Quake III Arena
"Quaver" Benchmarking demo
by Reverend
24 January 2000
============================

Thank you for downloading this demo.

==============
INTRODUCTION
==============
First of all, I made this demo specifically to test the performance of my new
nVidia Quadro 64MB DDR reference board (review forthcoming). More
specifically, I was curious to know how the additional 32MB of memory
(over any retail GeForce-based cards' 32MB) would matter.

=================
ABOUT THE DEMO
=================
This demo was recorded with the retail version of Quake3Arena, patched 
with the 1.15 point release. The map is in Q3DM9, specifically chosen
because the usage of large textures in the map, exceeding 30MB. Because
of this, it is a good test of the possible limitations encountered with using
a video card with 32MB or less of on-board memory.

The demo was recorded at 640x480 at the default "Fastest" setting. It was
recorded with 3 bots with a fraglimit of 15. Bots used were TankJr, Bones 
and Uriel. I did not want to use anymore than 3 bots since that could use up 
quite a bit of CPU time.

Cheats were enabled and used (GOD and all weapons incl. the BFG,
a weapon not found in the map).

=============
INSTALLATION
=============
Once you have downloaded the zip file and extracted it somewhere, you
will find within this archive :
1) readme.txt
What you're reading now
2) "demos" folder
This folder contains the actual demo file named "Quaver.dm3"

Assuming you installed Quake3 to :

C:\Quake III Arena

...move the "demos" folder to :

C:\Quake III Arena\baseq3\

(note : if you already have a "demos" folder, don't worry, nothing will
be overwritten unless you have a demo file also named "Quaver.dm3")

===================
RUNNING THE DEMO
===================
(A) To run the demo in *normal* mode, load up Quake3. Once you're in the main
menu, click on the "Demos" option and choose "Quaver", click the "Play"
button and the demo will run its course normally.

(B) To run the demo in *timedemo* benchmarking mode, load up Quake3. Once
you're at the main menu, pull down the console (press the "~" key) and type :

r_swapinterval 0 <ENTER>
timedemo 1 <ENTER>
demo quaver <ENTER>

...where <ENTER> simply means pressing the ENTER key. The demo will
then load up and run as fast as your system can manage. Once the demo ends,
you will be back at the main menu. Pull down the console again and read the
last line - it will tell you the benchmarking statistic/framerate.

=================================
PERFORMANCE & CONSIDERATIONS
=================================
I'm going to list benchmarking statistics on my system below. However, if you are
to benchmark your system and compare your stats with those listed below, bear in
mind that we want things to be almost identical in terms of set up. For that purpose,
please adhere to the set up info provided below.

My system essentially is :
Intel Pentium3 500 overclocked to 620MHz
384MB PC-133 Micron RAM
nVidia Quadro 64MB DDR reference board

Latest drivers were used for all components. The Quadro was not overclocked for
the benchmarks below, with default core and memory clocks of 135MHz and 333MHz.

To benchmark, I disabled VSYNC for my Quadro. I also did the same in-game using
the variable "r_swapinterval 0" (as mentioned earlier). High quality sound was used.
Quake3's "Game Options" had everything turned on EXCEPT :
Simple Items
Sync Every Frame
Force Player Models
Draw Team Overlay

Note that the particular option of "Sync Every Frame" is not related to VSYNC despite
its name. It is something that is supposed to fix certain hardware-related driver anomaly
where there's an occurance of timing differences between input and actual screen
update. 

==Benchmarks==

I used the game's High Quality setting. However, for the purpose of benchmarking and
reason for recording this demo in the first place (testing the Quadro's 64MB DDR), I
changed the following High Quality settings from their default :

- Geometry Detail HIGH
- Texture Detail slider EXTREME RIGHT **very important!!**

As mentioned, High Quality Sound was used but A3D was not.

The figures in brackets are those of a friend whose system comprises :
Intel Pentium3 600 @ 618MHz
128MB RAM
nVidia GeForce-based 32MB DDR card

...which basically means that the only important difference between his and my system
is that my Quadro has 64MB DDR and his GeForce has 32MB DDR. He also overclocked
his GeForce to 135MHz core to ensure similar clock rates for both core and memory.
Thanks, Angelo :-)
The difference in system memory shouldn't matter, as long as there's at least 128MB, it
should be adequate and will not influence benchmarks.

Also of note, is that you have to ensure that you're running in 32bit Color and Textures
mode if you want to compare.

32bit Color Depth & Textures, maximum Texture Detail, High Geometry Detail
==================================================
Resolution	Framerates : 64MB DDR (32MB DDR)
==================================================
640x480			68.5 (60.2)
800X600			67.2 (42.3)
1024X768			53.0 (15.1)
1280x1024		32.8 (7.6)
==================================================

I also did some benchmarkings with 16bit Color Depth and Textures, just in case you
have a card incapable of such bitdepths. Only for my Quadro, not the 32MB DDR GeForce.

16bit Color Depth & Textures, maximum Texture Detail, High Geometry Detail
=======================================
Resolution	Framerates : 64MB DDR
=======================================
640x480			74.3
800X600			74.6
1024X768			71.8
1280x1024		52.0
=======================================

===========
COMMENTS
===========
The above benchmarking differences are incredible to say the least. I remind you that
the 32MB DDR GeForce had its core frequency overclocked to match my Quadro's,
i.e. 135MHz, so the fillrates of both cards are the same. Just take a look at the 800x600,
1024x768 and 1280x1024 scores!

All video cards should come with at least 64MB of memory I say :-)


--
Anthony)))Reverend(((Tan
http://www.voodooextreme.com/reverend
24 January 2000

*All trademarks mentioned herein are properties of their respective owners, blah, blah.
