codecreatures

[Che Guevara]

How do you guys score in codecreatures benchmark?


Just downloaded it some weeks ago, and I was quite amused seeing that my P3 1Ghz with parhelia reaches avg. 17,9 fps in 1024x768, my brothers p4 2.4Ghz(C) with same memory amount (but two rambus channels) and GF4 Ti 4200 only gets 10,7 fps !!!!!!!!



Is it really such a gpu-limitated bechmark?
Even more than 3dMark?

Hm...

[Indiana]

Codecreatures Benchmark is heavily GPU-limited, yes.

For the question "Even more than 3dMark?" the answer is not really clear. Which 3DMark do you mean?
3DMark2001SE is nowhere GPU-limited with todays faster cards, but mostly CPU/RAM-troughput limited.
3DMark03 is as GPU-limited as it could get, up to the point that you get same scores on an Athlon1400 and a P4@3.2GHz.

For the Codemark results:
With a Radeon8500 I got 12.3 fps in the 1024 resolution, the Radeon 9700Pro (oc'ed to 385/337) gets 44.3 fps.

[Che Guevara]

Ok, It seems this bench is really heavy gpu-limitated (more than any games we see today ).

Thanx

[Chrono_Wanderer]

IMO these GPU-bound benchmarks are actually kinda useful. They tell you the raw GPU power, assuming your GPU has all the features it take to actually run the benchmark. It should be a good measure for GPU makers. i.e. comparison between GFFX 5200 vs. R9800... R9800 is like 4 times faster, and the price also justifies that.

But to us, that's different, real world performance is more important... I think 3DMark2000 is THE benchmark that really test gaming performance. 01 is ok, and 03 means almost nothing at all LOL. I mean, my GFFX5200 is beating my R8500, but in real life that's not true.

LOL just my 2 cents

[Che Guevara]

You're right, gpu-benches are interesting, but not really useful for us.

In my personal opinion 3DMark2000 is no longer actual at all, 2001 SE is a good choice today.

The 5200 really beating a R8500 in 03? Thats a good joke of futuremark/madonion, eh? Thinking the reason are the DX9 benches (a bad result is better than no result in 3dMark ).

[Indiana]

The FX5200s "good" result (if you can call that good) is totally blown up due to nVidias cheating drivers. See one of of the tests with anticheat tools and you know where the FX5200 really stands...

[Che Guevara]

Of course I know about nVidiots cheating drivers, but I thouhgt that was removed in latest driver release?!??


-nVidia hates us all-

[Chrono_Wanderer]

Nay... even the latest drivers (45.23) still doesn't do trilinear filtering... grr...

anyways...

the major reason why 5200 can perform better is because it actually run the Mother Nature scene. For DX8 cards like R8500, it scores low because it gets 0 due to no DX9 support.

However, fill rate and stuff on 5200 is definitely slower than R8500. That's why I said "comparison between DX9 cards", like R9800 vs. GFFX5200. Its not fair to compare a GPU's RAW power if they are of different generations some benches like Mother Nature may not run.

Note that I said "raw" power. Which didn't take the supported feature set into account. If you are a newbie in DCC and want to do some very simple stuff with the latest API, then that's a different story. I think 5200 is actually a good solution for those who want to do a little shaders programming, but is not even close to being a 3d artist. Then in a sense the 5200 is pretty good for 3d artist students... they don't need anything close to Quadro FX 1000 / 2000 / 3000, but just to get the support for the feature set.

Of course, I am not doing DCC, so I may have no clue about it, but that's jsut what i think it is.

[Chrono_Wanderer]

oh yea... i am using 45.24, a leaked beta... but i am not sure if it does trilinear filtering or not (which i doubt)... i dun have any game installed on my machine.

[Che Guevara]

@Chrono_Wanderer

"The major reason why 5200 can perform better is because it actually run the Mother Nature scene. For DX8 cards like R8500, it scores low because it gets 0 due to no DX9 support.

However, fill rate and stuff on 5200 is definitely slower than R8500. That's why I said "comparison between DX9 cards", like R9800 vs. GFFX5200. Its not fair to compare a GPU's RAW power if they are of different generations some benches like Mother Nature may not run."


Just like I said .

"Thinking the reason are the DX9 benches (a bad result is better than no result in 3dMark )."

By the way, nice Avatar .