5900SE Users: What do you get in 3dMark03?

Grammar Nazi

Limp Gawd
Joined
Mar 9, 2003
Messages
282
I've just finished overclocking my components, and after extensive testing, the highest I can get is around 5600 in 3dmark03 and around 40,000 in aquamark3. This is also with a zalman cooler w/ the optional fan. Clock speeds are 480/790. My 2400+ is also overclocked to 166x14 = 2333 mhz. (I'd overclock it more, but I am using shitty kingston valueram, and it dont work good). How does anyone else's score compare to this?
 
wow higher than my score and i got a 5900 flashed. i only get 5400 i think its because of my ram though.
 
Nexu: Is your 5900SE overclocked?

my score is on 53.03 drivers, I just realized that I remembered hearing that someone thought they were "cheat drivers". The reason why I used those drivers is because my card is extremely flaky when it comes to letting me overclock (It will only let me about 1/10 times I try) and with the newer drivers(56.whatever) it wont let me at all.

Aquamark3 scores would be good to hear as well.
 
yes its slightly overclocked. I think at stock I got 4800, my card didn't overclock to well, my best score was at about 440/750. Higher than that I got artifacts and lower scores. My computer is pretty old so memory, pcu probably isn't helping (ino, ino 3dmark03 is all gpu :)
 
The highest I have got so far is 5600.
I may bump my cpu up some and also try bumping the GPU up some more and see what happens. I have had both CPU ands GPU faster than I have it now but keep them below the top end on a day to day thing. I have had this MSI 5900XT for a week tommopw and have had it to 480/910 I think it was. At that speed it ran ok until the last test in 3DMark03 and system rebooted at the start of last test. I figure it was because the ram got to hot. I read a review on this card that they had the same problem at about the same speed.

I ran it for 2 days at 475/840 with no problems at all. But until I get some ram sinks on it I want to keep it running at a slower speed so I do not have to worry about messing anything up.
 
I would say that the score he linked to is a regular 5900 and not a 5900XT or se. Wish everyone would use the same thing. XT and SE cards are the same cards but it can get comfusing to a lot of people when they see the different terms.
 
I am using an eVGA 5900SE. You can tell by the memory clock speed (remember, SE/XT cards have the slower 2.8ns memory). I was at 450/760 for that benchmark.
 
Originally posted by micah
I would say that the score he linked to is a regular 5900 and not a 5900XT or se. Wish everyone would use the same thing. XT and SE cards are the same cards but it can get comfusing to a lot of people when they see the different terms.
My evga5900SE (yes, it definitely is, its not a regular in the wrong box) shows up as a 5900 in the BIOS and display properties because there isnt a driver for it.
Originally posted by onetwo
I am using an eVGA 5900SE. You can tell by the memory clock speed (remember, SE/XT cards have the slower 2.8ns memory). I was at 450/760 for that benchmark.
damn, I would like to know how you got such a higher score. Do you have fast writes on, have you done anything else?
 
Um, GrammarNazi also used the same 53.03 drivers, so the scores are comparable. And BTW, Futuremark already 'fixed' the NVidia 'cheat' with the 3D Mark 03 ver. 330 patch.

Key stats:

-2.8C @ 3.03GHz
-FSB217 (DDR433) 1:1, 2-3-2-6, 2.8VDimm
-eVGA 5900SE 450/760 :D

And trust me, the card has real world performance to match.
 
Handy dandy chart I made:
5950vs5900bios.gif


The top row makes me sad. Why run a 5900SE/XT at default speed? ;)

The reason I posted the whole set was because I wanted to show it's not always best to go for just the highest core speed, especially if you're going to flash the 5950U BIOS to do it. Stick with the default BIOS and overclock for the best performance. These benchmarks were run on a BFG 5900 (default: 400/850)/P4 3GHz system in my sig.

I found it strange that the memory with the tighter timings using the default BIOS is virtually the same as with the 5950U BIOS. 980MHz was the max stable for me with either BIOS.

The core overclocked better with the BFG 5950U BIOS, but I think I'm going to stick with the BFG 5900NU BIOS. <3 my $170 BFG 5900 w/VIVO.
 
I dont think anyone here isnt overclocking their 5900SE as much as they can... there is no way in hell anyone is going to get to a 950mhz memory clock with the 2.8ns ram in it.

onetwo- I am only left with the assumption that your score dominates everyone else's because you just have a fast computer with fast ram, while I do not.
 
strange.. when i ran 3d mark on my 5900 ( NU ) i got a score of 5111 a tad bit low. then i flashed my bios and got a 5463. i only got 512 mbs of ram ( not duel chanling ). i think thats why my score is so low but diffent story for me pxc but very interesting.
 
wohhhhh 6012, what drivers are u using, i get 5888 with my bfg 5900nu flashed to a 5950 ultra at stock speeds of 475/950.
 
Originally posted by chuckm1020
wohhhhh 6012, what drivers are u using, i get 5888 with my bfg 5900nu flashed to a 5950 ultra at stock speeds of 475/950.
There has to be some "cheating" there to get 6000 off a 390/700 or 400/700 5900XT. With default settings in the driver (Quality), it's impossible to get 6000 on that card.

Bodega: upload your project. It's easy to tell the true core speed (at least) from a published project. I doubt it was at 390MHz or 400MHz to get that score.
 
when i ran 3dmark03 i was using 53.03 and i wasnt online ( i wouldn't have ran the test if i wasnt so bored that i put my roommate's card in my computer) lan in whole dorm was down. I'll see if i can borrow the card again tomorrow or something and i will give link to the project
 
Originally posted by Bodega
6012 all stock speeds with a 5900xt

3.0c, 875p, mushkin 2-5-2-2
wow, the 5900XT/SE appears to really like intel chipsets with fast RAM.
 
do you have a cold? i already said i would as soon as i can get the card back
 
Originally posted by @trapine
Stock cough proof please cough!
apparently you missed earlier when someone with a score thats 100pts higher posted his project link.
Originally posted by AMDoido
http://service.futuremark.com/compare?2k3=2144161

With a XP2000@ 1790Mhz... (fsb: 143)
256mb ram...
And the XT running on: ~430/830
828mhz memory? Thats pretty damn high, but your score isnt that great.




CONCLUSION: Fast/good ram makes a huge difference for some reason with this card
 
i get 4812 with stock 2500+ and stock XFX 5900XT. I have reinstalled the OS and will try again to see if that made a difference.
 
Originally posted by blvdKing
i get 4812 with stock 2500+ and stock XFX 5900XT. I have reinstalled the OS and will try again to see if that made a difference.
thats what I got with my card on stock, there isnt a worry. I get a very large improvement in score from overclocking.
Originally posted by I(illa Bee
can i come and play too?
k
 
Originally posted by Killa|3yte
why dont you run it with approved drivers. ;)

I agree. If you're going to take the results seriously, then you need to use approved drivers. Not some that falsy inflate the score.
 
Originally posted by fallguy
I agree. If you're going to take the results seriously, then you need to use approved drivers. Not some that falsy inflate the score.
what drivers are approved?
 
Originally posted by Grammar Nazi
what drivers are approved?

I did this a few weeks ago:
3dmark03.gif


There are no obvious quality differences between 3DMark03 340 on 52.16 and 53.03. But fanboys like to complain. :p
 
I save my 3dmark score to file whe i didnt have internet for a wile, (*.3dr) how can i upload it?

its 6,930 with aproved drivers...

BFG5900nu @ 610/1020 (1.62v core / 2.81v memory)
 
when I gave you permission to play, I didn't realize you were using an INFIDEL VIDEO CARD.

Also, what kind of system ram / speeds are you using? I want something to back up my theory that system ram speed has a lot to do with 3dmark03 performance while clock speed doesnt have a lot to do with it.
 
Its not about quality, its about taking shortcuts, that are against the rules. Drivers cannot detect the benchmark, or they are invalid.

Its not odd that going from 52.16 to 53.03 gives you 1k more or so 3dmarks, but no more frames in games?

Nice name calling as usual. Its too bad they dont enforce the rules around here.
 
I think the point that you are missing is that I dont give a shit about how my card compares to different cards, I want to know if it is as fast as it should be.

Also, if there is no quality loss, what is the problem?
 
Back
Top