Hey everyone.

I was running a radeon 9000pro 64mb. I had a friend who was upgrading his video card so i scored his replaced 9000pro II 128mb. (same brand)

I ran some beanch marnksin ut2003 with my old card. Then swaped cards (deleteled all the old softwear, installed new card, unstalled latest drivers) the beanchmarked again. here is what happened.

64mb card:

under clock (artifacts)
11.728424 / 46.094494 / 159.620041 fps
Score = 46.152962

normal clock 15.617153 / 52.082180 / 159.765305 fps
Score = 52.151428

oc'd 8.403242 / 57.325413 / 173.260468 fps rand[12139]
Score = 57.422844

really oc'd (artifacts)
11.277717 / 59.306988 / 187.166840 fps rand[12139]
Score = 59.405712

speeds of 64mb card: core 275mhz memory 550mhz





speeds of 128mb card: core 275mhz memory 500mhz


new card, no OC
19.683710 / 44.787083 / 151.381226 fps rand[12139]
Score = 44.845081

new card oc'd to old card speed
15.212376 / 47.021442 / 253.746094 fps rand[12139]
Score = 47.095764

new card, max oc: (Artifacts)
10.936104 / 51.244102 / 163.098740 fps rand[12139]
Score = 51.332920



yeh, so if anyone can explain to me why a 128mb car is worse then a 64mb one, or why oc appears have no consistant effect on FPS it would be cool

unreal tournament 2k3/4 trys to automatically scale the graphics within your initial video options...

so if you have only 64 meg of VRAM it uses the appropriate textures for that class of card. 128meg it scales up to the next class so more complicated quality (theoretically, im not an expert on the workings of the unreal engine) its still with in your max/min fps but slightly lower due to the increased ammount of data being sent.

again this is probably why OC seems to have no effect - its automatically making the graphics slightly better

again, im not an expert - i could be just dreaming that such a feature exists - but i distictly remember tweaking this feature using a website guide and UT2004.ini

Yeh, i set all my details manually to maximum to avoid that problem. When i ran un after swaping hardwear it asked me if i wanted to auto config for the new hardwear, i said no.

This was just a little fiddle really. I'm some mad keen beanchmarker, i used ut2003 beacuse it's the only beanch mark utility i have.

I'm not here to sulk that i score 2.43 less 3D marks :p Just interested is all.

i think you'll probably have a 'higher' lower fps with the 128mb card.

I think a majority of cards are faster with less ram, but then new games come out that need the ram and you'll start to notice.

Years ago when i got my Asus GF2MX everyone said it was shit cause it bench marked the same as a GF1 in most benchmarking utils at the time. A year later and i was still using the GF2MX meanwhile GF1s had nothing left in them, they just couldn't handle the newer stuff.

Personally i would stick with the 128mb card really, because if you are planning on keeping it, it will most likely handle the upcoming games a bit better.

at least read what you type before posting it. and how can you live with such a slow videocard? i brought a 9600se for $100 at the fair just for my moviebox shuttle and thats like three times faster than a 9000

im still using a 9200 se =\

i get my 100 FPS in cod, that's all i ask for.

thats because the 4200 is too slow to make good use of the extra ram

Agree with FattyB on this one, ripped this from /.

After seeing the press releases from both Nvidia and ATI announcing their next generation video card offerings, it got me to thinking about what else could be done with that raw processing power. These new cards weigh in with transistor counts of 220 and 160 million (respectively) with the P4 EE core at a count of 29 million. What could my video card be doing for me while I am not playing the latest 3d games? A quick search brought me to some preliminary work done at the University of Washington with a GeForce4 TI 4600 pitted against a 1.5GHz P4. My Favorite excerpt from the paper: 'For a 1500x1500 matrix, the GPU outperforms the CPU by a factor of 3.2.'

Paper can be found here: http://www.cs.washington.edu/homes/oskin/thompson-micro2002.pdf

last edited by Punker at 12:04:45 11/May/04

Alternatively, just create an interface to run 2 GPU's, and create a soft-mod for the instruction sets of a CPU to run them on a GPU (surely there is enough processing power to manage it and still be faster than current CPU's).

Anyway... it would be mad, but it isn't likely anytime soon.

Nice!! That would be sweet. Now, just to get multiple (fast) GPU's happening in a system... Bring on PCI-Express... ;-)