🚀 go-pugleaf

Hi Inglo:

Well, you were of course correct.  I found out some interesting things about
my particular system, although nothing really new considering the extensive
benchmarking and comparisons done by the "pros" out there.  I just got done
wasting about an hour or more running the timedemo demo1 with several
different graphical options.  Here's what I got:

My system specs, for reference:

P4C - 3.2 MHz HT (Northwood)/800 MHz FSB/ 512 kB L2 / 875P Intel Chipset
2048 MB DDR - SDRAM @ 400 MHz w/ 3-3-8 timing
Radeon 9800 XT 256 Mb graphics card w/ 400 MHz RAMDAC
500 Gb RAID 0 7200 rpm SATA storage

(for all of these the values correspond to Graphic Option/Resolution/Advance
Options Details/FPS average).  In all cases I used fullscreen, no vsync and
ran timedemo demo1 5 times, averaging the last 4 which is the value reported
here.  For advanced options, if not mentioned it is enabled (except for
vsync).

1.  Ultra / 1024x768 /  no AA / 39.7
2.  High / 1024x768 /  no AA / 43.7
3.  High / 800x600 /  no AA / 53.5
4.  High / 800x600 /  no AA, no shadows / 63.3
5.  High / 800x600 / no AA, no specular / 54.7
6.  High / 1024x768/ no AA, no shadows / 54.1
7.  Ultra / 1024x768 / 4x AA / 24.4

for the next tests I used the settings for a 128 Megabyte vid card  from
here:  http://ucguides.savagehelp.com/Doom3/FPSVisuals.htm
I accomplished this as described there by creating an autoexec.cfg file in
my doom3/base directory.  IMO the biggest effect of these settings is
removing image anisotropy completely, as well as some of the shadow
settings.  Here's what I got with a few variations of these settings.

8.  High / 1024x768 / no AA, no shadows (redundant, I think) / 60.3
9.  High / 1024x768 / no AA, no shadows, set anisotropy to 8 in console /
47.8
10.  High / 1024x768/ no AA, no shadows, set anisotropy to 4 / 51.2

Just going on these results one might think  using the autoexec.cfg settings
with High quality and 1024 x 768 would be the best solution, but this isn't
my opinion.  I noticed during the timedemo runs that the use of this cfg
file really diminished the "beauty" of the game.  It just looked noticeably
worse, even at the pace of the demo.  I wanted to see how much of a
difference these settings would make in actual gameplay, so I switched back
to the settings from run 6 (namely, High quality, 1024 x 768 resolution, no
AA and no shadows (but no autoexec.cfg file either).  I then loaded up the
Alpha 1 save point and used FRAPS to save FPS data to a benchmark file as I
played for about 2.5 minutes.  This is line 11 below.  I then repeated this,
but using the autoexec.cfg file (line 8 above).  I tried to play at a
similar pace and everything to be as closely comparable as possible.  This
is line 12 below.

11.  Max FPS:  62;  Min FPS:  35;  Average FPS:  56.5
12.  Max FPS:  62; Min FPS: 29; Average FPS:  57.8

And, as I stated above, using the settings from that autoexec.cfg file
really made the game look worse.  I don't really know if this is all from
the image anisotropy or what, but as you can see whatever benefit I had in
the timedemo run (60.3 fps vs. 54.1 fps) was completely insignificant during
gameplay.  Probably not that surprising to most of you, but I found it
somewhat interesting.

I also found it interesting that I played the Mars Base levels using the
settings from line 7 above (Ultra/1024x768/4xAA) where the demo gave 24.4
fps.  When I turned the AA off I definitely noticed in the gameplay, but I
found the game to reasonably playable up until that point anyway.

So, it seems that if I'm worried about parity in online play I should be
able to keep High quality and just bump the resolution down to 800 x 600 and
turn shadows off so I can maintain 60 fps as much as possible.  I'm still
debating if keeping shadows enabled while running High quality is worth the
visual enhancement vs. fps deficit.

At any rate, I think I've wasted enough time thinking about this instead of
playing the game.  But, at least I know where I stand and I'm not so upset
that I bought my "ultimate PC" 6 months ago and worried it might totally
suck for Doom3 and Half-life 2.

Mike