Tips for GF2 Ti450?

[nailbomb]

Jah its old, don't bag on me ;) Gainward Golden Sample. I just wanna see what I can get, new to 3d benching.

Running on a 1.8a@2400 mhz 2-2-2-5. Will have a 2.8C and a DFI 875B to play with on Monday, and hopefully I can avail myself of that cheap 9800Pro over on Newegg. Then the 1.8a will be relegated to filserver/Folding box.

So:

What Det's should I use?
What RivaTuner settings?

Diagnostic Startup/Realtime Priority I gather?

Any other advice?

Right now I"m getting 280/450 card clocks on stock cooling, with no artifacts.



Thanks gents, and cheers!

[lalPOOO]

I've acheived the best results with an LOD of 15. I'm still not sure excatly what dets are the best, however I find most of them give a pretty similar score. Prolly want to set prerender limit to 2...
You'll have to play around with it a bit to see where you are at. (I'm smart eh? :D )

[nailbomb]

Thanks :)

Anyone else?

Here's what I was able to get with card clocks/diagnostic startup. DX9 with latest Det's, which I'm sure probably hinder my score. Like its anything to write home about anyway ;)

http://service.futuremark.com/compare?2k1=7555993

[Geforce4ti4200]

I am gonna try my geforce2 ti at 250/440 but I doubt ill break 5k with a much faster cpu. sucks I got the sdr verson. my dragothic high details probably will be in the low 30s

[nailbomb]

I manged to pull 275/500 card clocks earlier, but it scored barely over 5k.

Geforce,

Any tweaks you could suggest? I am new to Rivatuner so....

[Geforce4ti4200]

set lodbias to 3. you got near 6k, not bad. too bad I have the sdr verson, my clocks arent far from yours but your blowing my score away. see my new thread

[nailbomb]

+3?

I just yanked 6143:

http://service.futuremark.com/compare?2k1=7556977

This is allready getting addictive ;)

[r3b0rN]

benchmarking is VERY addictive, well not just the benchmarking, but the overclocking and tweaking that come with it as well.

[nailbomb]

Score: 6152 nailbomb@xtremesystems.org
User: nailbomb1@hotmail.com
Res: 1024x768 32bit
OS: Microsoft Windows XP

Date: 2004-2-14
CPU: Intel Pentium 4 2484 MHz
GPU: NVIDIA GeForce2 GTS/Pro/Ti

2nd when doing compare same CPU/GPU :)

305/515 clocks. If I exceed those, I start to see artifacts when I get past the car tests. What's the solution to that?

Lower core?
Lower memory?
Better cooling?

Cooling on the card is stock. Does anyone know if the Blorb/Crystal Orb or the Vantec Copper Icberq will fit on my card? Kind of a general question I know.

Damn, THREE more points and I have it, by CPU/GPU!

http://service.futuremark.com/compare?2k1=7557763

[lalPOOO]

Next step is to put better cooling and then do a volt mod. Also try raising your 3.3v line, because it probably controls mem voltage on the card. You would also gain a fair amount of points if your cpu was faster, but this is just a backup system right?
http://service.futuremark.com/compare?2k1=7481927
compare to that and you'll see what I'm talking about with the systems thing.

Also, I have a sneaky suspiscion that all the geforce2mx cards are just failed geforce 2 ti/gts cards, because on both my cards whenever I pass a certian core speed I get checkerboards in games. Not sure why that affects you in anyway, but I figured I'd post it anyways.

also, you know you only need to run the game tests right?

[nailbomb]

I got it: 6274 :)

http://service.futuremark.com/compare?2k1=7561492

[zakelwe]

Try 40.52 for XP, try 42.01 as well.

Move LOD to 3.8-4.

Regards

Andy

[Mathias-K]

Originally posted by lalPOOO
Also try raising your 3.3v line

How? :confused:

[lalPOOO]

I would/did do it using the mod to the 3.3v sense line on the power supply... Check the Xtreme Mods section, theres a bunch of threads about it. Oh and btw, Mathias-k, I cleared your 2k3 record by a bunch more :P

[nailbomb]

Suggestions on test order anyone? :)

It sounds like Car Low can give rather variable results....

[lalPOOO]

Well I thought the same, and so I decided I'd find out excatly how much. On my card, which is pretty similar. Most of the tests tend to vary by about a max of .7 fps for me, and so I just set the machine up with clocks very close to the max and did about 5 runs on each test (individually) and writing down the highest result. Then I ran the lobby high (because I was having stability problems with it alone and needed to get it out of the way) and then went on a ran a couple more test and so on, until I had all the tests done with results. Then I went back and checked for low looking scores compared to what I had written down.

Timetaking? yes.
Sucessful? yes.

I also found having 3dmarks priority set to realtime did decrease the amount it varried by.

I havn't worked out any super test order, but I havn't seen a large difference running it this way, and its the only way I can get the higher marks, so I really dunno.