• Forum Moderator applications are now open! If you're interested in joining an active team of moderators for one of the biggest Pokémon forums on the internet, click here for info.

Video/3D Card Help

Status
Not open for further replies.
V

Vereux0

ヴォルクス
Joined
Jun 21, 2007
Messages
325
Reaction score
0
How do you measure how much rendering power a 3D card has? Is it the clock speed?
 
Clock speed and memory statistics are only useful if the video cards you're comparing have the same graphics chip. Otherwise, they're useless values.

The only way to truly measure power of a video card is to simply run a bunch of tech demos and measure the average frames per second. It's likely you don't have all the prospective video card purchases in front of you so, the second best way to do it is to look at various hardware sites that will do the benchmarking for you. I happen to use Tom's Hardware's VGA Charts for the latest video card benchmarks.
 
Last edited:
Also, features such as vertex/fragment/geometry shaders are a plus, and in some poorly thought out cases required to run games (NWN2, I'm looking at you)
 
I just need something that can run Spore and Oblivion.
 
Again, look at the benchmarks I linked to. They use Oblivion for one of their benchmarks.
 
If you wish to run Oblivion nicely, you will need at least a GeForce 6800, GeForce 7600 GT, Radeon X800, or Radeon X1600 Pro.

If you wish to run Spore nicely, you'll need a GeForce 8 or a Radeon HD 2000 card. Spore takes advantage of the new features these cards have. While it's possible to run it on older hardware, it's not going to be as nice.
 
You might also want to take a look at this site.

Personally, I wish the wide world of video cards wasn't so damned obscure and arcane... but it is.
 
Actually you can measure a graphics card power using some simple formulas.

For bandwidth of its VRAM (which is important for high resolution textures and effects) it's: ( bit width/8) * effective clock speed. So for example, if you have a 128-bit bus using RAM clocked effectively at 1.2Ghz, the bandwidth would be (128/8)*1.2 = 16*1.2 = 19.2GBps.

For pixel shader power, it's: number of pixel pipelines * GPU clock. So if you have 24 pipelines and a clock of 450Mhz, that's 10.8GPixels/sec.

For vertexes per second, take the number of vertex pipelines * GPU clock. So if you have 8 pipelines and a clock of 450Mhz, that's 3.6GVertexes/sec.

This is just theoretical of course. Real world performance depends upon drivers, API used, etc.
 
Status
Not open for further replies.
Back
Top Bottom