3D Mark 03 odd results

BeatFreak

Right, I've never really bothered with benchmarks much mostly because there was always something lagging in my system which I knew would give a dodgy result (no I can't handle the truth).

But I recently got my hands on a shiney new Radeon 9800 XT (finally dumped the Nvidia 8) and thought it was about time I got some results.

Downloaded 3D mark 03 V.340 fired it up, looks very pretty and was blasting out frames which didn't drop below 50fps or so, which was nice, but as soon as I got to the cpu tests my smile soon changed to a frown to say the least.

I was getting on average 10 fps. Sometimes less 8( which lead to the poorly score (I think correct me if it's not) 5980. Regardless a 2600xp (2083mhz at stock) OC'ed to 2550mhz (212x12) should be able to get a better output than 10fps should it not? Im aware that it's not the greatest chip (especially for OC'ing) but I thought it would preform better than that.

Please don't tell me that more upgrading is in order (the FX's are a teensy bit way too expensive atm) Or does how does this test relate to real world performance?

Sorry for long and boring post but I'm curious 8).
Thanks dev.

Col_Loki

Most people dont really use the 2003 version as the score is much to heavily based on graphics cards.

Try running 3d mark 2001 and see what that will tell you, its a much more realistic view of how your system will perform in games.

An athlon @ 2550mhz is very fast, comparable to over a 3ghz P4.

Giblet

You should get about a 12000 at the very LEAST in 3DMark2001.

CivilServant

The cpu test in 2003 gets the cpu to draw all the frames in software mode without any hardware acceleration. It's normal for it to be slow, especially in game 3. What score did u get?

jesus_thats_gre

I think the score is pretty much spot on. Go to the Futuremark web site to compare similar spec'd machines and their scores.

Matt Simis

Originally posted by COL_LOKI
Most people dont really use the 2003 version as the score is much to heavily based on graphics cards.

Try running 3d mark 2001 and see what that will tell you, its a much more realistic view of how your system will perform in games.

An athlon @ 2550mhz is very fast, comparable to over a 3ghz P4.

Thats a bit optimistic in this case. For games and especially software optimised for them, P4s are rather good. The "raw" CPU performance of a 2550MHz Athlon XP is most likely higher than a P4 3GHz for "real" work and unoptimised games. When I had P4s, they always shone in 3DM2k3, much more so than 2K1.


Matt

Woden

as a comparison i get 17500 or so in 3dmark01 with a 2500+ at 2.2Ghz and a radeon 9800pro

thats some overclock you have there how do you cool that?

netman

Don't be comparing the 3dmark01 and 3dmark03 results 1:1, they're simply not the same. 3dmark01 is a directx8 test, and not really significant for today's games.

And the score you got on 3dmark03 is pretty damn good. don't worry about the 10 fps, some of those benchmarks are designed to stress out the future setups and will crawl on anything that's out today.

I got a watercooled P4, running at over 3 Ghz, with a watercooled Radeon 9700 Pro and I get a lesser score in 3dmark03 than you. At the same time I used to get over 20000 3dmarks in 3dmark01.

Also, if you're comparing the benchmark results, make sure you compare them at the same patch level. The graphics card manufacturers have been trying to "trick" 3dmark to get a better result, that's been fixed with subsequent patches to 3dmark which will cause even lower scores on some cards.

Woden

Originally posted by netman
3dmark01 is a directx8 test, and not really significant for today's games.

i'd have to disagree with this somewhat in that the sense that a large amount of the games played at the moment are direct x8 i would think

also as col loki said the 2003 is heavily based on the graphics card

unkel

Originally posted by Beat Freak
5980

That is a pretty good score m8 - why you worried?

BeatFreak

I had a feeling I was missing something but I wasn't surehow closely related 2001 SE and '03 were, so thanks for clearing that up. As regards cooling I gotta vapochill all doing the work for me.
As regards why I'm worried, I've spent alot of time (and money) fine tuning this set up and it's a tad dissapointing to say the least when you see your pride and joy running a cpu test at 5 to 10 fps. But I have since been informed that it's ment to stress the cpu and that its not as bad as I thought it was. Which is nice.
From the looks of it it's not a bad score but I've got some tweaking to do possibly grab some omega drivers and I think a reformat is in order too so hopefully I can push that score up another few hundred.
Thanks for the help,
Dev.

shinzon

3d mark 2001 has a patch out now that will test for direct x 9.0 games, version330 its called if thats any help to you

Shin

mr_angry

5980 is a damn good score! I admit, the framerate was poor, but I wouldn't worry about it on the whole.

I got 5775 with a thoroughly un-optimised P4 3.2GHz and Radeon 9800.

Creamy Goodness

Originally posted by netman
Don't be comparing the 3dmark01 and 3dmark03 results 1:1, they're simply not the same. 3dmark01 is a directx8 test, and not really significant for today's games.

have to disagree with this because why is there a patch to enable directX 9.0 support

netman

Originally posted by Cr3m0
have to disagree with this because why is there a patch to enable directX 9.0 support

3dmark 2001 is a directx 8.1 benchmark. it does not support any of the directx 9.0 features. the patch is out simply to allow the benchmark to run on a machine with directx 9.0.

without the patch all you get is an error message that directx 8.1 is required.
as directx are downward compatibile, futuremark fixed this with a simple patch.

Woden

just something else to note but i've just found out recently how driver depend 3dmark scores are.

as i mentioned above i get 17500 roughly thats with cat 3.8. i installed 3.9 and my scored dropped a whopping 6000 points. I'm not sure if this is an issue so much in 2003 version but as your running an XT perhaps you could try the two different driver version and see what works for you.

i'm also gonna give the 3.7's a try.

cheers

data

Gerry

Maybe you should consider playing some real games, and using benchmarks from them, instead of worrying what a corrupt synthetic benchmark thinks. How is quake running for ya?

Woden

Gerry not sure if your referring to myself or beat freak. but if its myself i don't have/play quake, i could use unreal tournaments 2003 in built benchmark though for a real game.

cheers

data

BeatFreak

I haven't run quake in a while tbh but BF and DC and the likes run perfectly at 2048 x 1536 which is all I ask really, all I was worried about was whether it was time for a cpu upgrade or not. I think I'll hold on for another while hopefully the FX's will drop in price soon 8)