Quake II performance were great but I can't say the same for Quake III. Why? Help!!

[ManojM]

HI!

Early today I posted my Quake 2 results using the TurboGL drivers on my system.

I have a Pentium III 450 overclocked to 558Mhz with my Matrox G400 Max (32 bit Z buffer enabled)

Using Quake 3 which I have set all the options to high.

Color Depth - 32bit
Full screen - on
lighting - vertex
geometric detail - high
texture detail - full
texture quality - 32bit
texture filter - trilinear

my results:

Demo 1

640 by 480 - 56.3 fps
800 by 600 37.5 fps
1024 by 768 - 23.7 fps

These are way off from what I am seeing from the reviews.

My Quake II results were very very good and my results were same as the ones on the reviews.

So why aren't my quake 3 results good? TurboGL drivers are enabled in Quake3. I am 100% sure it is.

Regards,
Manoj Mahtani


[This message has been edited by ManojM (edited 10-09-1999).]

[Ricardo]

Well with those sethings I get 28.7 frames
at 1024x768x32 (P3-500 not oc). Not bad but
still way far from the 40+ frames that Ant had. Ant can you try the released TGL and post the results ?


Regards

------------------
P3-500, Shuttle 661P, 512Meg HSDRAM,WD Expert 18GB, Segate Medalist 6GB, AOpen 40X CDROM, HP 8100i CD-RW, G400 MAX, SB Live Value, 3Com 10/100 (3C905B-TX) , Sony F400 19"

[ManojM]

Let me double check if this is how you benchmark in Quake 3.

I have to go to the console and type:
timedemo 1
demo q3demo1.dm3

Is this it? If yes then why are my quake 3 scores way off?

Ricardo:

How come you get about 5fps more than me in 1024 by 768 32bit. We both get our scores way off from the reviews.

Can someone help!!!!!

Regards,
Manoj Mahtani

[Ricardo]

Hi Manoj,

What is your sound settings ? I have it on low. Maybe the released TGL is slower than the beta TGL Ant tryed. That would explain wy so many people are having lower scores than Ant.

Maybe Matrox can bring some light into it

Regards

[ManojM]

Ricardo:

My sound settings are all set to high.

Quake II showed the same improvements as the ones on the reviews but the Quake III results are way off!

Regards,
Manoj Mahtani

[Ant]

Try disabling sound and see what that does to the scores. I always benchmark with sound disabled as it can have a big impact on performance and there is so much variation between different cards and different driver versions.

[Ricardo]

Hi Ant,

Q3demo1 1024x768x32 everything max:
with sound enable 28 fps
without sound (s_initsound 0, snd_restart) 29 fps.

Did you already tried the released TGL ?


Regards

[ManojM]

Ok! how do I disable the sound in Quake 3?

But I don't think that is the reason for the big difference in our scores.

Regards,
Manoj Mahtani

[Ant]

Sound can make a differnece although I wouldn't have thought anything near the differences we are seeing here, but you never know.

From the console

s_initsound 0
snd_restart

[ManojM]

Ant:

I tried it.

I disabled sound and I still get the same result.

I only tried once:
at 1024 by 768 (demo 1) - 23.8 fps

with sound = 23.7 fps

This is strange. Quake 2 was great but quake 3?

Regards,
Manoj Mahtani

[Marmita]

I don't have quake2 but does it support Trilinear filtering ??? And if so, did you enable it ???

[Xor'Arch]

Hi ManojM,

Try Bilinear texture filter in place of trilinear, that makes a difference of about 12 fps on my system.



------------------
____________________________________

Xor'Arch

PIII 550Mhz, 256 MB, 10,2 GB 7200 rpm HD, Teac 6x24 cd SCSI burner, Pioneer 6x32 SCSI DVD ROM, Matrox G400 32Mb DualHead, SBLive! Player and lots of other stuff.
____________________________________

[Marmita]

Yeah I posted it another thread... I think the Bete TGL didn't support all the features the final relase has, like for example Trilinear filtering... from my testing I get the same result (~0.5fps) as Ant's tests but I use Bilinear filtering and the same system spec... with Trilinear I go way down. Mybe the Beta didn't support it so it defaulted to bilinear ?!?! I really have no idea but it seems reasonable doesn't it ??? Q3 is much more complex then Q2 and so I should have lower fps for the same resolution.

[ManojM]

Ok I got some new findings:

Let me give you my system specs and my Quake 3 settings again.I have a Pentium III 450 overclocked to 558Mhz with my Matrox G400 Max (32 bit Z buffer enabled)

Using Quake 3 which I have set all the options to high and I have sound disabled this time.

Color Depth - 32bit
Full screen - on
lighting - vertex
geometric detail - high
texture detail - full
texture quality - 32bit
texture filter - trilinear

my results:

640 by 480 - 63.9 fps
800 by 600 - 42.6 fps
1024 by 768 - 27.6 fps

When I changed Trilinear filtering to Bilinear filtering (while all the other settings are the same as above) here are my results.

640 by 480 - 78.6 fps
800 by 600 54.7 fps
1024 by 768 - 34 fps

When I disable 32bit Z buffer with Bilinear filtering (while all the other settings are same as above)

640 by 480 - 85.7 fps
800 by 600 - 66.3 fps
1024 by 768 - 42.8 fps

Now I get similar results as Ant. Only difference is that I am using bilinear filtering and I have 32bit Z buffer disabled.

Regards,
Manoj Mahtani

[Kenny[SP]]

So we see something interesting, those turbogl are not so great (you will see severly differences in 16 bits tests, one of the thing on that g400 was criticized if i am right ) and in my system (athlon 600...) i get 37 fps (g400 dh) using billinear filtering in 1024*768 32bit colors and this is with the 5.30 icd not the turbogl... so matrox did good work, but i don't think we could expect more than we have.

i am still happy