TNT2 vs 8500LE 128MB

[MikeD2k3]

Hey Everyone

I've got a 1GHZ TB Athlon with 256MB of DDR. I currently have a 32MB AGP TNT2 card. I know this is a very antiquated card. When I read reviews for UT2k3 they talk about how you need such an uber-powerful system to play it well. I run the demo of UT2k3 at 800X600 with all the settings maxed out and I get pretty decent performance. What's up with that? Will I notice much of an improvement with a 128MB Radeon 8500LE or 9000?

Thanks for your help
MikeD

 

[Brian48]

Yes.

 

[EdipisReks]

are you sure you are running with all the settings maxed out on UT2K3 with a tnt2? it would run very poorly at those settings on a Geforce2 GTS.

 

[MikeD2k3]

I have it on 800X600 w/ 32bit color (it's a tad smoother at 16-bit color), with all the pulldown boxes set to highest level and all the checkboxes checked for the Video section in the settings. I've never played on any other system but mine, maybe I literally don't know what I'm missing?

I don't believe the TNT2 supports T&L but I still get lighting affects (it's not all the same brightness everywhere). I don't have anti-aliasing but when I look at screen shots of high-end cards I still see rough edges sometimes. Another thing I have a question on is how do your measure how many fps you get? Is there some kind of macro you can run on certain levels that tests this for you? I remember running something like that for Quake 3 once.

Thanks for your help!

 

[Lonyo]

The demo has lower quality textures, and the max detail settings are not actually the same as in the final release, they don't offer as much quality. The final game needs more power than the demo. I should think if you can get a good deal on the Radeon then you should go for it, unless you want to wait for something like a 9500 (non pro) which should be a little faster and not cost too much more.

 

[MikeD2k3]

Ahh, I see. The two cards I was looking at were these: A 128MB Saphire Radeon 8500LE (http://www.sapphiretech.com/vga/8500le.asp ) or a Saphire Radeon 9000 Atlantis 128MB ( http://www.sapphiretech.com/vga/9000.asp ). The 9000 is $89 with free ship and the 8500LE is $94 free ship at Newegg. I'm not sure which to get. I read the R250 is technically inferior a little bit to the R200 because of the deletion of a texture pipeline. That would explain the slightly higher cost. Are there any benefits to going with the 9000 over the 8500LE? I don't see any special features really.

 

[MikeD2k3]

Just ran 3DMark 2001SE. I scored an 812. I compared it to similar systems to mine w/ 8500LE or 9000. The 8500LE got about 9X my score and the 9000 got about 8.5X my score. Boy do I feel slow.... The first run seemed simlar to what I get in UT2k3 -- it reportedly had about 15 fps. Once you get passed 30 isn't it kind of pointless though? Films at the movie theater are 24fps and TV is 30FPS--why do you need more than that?

 

[Lonyo]

Movies/TV are a constant 25 fps, and it's not quite the same as a computre game. Also, I think TV has a kind of blur, so the motion looks smoothe, even thos it isn't really totally smooth. You'd probably want at least 45fps for smooth game play, but 35 is (IMO) playable. The Radeon 8500 is bein gre-released as the Radeon 9100, so its number is more indicative of its performance. The Radeon 8500 IS faster than the 9000, hence why the new name has a higher number.
You will notice jerkyness at 30fps.

 

[rogue1979]

Try overclocking you cpu a little, 1200MHz is about the lower limit to get most of the benefit from a higher powered video card. My son is running a Morgan at 1350MHz with a Radeon 9000. Unreal 2003 demo plays smoothly at 1600 x 1200 x 32 no FSAA with 16-tap anisotropic. Not blazing framerates mind you, but still smooth. Unreal Tournament averages 100fps at the same settings. The new Catalyst 2.5 is a big improvement in real world gaming performance for all Radeons.

 

[BFG10K]

Yes, there'll be a massive difference between those two cards. A TNT2 is not good enough to run UT2003 at 320 x 240 x 16, much less 800 x 600 x 32. You should also think about upgrading your CPU because it's pretty underpowered, and upgrading your RAM because 256 MB is not really enough for UT2003.

Anything seems good if you've never seen any better.

Another thing I have a question on is how do your measure how many fps you get?

stat_fps gives you realtime framerate statistics.

Is there some kind of macro you can run on certain levels that tests this for you? I remember running something like that for Quake 3 once.

Yes, run the benchmark batch files. You should probably sit down before viewing the results because you'll be stunned at how slow your rig is. Everytime I tell someone how to run UT2003 benchmarks they can't believe how poorly their rig does even though they think it's smooth.

 

[MikeD2k3]

Thanks for the tip. I plan on upgrading the CPU in the near future to something around a 2200+ Athlon XP (that's the current max my mobo firmware support). I also plan on upgrading my RAM. I currently have 256MB of PC1600 (200MHz DDR), I am going to remove both 128MB sticks and replace it with a single 512MB stick of PC2100 so I can run at 266MHZ DDR FSB. If I could run full effects at 1024x768 than I would be very happy for now.

 

[MikeD2k3]

Thanks BFG10K, you caught me while I was writing the response before this. Where are thoes batch files located? Do I have to download them or are they somewhere in the UT2k3 demo folder? I will go try out that 'stats_fps' command now!

 

[MikeD2k3]

I found the benchmark files in the UT2k3 demo folder, couldn't get the stats_fps command to do anything in the console though. Trying benchmarks out now!

 

[LS20]

heck, yes, it makes a difference

my friend and i have similar rigs, except he has a 32mb tnt2 and i have a 64mb 8500le... huge difference!

 

[BFG10K]

couldn't get the stats_fps command

stat_fps, not stats_fps. Or it could just be statfps.

I'm not by my computer so I can't test it but it is something like that.

Trying benchmarks out now!

Be sure to run the individual ones for more meaningful results. The benchmark log file has the results appended to it each time you run one of them.

 

[MikeD2k3]

Thanks for the help! My results were dm-asbestos: 16.28fps and dm-antalus: 18.11 fps. What kind of fps would I get with a 8500LE?

 

[BFG10K]

And now look at your minimum scores (far left); they are the killers. I'd expect something like 2-3 FPS for your rig.

What kind of fps would I get with a 8500LE?

The botmatch demos are CPU limited and upgrading your video card won't help you that much in those situations. You need a very fast CPU with loads of fast RAM to run them well, and even then they'll slow down sometimes.

The video card will help you the most in the flyby demos because they're GPU bound, especially the intensive maps like Inferno, Phobos2 and Facing Worlds.

In terms of actual gameplay, both the CPU and the video card will help so it pays to have them both as fast as possible for the best possible gaming experience. I estimate that a R350 with a 4 GHz P4 should be fast enough to run the game comfortably in all situations.

 

[MikeD2k3]

I ran flyby on antalus and asbestos.

Antalus
avg = 18.11
lowest = 7.18fps

Asbestos
avg = 25.867
lowest = 11.01

 

[BFG10K]

A Radeon 8500 should easily triple your scores in the flyby benchmarks, if not more.

 

[tenoc]

The minimum card cited for UT2003 retail game is a TNT2. An R8500 sounds like your best option.

 

[CraigRT]

your resolution and settings will be able to be pushed alot higher on an 8500 or 9000 card.
I have an 8500 and i really like the way it plays ut2k3
my tnt2 was like 3 video cards ago.. its gotta go.

 

[MikeD2k3]

Sounds like an 8500LE 128MB is the way to go. I will probably wait a little while until they release the 9100, that should drop the prices on the 8500LE quite a bit don't you think?

 

[Lonyo]

The 8500 prices might drop because they'll want to get rid of old stock, so they'll sell it off cheap (I'm guessing). Give it a month or so and the 9100's should appear, then you can find out if they 8500's do drop for sure.