Nvidia vs AMD

Who makes the best video drivers for increasing FPS in games?

 

AMD and Nvidia are two companies whose names are commonly thrown around when talking about video drivers. You might want to look into one of them. I'm sure either company can provide you with suitable drivers.

 

Parsing error: Please reformulate question.

FPS are not the only thing to look at. Everything is a quality/performance tradeoff. And you can't divorce the hardware from the drivers. Most anyone would take a Radeon 5870 over a GeForce 220, or a GeForce 260 over a Radeon 5450 regardless of drivers, assuming game performance is the goal.

Really, the question you ask is silly. If you want to play games with PhysX Nvidia is the way to go. Right now AMD provides the best performance/watt from what I've seen, but things may change soon. Both play games quite well. Don't make your choice on drivers... make the choice on the hardware you can afford that does what you want. The drivers only have minor performance effects after that.

 

It all depends if the games you are going to play are optimized better on the Green or the Red's card.

 

Nvidia is worse because their chips make you back in time
/sarcasm

They are competitors, usually what is best is the top of the newest line. Theres not much more to it.

 

PhysX on Nvidia hardware is just a marketing gimmick. One of many physics engines and it runs just as well on a CPU as it does on a Nvidia GPU. If your CPU is the bottleneck in a game that uses PhysX and you are running a Nvidia GPU, it might give you a few more FPS tops. On the other hand, if you enable the GPU to run PhysX and the GPU is the bottleneck, it can take away a couple FPS. Otherwise it just nonsense.

 

Holy moly, you're right. PhysX has changed a lot in the past two years. It used to be a completely worthless gimmick, but now wow, in the few titles that utilize it, it seems to be a ploy to make ATI look bad.

 

In any case PhysX should not be the deciding factor between GPUs due the extremely short list of games using it.

Back on topic: the OP should clarify better what he wants to know. But by reputation many people say Nvidia drivers are better Ati, although I am not sure if that just means that Nvidia releases more frequently drivers than Ati.

 

I think a lot of that has to do with Linux support. Nvidia has traditionally been way better with Linux support than ATI/AMD has.

 

I used to think that too, but I did some research:

PhysX: An easy target? | Scali's blog
http://www.xtremesystems.org/forums/showpost.php?p=4562898&postcount=56

PhysX is likely actually faster on the GPU. SSE and SIMD really isn't as powerful as you would think. True, it's not optimized fully for CPU usage but you're talking about maybe a 20% performance gain in physics performance, whereas CPU PhysX vs GPU PhysX is showing much larger differences than that that kind of optimization would gain.

Now, as to whether you could accelerate PhysX on an AMD GPU to the same degree as you do on an Nvidia GPU is up in the air, but it's not just smoke and mirrors. It's one of those things that Nvidia has to differentiate it... they've paid game developers a lot of money to use PhysX and spent a lot of time helping them implement it properly so they can sell more cards. I'm still an AMD/ATI guy because they haven't pulled the same crap Nvidia has with lying about the chip bumps and such, but I'm not gonna begrudge them the facts.

 

as far as driver support goes, i have never really had an issue with either company providing drivers that negatively impact my computer's performance in such a way that i view the drivers as trash.

Nvidia has great driver support, but recently ATI has been working well to provide new drivers every month.

i feel performance only goes so far, especially since both companies seem to neglect older hardware in their newer drivers.

 

I have used both companies over the past year, and both have had driver support that was stable and quick. I would give the edge to Nvidia on the driver side as of now, only slightly.

 

Ok... so CPU and GPU calculation loads are different. Hardware accelerated PhysX deals with more particles and effects and yet CPU PhysX is still slower. Even optimizing it, the CPU dealing with fewer particles is still slower than the GPU accelerating more particles. There's no real credible source that can do ANY kind of physics simulation on the CPU (PhysX or not) as fast as the GPU can.

If you look at the Bullet physics whitepaper here (pages 43 and on) you can see almost an order of magnitude performance increase by using the GPU instead of the CPU in some tests doing the exact same work, and it's twice as fast at minimum just using generalized OpenCL techniques: http://www.nvidia.com/content/GTC/documents/1077_GTC09.pdf

I'm not really sure what you're saying "No" to? The facts are quite plain that GPU acceleration is quite a good thing for physics simulation and it is faster than a CPU can do it, even with SSE and other optimizations of the code.

 

That Youtube demonstration has no substance. There are no machine specs given, no comparison to GPU acceleration of the same effects.

I'm not saying the CPU cannot to physics. I'm saying that physics is a very parallel problem that is an ideal task for a GPU. A CPU can create beautiful rendering of images, more detailed than any GPU can. But it's not as fast at it. Physics (especially game physics where precision is not nearly as important as speed) is in the exact same domain of problems.

 

This topic is not going anywhere. Original poster, you're going to have to ask a more pointed question. What you asked doesn't make much sense.