
PC developers are traditionally the developers that do the worst on wierd console architectures, mostly because they aren't used with working without the straightforward tools that the PC brings, even tho they are allways developing for the most advanced hardware.
PD are expert in creating fake graphical effects (i.e. 2D trees in the distance that look like 3D trees) making simple things look complex.
Great simulation isn't about making simple things look complex, but making complex things look simple and natural. That's where they fail with almost everything related to simulation (real braking distances, excessive grip, etc).
Talent and money are two different things. They sure are talented with graphics but they're very untalented when it comes to other factors.
As for the money, undoubtedly there are instances where a lack of budget gives rise to creativity that would not have been necessary had the producers been able to throw money at a problem, which can in turn result in a stronger overall game (not gaming wise, a real life example of this, that classy and historical movie -Jaws-, is the best one that I can think of).
Your statement is an example as clear a sign of hype as you'll ever find: assuming that because a game has a budget, it's devoid of creativity. The paradox, indeed, is that in making such a statement, people are tacitly admitting to an inability to recognize creativity (or a lack thereof) on their own.
Last edited by a moderator:
I can't comment on the realism of GT, but all this talk made me remember an interview with a racing driver from a few years ago.
http://www.firingsquad.com/features/bobearlracinggt3/
He seemed quite impressed with it.
Similar threads
- Locked
