Benchmarking ut2k4 demo

mollmerx · n00b Joined: 19 Dec 2003 Posts: 41 Location: Cambridge, UK

Hi,

is it normal, that nothing actually happens in the benchmark? What I mean is that the benchmark starts a normal game, in which I can walk around and shoot. If I do nothing, I just stand there until one of the bots comes along and shoots me

Also, considering my system specs, I'm not too pleased with the result:

14.465229 / 27.761543 / 37.629829 fps -- Score = 27.775936 rand[1490739532]

this is on a Barton 2500+ running at 2200 MHz (=3200+), 200MHz FSB, 512MB 400MHz Dual Channel Memory and a Radeon 9600pro chip. Ati-drivers 3.7.6

Surely, there must be something wrong here? The game is playable in Linux, but not as smooth or good-looking as in windows.

So, any ideas why my system is so slow, or maybe an idea why the benchmark puts me into the game as a player and not a bot?

Thanks a lot - any help is much appreciated,

mollmerx

PS: In case it should be of any relevance: I have the German retail version of UT2004 and a German keyboard. I am not able to use the console - pressing ~ does not bring it up.

DocGonzo · Posted: Wed Apr 07, 2004 2:00 pm Post subject:

koubiak · Posted: Wed Apr 07, 2004 8:53 pm Post subject:

Somebody can do a benchmark script for the final version ?

I have switch with my own lib

I have a question about

mollmerx · n00b Joined: 19 Dec 2003 Posts: 41 Location: Cambridge, UK

Hi,

Thanks, DocGonzo, the Benchmark runs nicely now. Sorry I hadn't seen how to do in the thread earlier.

The other, much more critical, problem still exists, though: In Windows I get around 60 AVG FPS, running the same map, deetails, bots etc. in Linux I get exactly half that: measly 30 AVG FPS. All it takes is for a bot to come into view and it goes below the critical 30.

I know some people have been complaining about the performance being lower under Linux. But is it really meant to be THAT bad?

It really hurts, because I bought the game specially for the Linux support, and now I might have to end up playing it in windows. :cry:

Anyone experiencing similar problems?

mollmerx

KozmoNaut · Posted: Fri Apr 09, 2004 10:22 am Post subject:

(Replaced "true" with "1" in the benchmark script, since I'm using the full version)

Specs:
Soltek SL-B8E-F (i865G)
2.6GHz Pentium-4 Northwood-C
2x512MB Corsair XMS PC3200
Albatron Geforce4 TI 4200 128MB

UT2004 Build UT2004_Build_[2004-03-03_02.42]
x86 Linux
GenuineIntel Unknown processor @ 2594 MHz
GeForce4 Ti 4200 with AGP8X/AGP/SSE2

dm-rankin?spectatoronly=1?numbots=12?quickstart=1?attractcam=1 -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

11.443704 / 68.182480 / 173.010193 fps rand[1881639875]
Score = 65.493668

This seems to chime much better with the other results in this thread.
_________________
War. War never changes.

BlackTiger · Posted: Mon Apr 12, 2004 3:29 pm Post subject:

Benchmark on UT2004 Demo:

816134 / 45.914349 / 124.916580 fps rand[1814308868]
Score = 45.769192

I am using the 3.2.8 ATI driver (Newest 3.7.6 runs slower ...), but seems that this performance isn't well for my hardware:

AMD Athlon XP 2500+
Abit AN7
Sapphire Radeon 9800 Pro
SB LIVE! 5.1 Player
Gentoo Linux 2004.0 - Kernel 2.6.5 - KDE 3.2.1

any suggestions how i can increase my performance?

Mfg
BlackTiger

mollmerx · n00b Joined: 19 Dec 2003 Posts: 41 Location: Cambridge, UK

Hi Black Tiger!

a score of 46 is really not too bad. The game should be playable under most circumstances.

Because the benchmark runs at low resolution, it is mainly limited by the CPU, rather than the graphics card. Version 3.2.8 of ati-drivers is definitely the one you should be using, so things are ok there. One thing you could try is to run the benchmark without sound. Check some earlier posts in this thread to see how to do this. If there is a great difference (i.e. you start getting more than 50) try compiling your own openal driver. It is explained earlier on in this thread how to do this. I tried it and it increased the score by about 3 or 4.

I also have a 2500+, but I have the FSB overclocked from 133MHz to 200MHz. You could give it a go, if you're not uneasy about overclocking your system.

I get a score of 58 running at 3200+. This evening I'll clock my CPU down to 2500+ and check the score then, if you like.

MfG,

mollmerx

PS: I forgot to mention, I am using a Radeon 9600pro

BlackTiger · Posted: Mon Apr 12, 2004 7:12 pm Post subject:

Wow this is unbelievable ...

I just used the -nosound function

19.491327 / 56.998791 / 148.093552 fps -- rand[1814308868]
Score = 56.262215

Before i got a score of 45.769192

Really crazy

I'll try to compile my own openal now, and after this i'll overclock the cpu to 200 fsb to see if there is a difference too!

Hmm i think i did something wrong

Have no sound in game now ^^ I replaced the openal.so and libSDL-1.2.so.0 from the UT04 System dir with the ones from /usr/lib/(libopenal.so.0.0.6 and libSDL-1.2.so.00.6) Anyone knows what i did wrong? ^^

Thanks

Mfg
BlackTiger

Malakin · Posted: Mon Apr 12, 2004 8:22 pm Post subject:

BlackTiger · Posted: Mon Apr 12, 2004 9:13 pm Post subject:

Ah,

Thanks!

Ok my own openal and libSDL worked, but didn't increase my performance, instead it decreased it by circa 1 point.

So overclocked from a 2500+ to a 3200+ with 1,84 Vore (argh so much ... but system freezes in 3d games with lower core ... good cooling needed

) ... it increased my points from 45.769192 with sound to 54.741550 with sound :wink:

Yeah UT is a really heave cpu usage game

Thanks for ur help!

Mfg
BlackTiger

mollmerx · n00b Joined: 19 Dec 2003 Posts: 41 Location: Cambridge, UK

Well, there you go then.

Running exactly the same system as me, you're now getting similar results. Depending on what you want to do with your system you can follow a few different approaches concerning your CPU clocking.

I'm lucky enough to have a Barton that can cope with 3200+ without a voltage increase (thank god, as I have a superlocked version). When I plan to have a gaming session or think I'm going to be dooing a little compiling, I set the FSB to 200 MHz when booting, otherwise to 100MHz (thereby underclocking a little). That's useful for when I have the computer running at night for downloading etc. I have watercooling and can turn off all the fans in the system then.

Of course, if your system is a server or something like that and you can't afford to reboot frequently, you just have to stick to 2500+. I would think that a benchmark score of 46 is plenty for having fun online. Botmatches might be a little problemativ, though.

Good luck!

foosh · Apprentice Joined: 11 Jan 2004 Posts: 231 Location: STL

how about doing a timedemo style benchmark???? this would be good because the exact same frames would be rendered each time the timedemo was run.

i know how to record and play demos, but don't know how to get in a "timedemo mode" to play them back in order to determine framerate

anyone????
_________________
http://wustlog.blogspot.com

skuld · n00b Joined: 13 Oct 2002 Posts: 69

8.546578 / 28.552942 / 115.486938 fps
Score = 28.546787

Pentium 4 HT 2.8@3.5 / 512 MB PC3200 LL / Radeon 9200 SE / 2.6.4-ck2 / ati-drivers-3.7.6-r1

sokar · n00b Joined: 02 Nov 2003 Posts: 9

Intel Pentium IV @ 3207 MHz
ASUS P4C800-E Deluxe
ASUS V9980 Deluxe (nVidia 5950 Ultra)
1024 MB DDR PC400
2.6.5 Vanilla
nVidia-glx 44.96

filmore · n00b Joined: 16 Jan 2004 Posts: 18

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
AuthenticAMD PentiumPro-class processor @ 802 MHz
GeForce4 Ti 4400/AGP/3DNOW!

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=true -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

4.981127 / 14.836020 / 52.655788 fps rand[1814308868]
Score = 14.838710

-----------
Hehehe

jimbojetset · n00b Joined: 01 Feb 2004 Posts: 67 Location: UK

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
AuthenticAMD Unknown processor @ 2050 MHz
Radeon 9800 Pro Athlon (3DNow!)

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=1 -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

22.614853 / 77.858185 / 240.633606 fps rand[1814308868]
Score = 71.651192

AMD64 in 32 bit mode, not too bad

- c'mon ati with those 64 bit drivers
_________________
Athlon64 3200+, Asus K8V Deluxe, 160GB SATA, 512MB RAM, ATI Radeon 9800 Pro - on 2.6.5-lokean2, nptl, gcc-3.4.0, xorg-x11, reiser4

KKJasmin · Posted: Wed May 26, 2004 1:12 pm Post subject:

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
AuthenticAMD PentiumPro-class processor @ 2370 MHz
RADEON 9800 Pro Generic

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=true -benchmark -seconds=77 -ini=default.ini -exec=../botmatchexec.txt

15.432233 / 37.985462 / 131.072021 fps rand[1814308868]
Score = 37.876015

This is with the latest devel drivers, 3.7.6

EDIT: with 3.2.8r1, a near doubling in score. The strange thig is that i have better framerate in Enemy Territory with the devel drivers...

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
AuthenticAMD PentiumPro-class processor @ 2370 MHz
Radeon 9800 Pro Athlon (3DNow!)

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=true -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

20.250198 / 73.673218 / 198.219406 fps rand[1814308868]
Score = 69.408073

jimbojetset · n00b Joined: 01 Feb 2004 Posts: 67 Location: UK

I am using the 3.2.8 drivers - find them a lot quicker and less buggy. Only reason why i'd use 3.7.6 is if I wanted to use wine.

Kai Hvatum · Posted: Thu May 27, 2004 5:33 am Post subject:

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
Counterfeit AMD Pentium-Pro processor @ 2000 MHz
GeForceFX Generic 5900/AGP/SSE/3DNOW!

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=true -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

Then it crashes! Any clues?

energyman76b · Advocate Joined: 26 Mar 2003 Posts: 2048 Location: Germany

Hi,

AMD Athlon 2000+(1666Mhz)
K7S8X
Kernel 2.6.6 (vanilla)
Nvidia 5336
Winfast GF5200:

bash-2.05b$ cat .ut2004demo/Benchmark/benchmark.log

first run:
12.056080 / 40.724258 / 134.363358 fps -- Score = 40.633823 rand[823438765]

here I replaced the 'sdl in openalrc with alsa:
14.464461 / 37.987484 / 110.160828 fps -- Score = 37.977432 rand[823438765]

here I replaced openal.so and libSDL.so:
16.216087 / 43.653343 / 135.846268 fps -- Score = 43.554951 rand[823438765]

and here I removed openalrc:
13.797054 / 36.127335 / 118.590752 fps -- Score = 36.109406 rand[823438765]

All run, with this forum in a konqueror window

Slide · n00b Joined: 13 Jul 2004 Posts: 16

UT2004 Build UT2004_Build_[2004-02-10_03.01]
x86 Linux
GenuineIntel Unknown processor @ 2133 MHz
GeForce3/AGP/SSE2

dm-rankin?spectatoronly=true?numbots=12?quickstart=true?attractcam=true -benchmark -seconds=77 -ini=default.ini -exec=../Benchmark/Stuff/botmatchexec.txt

14.495691 / 49.816864 / 150.746872 fps rand[1814308868]
Score = 49.475773

Gabriel Shear · Posted: Tue Aug 10, 2004 10:39 am Post subject:

Hi,

whiskeypriest · Tux's lil' helper Joined: 05 Feb 2004 Posts: 91

I've also seen a marked improvement with the latest ATI drivers in UT2004 Final.

Previous benchmark (3.9.0):

Gabriel Shear · Posted: Thu Aug 12, 2004 7:26 am Post subject:

Hi,

what card you have? 9700 or 9500?
i have a 9700 and a Intel p4 with only 2400 Mhz, but i have more points then you?

mfg Gabriel
_________________
man believe what he sees and hears

whiskeypriest · Tux's lil' helper Joined: 05 Feb 2004 Posts: 91

For the record, it's an ATI Radeon 9700 Pro...though I was somewhat amused to see that it's now being interpreted as a 9500 Pro. Then again, I also saw that portage wanted to downgrade ati-drivers to 3.9.0-r1 after last night's sync. I suppose we'll see what the next few days bring.

Edit: My mistakethe downgrade was called due to a dependency issue I hadnt properly resolved. Nothing to see here, move along

As for your higher benchmark score, I'm not too concerned. Without going too deep into the vagaries of system configuration, my card currently runs in a dual-head configuration and I rarely bother to kill any of the information in the secondary monitor when benchmarking; this is primarily due to the fact that I expect this information to be there (especially during multiplayer games) and any variation from the configuration I use while playing would result in a misleading benchmark.

Perhaps you've got a similar situation and I simply haven't optimized and/or configured something correctly. One way or the other, it runs smoothly at 32-bit/1280x1024 resolution with no artifacts and I'm quite happy with it.