Having a look at benchmarks always helps. Sometimes the cold hard specs can be difficult to interpret, but real life performance never lies
I would advise this too.
As for Nvidia...
Nvidia has one of the worlds worst naming schemes ever, or at least they did (Nvidia has made it a habit to basically rename old graphics cards, so be wary).
For Nvidia I'm gonna leave out the 9xxx series out. It's you already have a 9600gt, so mentioning a 9xxx would be useless, and Nvidia has basically restarted the naming scheme (if they continued on and call the graphics cards after the 9xxx series then you would be looking at the 10 xxx series and so on, which would be difficult to say, not as catchy, and well, it would be annoying)
So, to what you want.
1)the shaders, just like AMD/ATI the under of shader processors on these cards has a direct correlation to performance. So the more shaders the graphics card has, the more zoom zoom. HOWEVER, don't go an make the mistake of thinking that AMD shaders=Nvidia shaders, na-ah, these are two different breeds of processors.
To make sense of it all, (if we assume that the AMD and Nvidia cards produce the same performance), think of the AMD as a v12, it has more cylinders, and produces 400kw of power. Mean while the Nvidia is a big V8 that also produces 400kw of power. So you can't compare apples with apples, because you dealing with lemons and oranges.
2)the memory. This is simple, the more bandwidth the card has, the better. However, unlike AMD, Nvidia doesn't use ddr5, they use a much bigger bus lan. The result is the same speed, but just a different way to do it. So what you looking for is a big bus width (I'm not clued up on the Nvidia bus widths, I think it's about 448bit-bus to 512bit-bus, just go for the bigger one). What you really want here is the Higher bandwidth.
3)Clock speeds, this is how many times the processors, well processes. So the higher the core clock speed, the faster the processor, but it's the same as the AMD, the number of shaders takes preference over the clock speed, when you dealing with the same number of shaders, then you can compare the clock speed.
4)Nvidia's new naming scheme is a little easier to follow then their previous one. Basically, you have you series (or generation) number up at front (for eg: the GTX280, the "2" means it's the 2nd generation). So the gt200 series means the second series (you might be wondering where the 1st series was, well, the 8xxx and 9xxx series were the 1st series). Then comes the 2nd number, which means how fast the card will perform, which goes in counting order, so 4 performs worse then 5, and 5 performs worse the 6 and so on (eg: GTX 280, the "8" signifies it's fast).
The 3rd number shows that it's an update, re-designed, re-worked, what ever, it's just a modified version of the "0" version, and they always use the number "5". So this means the GTX285 is faster then the GTX280. that's obvious.
Nvidia seemed to have paid so much attention to deal on the 200 series, that they even made a nice little addition to help you decide whats faster. Basically, when you see "GTS", it means slower then all other "GTX" 's. So a GTS250 WILL be slower then a GTX260, I kinda think it's useless, but go Nvidia.
What you really want is to start Graphics card reading. Just read reviews on cards, and you'll learn more. I could spend 6 pages explaining how the AMD graphics cards work, what makes them tick, how they process, what makes certain ones good etc, and the same for Nvidia, but alas, I'm not so diligent. And I'm not really inclined to spend 2 hours typing. So just use the internet and learn. read reviews. Hell I'm gonna be so kind as to push you into
http://www.tomshardware.com .
Cheerio.
hey guys i'll soon be in the market for a new gpu( upgrading my 9600gt) and i often struggle to understand gpu's. my question is: what are the most important bits to a gpu?
