Right ladies, Let's see all the tweaks you've got for your machines for getting that ubar1337 3DMark score Currently sitting on 11234 3DMarks using: GeForce 4 Ti4400 @ 320 / 620 (still some tweaking required with THAT clock speed ) ^--- Using Omega Tweaked 29.42 Dets AMD MP1600+ @ 1.673Ghz (Unlocking should be happenin soon, get a few more mhz then) 256Mb PC2100 Crucial @ 159Mhz I've done absolutely everything I can think of to optimize before I ran the 3DMark test, defragging, closing EVERYTHING running in the background, etc... but still think I might of missed something that you peeps can point out to me...
If you take all your PCI cards out, you might be able to get a higher fsb Obviously, this isnt a permanent measure
Might invest in a block of copper and a big vessel of LN2 aswell at this rate XD I have a very small spare HDD that I'll wack XP on tomorrow and use it as a Benchmark disk
Where 3DMark is concerned EVERYTHING matters! The board leaders individually tune their system for each benchmark in the suite. Turns out you can do that. My impression is they also run them over and over again when they're unstable, waiting for just the right run to make it through. These guys think nothing of wiping these systems clean and going for it, just to win. Some have a spare hard disk. One has their day to day stuff, the other is strictly for benchmarking and is almost bare. They may run a totally different OS. I've heard a different OS is faster for AMD versus Intel. Enjoy!
Hmm, interesting. THanks for that, I might go run a few searches and find out what OS/CPU combos seem to get the best results etc. If I find anything that looks like it might be useful, I'll post it
Setting the resolution in Windows to 1024x768 and the refresh rate to 60Hz has always given slightly higher scores for me. Plus obviously disabling the splash screens in-between tests. Also disabling all of Windows' fancy visual effects and having no desktop background might help.
Setting the resolution to 640x480 with 16bit colours gives a much higher score, but some people (myself included) consider this cheating An example: defaults give me just under 4000, in 640x480 I get 5700.
Low res benchies aren't cheating--Mad Onion identifies them as such. It'd only be cheating to take a bench at low res and then tell your buds it was a high res. But said buds can find that out in two shakes looking at the board. BTW, you gotta test every driver set you can get your hands on for the vid cards. You don't always want the latest greatest. Ditto for other drivers. Lastly, everyone says the way to win 3DMark is cranking the FSB, not the cpu. Unlocking is therefore a requirement to serious contendas!
