Darkoz

It definitely goes to show how different benchmarks can produce completely different results.
This benchmark doesn't paint the same picture.........

http://www.vr-zone.com/?i=1181&s=1

Veridian3

Sounds like you dont have reflect all on. Also CPU has little impact on the benchmarks at these settings...this seems to be a GPU limited test more than CPU.

Also there are quite a few factors that affect the performance of Source which i noticed in my testing however i am 100% sure that the results between the 2 cards on my system are comparable.

Veridian3

Probably also due to them using CS and me using Source with much more demanding environments. Also the Video Stress Test doesnt have a fps cap...i think CSS might... i'll check.

Risiko

Yeah, there is something absolutley wrong with the 6800Ultra graphs. It just doesn't add up.

digitalemperor

updated my thread a little bit. i did indeed turn on reflect all.

i also realize the only two posts i have are today but as you can see, i've been registered for almost a year. i don't understand the benchmarks at all and I've tried multiple times to replicate the performance you are seeing and I just can't get it to add up. i've trusted this site for fair benchmarking for some time but this does raise a little bit of concern.

Hamidxa

Thats the OLD benchmark for CS Source.

This isnt


That was a lot less taxing on the GPU, this once again isnt.

Im not surprised at all by the results, ok I am, but pleasantly surprised.

ub3rs0ldat

Uh, you must be a fanboy or something. I got at least a 180 fps on extreme quality. God, you are not doing these benchmarking currect. I have a Nvidia Geforce 6800 Ultra. Stop bullshitting nvidia and get it right. BTW, CS: S is awesome. You should be surprised with the results. I saw the exact same benchmark with same cards and the results was that ati and nvidia is identical. ALso this site is know to be the worst for benchmarking. Why do you rely on this site?

digitalemperor

Actually if you visited this site regularly, you'd see that Veridian3 is not nearly biased enough to be clasified as a fanboy. Infact, the only time i've even questioned benchmarks posted on DriverHeaven is today. I honestly think something went wrong with this benchmark. Call me a conspiracy theorist but it just doesn't smell right. I don't think trying to bring this thread down to a "you're wrong idiot" is at all the right way to go about things. Does anyone else have some information (maybe some more personal benchmarks) that we can compare with?

Erroneus

Extreme quality.... funny beacuse there is no such thing as extreme quality in the source options

So i dont find your results trustworthy, if you don't even know the settings in the games.

Looks to me, a nvidia forum heard of these results and now they are all signing up in here, to cry about how eviiiiiil DH is and bla bla bla

btw. how do you test, i would really like to know this? net_graph 1 ??

__________________

realjessy

no no no i dont believe this....this is like wanking ....honestly....6800 shouldnt be that slow, coz other site showed different results LOL...

i just say a honest word..dont bash me plz

Erroneus

Be noticed, that many of the test other sites used, was an "old" source with a test, limiting the GPU.

__________________

realjessy

oh...ok...no biggie, ummm can u plz explain wat is old source?? wat is the difference?

digitalemperor

above all else, we should keep in mind that this is a bizaare tech demo which does not feature any charcter models, gameplay, or environments like we are going to be seeing in HL2. It's a little presumptions to say that this test should define a "huge victory" for the x800xt PE. i have no doubts that the x800xt PE will end up being the card to have for half-life 2 but the benchmarks posted do not jive with what i've personally benchmarked. now i do not benchmark for any large sites, but I like to consider myself atleast competent when it comes to the task. i think if we let the chips fall where they may, we'll get a pretty good idea how the #'s posted compare with the masses. valve is planning on releasing a nice bundle of information about everyone who's run this test and submitted their scores. i don't know how they'll handle overclocking or anything of that nature but hopefully it will show that the 6800U performs "reasonlbly" better than the test shown here.

germanjulian

I dont understand how doing a typical male sexual act has to do anything with benchmarks?

people trust me veridian 3 is propably checking out why the benchmarks are so low and zardon or so propably is testing as we speak....

you know this site. every time something is questionable they triple check their results.

__________________
/|\ Asus P5W DH Deluxe, Intel C2D E6600, 2GB Corsair XMS2-6400C4 DDR2, E-VGA GeForce 7800 GT, Creative X-Fi Extreme Music, 500GB Seagate 7200.10 SATA, Lian Li PC-V1100 Aluminum Case Black, etc. http://germanjulian.com /|\

BWX

http://www.vr-zone.com/?i=1181&s=5
"In order to benchmark using the Source Engine and Counter-Strike: Source, we used the included recording and demo playback features while running FRAPS in the background to record frame rates."




http://www.vr-zone.com/?i=1181&s=6
"• Windows XP Professional SP1
• VIA Hyperion 4in1 4.53
• Detonator 61.77
• Catalyst 4.8
• Desktop resolution 1024x768x32, 75Hz"





http://www.vr-zone.com/?i=1181&s=7
"Our demo was recorded on a 24 player Counter-Strike: Source server using 4 minutes of actual game play. "




http://www.vr-zone.com/?i=1181&s=9
"Well it seems that there is a cap of about 90 frames per second in the build of Counter-Strike: Source"



"The ATI Radeon X800 XT Platinum Edition was able to run around 7 FPS faster at 1600x1200x32 with 4X AA and 8X AA enabled, compared to the approximate 1 FPS difference between the two cards at every other setting.

This is a good indication that ATI probably has a winner on its hands with games based on the Source Engine, but really it's a little too early to tell."





I'd trust DH over that sketchy benchmark any day... It looks like they did everything possible to get the CPU/system bottleneck to have the greatest effect on the outcome.

__________________
One Big Ass Mistake America

realjessy

hey..ye i know..i just dont wanna see a biased benching

mini

Hehe... That test on VR-Zone isnt that on a de_dust sort of map... So there is NO demanding graphics... AT ALL...

They wrote: Cap of 90 FPS... How the hell can they see that??? They have in 800x600 74 FPS... Lame test...

This is a great 1... Tests alot of nice graphics... But hey great to see this...

So:
Doom3=Nvidia game... 19% performance with 4.9 beta (14% with humus+4.9beta)
HL2= ATI game... 65% performance... Hmm...

I dunno bout you guys but im goin with ATI...

I was thinkin about goin Nvidia... But after trying doom3 i realized its 1 of those games you just complete and throw away... The Multiplayer is BORING... And the singleplayer isnt very challenging... You dont have to do any thinkin... No tactics just shoot...

If HL2 is half of what HL1 was then im with HL2...

s-flow

Looking at the charts I am happy to own a 6800GT (400/1100), at res. 1600x1200
with 4X AA and 8X AF at max detail this card runs at avg 50 FPS.

Now, thats fantastic.. It is fast!

Why? As long as the framerates are high and games runs smooth
there is no need to be dissapointed! Or is this all about fighting graphics card
brands.

As I see it, this is great for NV users and even greater for ATI users.

Demigod

It should also be noted that using 8ss on the nvidia cards has a much greater perfomance hit than 6aa on ati, the type of anti aliasing isnt comparable. Ati said last year they wont use ss as the gain in aa quality isnt worth the loss in performance (see this thread http://www.rage3d.com/board/showthre...super+sampling)
So those results arnt truly comparable, its a pitty nv doesnt offer a 6aa similar to ati's. The test @4aa is much more comparable imho

Zardon

im sure veridian3 has ran numerous tests to verify results, I really have nothing to do article so I cant "check" anything. If anyone has questions its best to ask veridian3 no need to PM me guys.

Veridian3

Sorry for the non response to questions over the last hour or two, i've been off doing some further testing...

A lot of people questioned the NV results so i've concentrated on that...

The graph below now has the original 2 results and in addition i've added results for the NV card using forceware 65.62 then some with an extra GB of ram added to that system, then the result of the card overclocked to its max using coolbits +1gb extra + 65.62. I also reinstalled windows, SP2, Hyperion 4.53, and Source to be safe...

1600x1200 4aa 8af


Probably worth noting that all optimisations were on for the 65.62 and i tested with vsync disabled in the driver and app and also aa/af enabled in both to check the results were ok no matter what method was used. Finally, my original findings have been replecated by 2 organisations outside DH.

Demigod

Cool thanks for the work. Can you tell me , if you know, does the benchmark use 3dc, as that would cetainly explain the diferance as it can be much faster then dxtc5

Oh and for me (with a gt) Im quite happy with the results they look playable to me which is all that matters to me

HawK

Ah yes maybe you like tests runnin a 2500xp with a 6800u vs a X800, like I have seen somewhere (luckely I forgot where).. ,
yeah that would not make much diffrence then.
these are top cards ! no sense in testing with a midrange proc.
@ s-flow: Was refering to the few here on DH, who'd changed their X800 for a 6800 to play Doom3 (for 2 day's)
Myself ? still on a 9700 wich will do for HL2 I bet

BWX

I agree with that, but I also think that ATI's AA looks just as good if not better (in the latest generation of cards- earlier generations ATI wins hands down)- Nvidia's AA has always looked more jaggy to me..

edit- spelling

Poppi

. And I was still able to play Doom3 at 1600x1200 16x Aniso in the CP, with High Settings with all Doom 3 quality turned on.

__________________

BWX

I bet you are right, I did some "testing" with my OC'd 9700pro and I think HL2 will be playable even with some AA and AF on a 9700p at 1024 x 768.. With still looks great.. An upgrade to a 9800pro with an OC'able r360 core would probably be quite worth it if you can't afford an x800 yet though... depending on how far you 9700p OC's..

I don't think HL2 will run nearly as fast as this game though... could be wrong..

Risiko

Veridian3, I'd like to ask you to redo the 1280X960 w/ 4X AA and 8X AF benchmarks. The original ones are simply wrong, and it's therefore very misleading. I was able to get 83.5 FPS every single time I benched with the exact same setting as you. How could you get ~64? Please, if you can find the time, redo those.

HawK

????? Why? you can't blame ATI for that can you?
I Did not see those remarks after the Doom3Benchmarks either.
ATi has an Advantage vs NV in HL2 because Ati keeps itself to the DX.x.x API's, not because it is optimized for the Source Eng., but NV not sticking to the rules of that API is what gives a lowerBM

@ BWX: counting on the 9700 beeing 1/2 the speed of X800, i'll be happy with 40-50fps on 1280

Demigod

Its not the stress test that comes with cs:s though that Veridian3 used is it?
The picks look simialr to those Ive seen but Im not sure?

(hay I may make 20 posts if I keep this up )


Oh and above I wasnt blaming anyone for anything it was just an observation, and I didnt know benches used aa for doom3 at least the hardocp ones didnt.

Veridian3

With the extra gb of mem and the 65.62 drivers i'm now getting 76fps at 1280x960 4aa8af (stock clocks).