quake3arena v4 benches...

[hans007]

Hmm well i have a "burned" copy of q3a. I told bfg10k i'd run q3a demos to see what difference it made on my v4 in this thread Anyways, i cant even get a demo to run, i installed 1.17 patch, and 1.25 on top of it. I type

timedemo 1

demo demo001

and it says something like "tried to load VM failed, dll something"
in the console.

I recorded my own demo and that worked and ran

Anyways, my computer isn't even overclocked so i have no idea wtf is wrong. I'm figuring maybe its because i run multi monitor or something, anwyays, i'm almost through downloading q3a demo version to see if this works, I just reinstalled 2 weeks ago. Maybe its because i have no cd key and my CD is a burn i dont know. Anyways i'll post more if q3a demo version runs demos, dont know if those benches are comparable oh well. Also if i exited q3a after trying to run one of ID's demos it would hard lock. Hard lock win2k, i have no idea wtf is going on

 

[RoboTECH]

timedeom function doesn't work with 1.25

delete pak4.pk3 and reinstall the 1.17 point release

 

[hans007]

ok i ran the q3ademo (demo version of quake3) at start up it says 1.11 in the lower right. Oddly here were my benchmarks. Again nothing overclocked, all at defaults, and the 3dfx driver was untweaked.

Normal quality

640 = 79fps
800 = 65.7 fps
1024 = 41.4 fps

Hi quality

640 = 78.8 fps
800 = 65.7 fps
1024 = 41.4 fps


Ok yeah that looks really weird, but those are the numbers i got, i guess i'll try it again with the full version.
Anyways, i dont know how to turn vsync off, so i dont know if its on or not. this is with the win2k whql 1.04 version drivers. tbird 800, with sound on, and my sound card is the VIA ac 97 codec on my motherboard, i dont know what effect that is but i'm sure its bad. Anyways, the HQ seems pretty fast.

I have no idea how to compare these to anand's benchmarks in the latest budget card roundup. However did those benchmarks in my opinion is a really crummy writer. First off he never says what version of quake3 he used. second all it says is 1024x768x32 with no mention of HQ or normal, third no mention of sound. My 41.4 fps at HQ in 1024 is better than their 39.9 or whatever and i think thats pretty good considering the a7v is usually faster than my gigabyte ga7zm, they were using a non ac97 sound card, and they use a tbird 1000 (vs my 800) and i'm using pc100 cas2 and they used pc133 cas2. Ok time to try it in 1.17 quake 3 retail . In their original v4 review, in win98, they used a 1.1 ghz tbird and that produced 1.1 fps more at 1024, with everything being the same except the CPU and driver version btw.

 

[hans007]

ok i ran it in q3a 1.17 retail (burned copy) . The HQ and normal having the same numbers wierdness was still there. ANyways here are the numbers thanks to robo and his tips on how to get timedemo to work.


normal
640 78fps
800 65.5 fps
1024 41.5 fps

HQ
640 77.9 fps
800 65.5 fps
1024 41.5 fps
1280 23 fps


Anyways i ran these 2 times each, because well the first time it would load all this crap from the CD and slow it down since i did miniumum install, so i took the numbers from the 2nd run. I guess considering my system is slower than the one anandtech used, that 1024 HQ could probably hit 50 fps on his setup. Anyways, everything i did was untweaked , and lastly my desktop resolution was 1600x1200x32x75hz which might be the reason the 640x480 numbers cap out at 78 or so. Anyways, hope you guys can analyse these numbers, i'm thinking they're not so bad considering my system is worse than anands and it still got higher 1024x768 framerates at HQ. Oh well there is an hour out of my day spent. i guess the v4 isn't as fast as the mx yet, HSR will probably help at the higher res where it matters, but then again at least i'm not going blind at my desktop resolution, stated above. I am not a demo running, 3dmark wacko, so i dont know what the tweaks are and fastest settings. I remember when the original q3demo came out and i was on a k6-2/400 with a rage128. Wow 640x480 normal 29fps, and i thought that was plenty fast.

Oh yeah and lastly i noticed a couple things in q3a. In a game that fast paced, i dont think i could even notice FSAA since i wouldn't be paying attention. In a much slower game like CS i can . Also the movie at the start of q3a is pretty cool

 

[RoboTECH]

the only difference between "normal" and "high quality" is that "high quality" forces 32-bit textures and color depth and also trilinear filtering

"normal" just uses your desktop color depth

 

[BFG10K]

First of all Hans007, thankyou for taking the time and the trouble to run these benchmarks for me. I really appreciate it.

Right, now onto the results. Anandtech used 1.17 Quake 3, ran at HQ settings and used an Athlon 1.2 GHz.

Your results:

640 = 78.8 fps
1024 = 41.4 fps

Anandtech's:

640 x 480 x 32: 96.6 FPS.
1024 x 768: 41 FPS.

Your scores are the same (except in low res where you are CPU limited) as Anandtech's. I'm not sure what Dave's going on about because there is no difference between the 1.03 and 1.04 drivers.

 

[RoboTECH]

1.04 + DX8 + Win2k show a nice performance increase over DX7 + 1.03 + Win2k

that's about all I can think of.

 

[hans007]

bfg10k i've got a much crummier setup than theirs.

 

[BFG10K]

bfg10k i've got a much crummier setup than theirs.

At high resolution it doesn't matter.

 

[Nesta]

Can you run the time demo with the Q3TA demo?

 

[hans007]

what is q3ta?

 

[RoboTECH]

Quake3: Team Arena

it's iD's attempt to win over the Unreal Tournament crowd.

it's a bigass download

if you like UT, you'll like Q3:TA

if you don't like UT (and NOT because &quot;UT $uX0rZ, Qua|<3 r0000Lzzzz) you may not like Q3:TA

if you have a good internet connection, you may want to check it out

 

[hans007]

hmmm i liked UT, but currently i play one game and that is counterstrike and most recently maybe i play that an hour a week or less. I'm on a t3 though so i might give that a try, then again i dont have a CDkey for q3a and i'm not about to pay money for that game, at least you could warez UT and play online. Neither of those games i'd say was compelling enough for me to actually buy. I bought halflife for $15 used, that game is by far better than q3a and UT especially with CS. Prooves once again that pretty graphics only woo people who are impressed by mere pretty graphics.

 

[Sunner]

Robo, DX8 shouldnt matter compared to DX7 when it comes to Q3 since Q3 uses OpenGL.

 

[RoboTECH]

Sunner, I'm talking about overall performance, not just Q3. Should've clarified that.

However, the GRAPHICS of Q3 uses OGL. Sound and controller input are greatly affected by Direct Sound, Direct input, etc.

So DX8 does have an effect, tho obviously not as great in the graphics arena (specifically)

 

[Nesta]

Still no answer after the above debate (UT/Q3)...

Can you run timedemo on Q3TA?

 

[RoboTECH]

timedemo 1
demo demo0000

 

[Nesta]

Thanks again ROBO.

What driver do you use in Q3A. The WickedGL or the latest 3dfx?

 

[hans007]

i dont have time right now to run this, finals week is next week. i'll do it later, or when i'm procrastinating.

 

[BFG10K]

RoboTECH:

However, the GRAPHICS of Q3 uses OGL. Sound and controller input are greatly affected by Direct Sound, Direct input, etc.

Whether or not you use 3dfx's 1.03 or 1.04 drivers is irrelevant because they have nothing to do with sound or keyboard speed. Anyway in order for any speed gain to be possible Microsoft must have put major performance enhancements in Direct Sound and Direct Input.

I'll ask a simple question: did you see a boost in Quake 3 when upgrading to Direct X 8 and the 1.04 WHQL drivers? If you did, how much was it?