Jump to content
  • Advertisement
Sign in to follow this  
schupf

GPU History/Performance

This topic is 3895 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hello, I am searching a site/document which shows the increase of the GPU performance (for example the difference between a Geforce4 and Geforce8800) in words or with a chart/histogram. If possible the comparison shoulb be made by "triangles per second" or "vertices per second". Moreover the source should be reliable - the nvidia site would be perfect (but a forum where people discuss about benchmarks wouldnt be usefull) I searched for hours but couldnt find anything :( Thanks for help!

Share this post


Link to post
Share on other sites
Advertisement
I think the problem is probably that people stopped paying attention to polygons/vertices a long time ago. Vertices just aren't the bottleneck any more, the real bottlenecks are: batches, bus transfer rate, fill rate, texels, pixel shader ALU instructions, and pixel shader texture fetches. And since there are so many bottlenecks, it's a little hard to do apples-to-apples comparisons.

- Josh Yelon
Teacher, Carnegie-Mellon Entertainment Technology Center

Share this post


Link to post
Share on other sites
Not really what you asked, but this could be interesting:
3d cards specs
and
probably the same informations

I don't think that it is possible to compare all cards togheter: there aren't tests performed on them all and the declared performances often are more theoretical than real. For example until the first geForce cards didn't perform transformation and lighting, wich had to be computed by the cpu. This also means that a game was often highly cpu bounded.

Lately nvidia and ati stopped declarations about vertices per second and so on, because it is no longer the main parameter (and in my opinion they didn't increase that value very much in the lastest cards, focusing more on shaders).

The best you can do is searching old benchmarks and trying to infere the relative power from different benchmarks.
ie:
First benchmark:
geForce 4 with Quake 3 50fps
ati radeon 9800 Quake 3 42fps

Second Benchmark:
ati radeon 9800 Half life 2 31fps
ati readeon x800 Half life 2 55fps

Third Benchmark:
geForce 6800 Far Cry 39fps
ati readeon x800 Far Cry 33fps

then you can say (oh, well, so to say...) than the 6800 is 20% faster than the x800, the x800 is 80% faster than the 9800 and the 9800 is 15% slower than the geforce4. You should be able to make a (probably wrong and surely unfair) comparison between the 6800 and the geforce 4. If you are lucky you could find benchamrks that already cover a long period of time. Rather than with games, you may want to look for 3dmark2001 and later tests.

Are you going to make your results freely avaiable on the net?

Share this post


Link to post
Share on other sites
ShaderX3 and ShaderX4 have articles in there that benchmarked graphics cards. Those benchmarks are really interesting because they are raw numbers. So you can nicely see when the performance breaks in with a certain texture size or what texture format is implemented in the driver.
This allows you to optimize your game. Obviously PIX gives you also a lot of knowledge.

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

We are the game development community.

Whether you are an indie, hobbyist, AAA developer, or just trying to learn, GameDev.net is the place for you to learn, share, and connect with the games industry. Learn more About Us or sign up!

Sign me up!