VOGONS


First post, by auron

User metadata
Rank Oldbie
Rank
Oldbie

one thing commonly said about ut2004 is that it's "basically a dx7 game with dx8 used just for speed", in other words that every effect should render correctly on a dx7-class gpu. as the game came out quite late for such a statement i wanted to put this to the test, and for this i used an asus v7700 (geforce 2 gts 32mb sgram). the rest of the system is a piii 1100, 512mb sdram, win98se and dx9.0c. four popular drivers were used for the test: 30.82 (August 7, 2002), 45.23 (August 13, 2003, last detonator), 61.76 (July 20, 2004), 71.84 (March 11, 2005, final driver for this gpu). i'll spoil right away that there doesn't really seem to be any difference between the two detonators. also this test is only about image quality, not performance.

i've incuded a picture of the used settings, however i haven't found any setting that would alter the effects in question at all anyway. cubemaps, hw t&l and vertex shaders were enabled in the ini.

https://nofile.io/f/hbgOSyqs6z4/ut2004gf2gts.7z

25 health:
a clear progression visible here. the latest driver manages to render this correctly.

100 health:
very similar progression here. the detonators render it transparent, the 2004 forceware renders the effect to black while the latest driver gets it correct.

dd:
this is where it gets interesting: all drivers fail at rendering the "ripple" effect correctly up close, however with the 61.76 i was able to see it from afar. 71.84 had the same result as the detonators in this case (forgot to make a screenshot here). epic also screwed this effect up in the dx9 renderer.

shock:
translucent purple ammo canister effect broken in the same way on all drivers.

the shield pickups appeared correct on all drivers from what i could tell so no pictures of those.

so the verdict, no, the gf2gts does not display all effects 100% correctly even with the latest drivers. considering that the game came out in march 2004 it's quite remarkable that even after the july drivers there have been such changes (to the healthpack effects, namely). if anyone has a setup with an original dx7 radeon i'd be interested in seeing the results for this game.

Reply 1 of 7, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

UT2003/2004 need vertex shader support for proper rendering of all effects. Emulation of vertex shaders in Catalyst drivers is worse, if I remember correct.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 2 of 7, by leileilol

User metadata
Rank l33t++
Rank
l33t++

It only uses shaders to accelerate terrain verts. No effects are shadered, just lots of multitexture combine ops and a cubemap here and there. HWT&L only applies to rigid meshes (and is why many many players are made out of segmented pieces of hard armor).

Forceware 61.76 and newer on Win9x regresses the Geforce2 in my experience, despite allegedly supporting it. It's supposedly better behaved under XP....

apsosig.png
long live PCem

Reply 3 of 7, by Reputator

User metadata
Rank Member
Rank
Member
leileilol wrote:

Forceware 61.76 and newer on Win9x regresses the Geforce2 in my experience, despite allegedly supporting it. It's supposedly better behaved under XP....

I haven't gone through each and every effect being done in the game but this was certainly my experience. 71.89, the last official driver for GF2, is rather miserable at a number of things.

https://www.youtube.com/c/PixelPipes
Graphics Card Database

Reply 5 of 7, by lordmogul

User metadata
Rank Newbie
Rank
Newbie

Well, I've been running UT2004 with a GF 3, same goes for Halo. (Using driver 41.09, in 98 SE and XP)
If it is about image quality, there are DX10 renderers available for Unreal and UT99.

EDIT:
Ran both at 640x480 with lowest settings. For Halo as comparison to the performance on the Original XBox which basically runs the same hardware, although slightly slower.

P3 933EB @1035 (7x148) | CUSL2-C | GF3Ti200 | 256M PC133cl3 @148cl3 | 98SE & XP Pro SP3
X5460 @4.1 (9x456) | P35-DS3R | GTX660Ti | 8G DDR2-800cl5 @912cl6 | XP Pro SP3 & 7 SP1
3570K @4.4 GHz | Z77-D3H | GTX1060 | 16G DDR3-1600cl9 @2133cl12 | 7 SP1

Reply 7 of 7, by auron

User metadata
Rank Oldbie
Rank
Oldbie

35-50 while just running around, 20-30 in firefights. that's on a default rankin botmatch and needless to say that's about the best case scenario for the game, bigger maps and other game modes like assault choke a pentium iii really hard. i've tested this under 800x600x32 with low-medium textures and everything but trilinear filtering turned on, going 640x480 gets slightly better max fps but still same lows.

since my initial testing i got an original radeon ddr 64mb, tested this with catalyst 3.7 and 4.3, the former of which being really weird in that it requires installing the default windows vga driver before driver install, and then includes no control panel whatsoever... anyway those two drivers seem to be pretty much the same for ut2004, very similar performance range to the geforce2 gts. i was expecting the memory size to make a bigger difference than it did, however the aforementioned settings on rankin don't seem to be enough to cripple performance on 32mb cards, so the advantage is likely more visible with things like higher texture settings or other ini tweaks.

effects wise it has all the same problems as the gf2gts except for the kegs, but on a quick test i noticed a few extra glitches, namely the bluish spinning light effect on ctf-bridgeoffate weapon spawns being rendered on top instead of below the ground, and some flickering large static tree shadows on dm-antalus (dynamic player shadows seemed sometimes cut off as well).