Page 1 of 1
3DMark05 is released!
Posted: Wed Sep 29, 2004 7:05 am
by Illuminati
Posted: Wed Sep 29, 2004 7:39 am
by Bio-Hazard
Another one of those huge files that's going to give my sattelite connection max load and use up all my band width for over 4 hours..........

Wish I had some other type of broadband out here.
Posted: Wed Sep 29, 2004 9:22 am
by eric m.
awesome. let's all run the program and compare scores or something.
Posted: Wed Sep 29, 2004 10:54 am
by Apoptosis
3.4 Extreme Edition (Socket 775)
ABIT AA8-DuraMax
Corsair XMS2 5400
Geforce 6800 GT 256mb (w/ 66.59 drivers)
Score 4497
Posted: Wed Sep 29, 2004 2:09 pm
by eric m.
ok nevermind. none of us have computers that fast. you win.
you get a free tshirt.
Posted: Wed Sep 29, 2004 8:00 pm
by Illuminati
2.6 C at 2.88
MSI Neo2-FIS2R
Kingston PC3500
Gainward 5900XT w/ 66.59 drivers
1024
Never got above 4fps on any test

Need to have a bada$$ vid card to rank anywhere on this benchmark.
Posted: Wed Sep 29, 2004 8:29 pm
by eric m.
ok i have run the test no less than 8 times already, each at different settings. i also ran reduntant tests with the same settings to make sure there wasn't any other variables i am not aware of. here are my results so far:
system:
2.8e
ASUS P4C800-E Deluxe
2x 256mb HyperX PC3200 (dual channel)
ATi radeon 9600XT 128mb DDR with latest drivers
ok, i am new at a lot of this overclocking stuff so bear with me if i did something stupid here. all tests were run using "turbo" mode on the mobo, if that makes a difference.
#1 i ran the tests with my CPU running at 3.2ghz and my GPU at the "overdrive" 527mhz setting. the video ram stayed at 297.
scores: 1627, then 1581 running the tests one after the other.
#2 CPU turned up to 3.5ghz, and all the ati settings turned to the highest "performance" settings (turns out this doesn't make a big difference in the benchmarking, but it does make the video look really crappy). i overclocked the GPU to 577mhz and the memory to 312.
score: 1329.
#3 CPU at 3.5ghz again, but all video settings at their default balanced settings. no overclocking of GPU at all (500 mhz).
score: 1230
#4 CPU running at stock 2.8ghz, all video settings at their default balanced settings, no overclocking of the VPU at all (500mhz).
score: 1598
#5 CPU still running at 2.8ghz, "balanced" settings again, but with the GPU overclocked to 580mhz and 320 for the video memory.
score: 1724.
EDIT: ok one last test. i remember rich saying that around 232-235 is a sweet spot for my motherboard so i set it to 235 (around 3.28ghz) and used 580mhz on the GPU and 300mhz this time for the video memory just to see what kind of score i would get.
score: 1582.
looks like keeping the CPU at stock speeds and cranking up the video is the bet bet.
i'm still tweaking things to try and get the score over 2000. this is a really, really resource intensive program, and it makes even fast computers look slow. the highest framerate i saw was 14, and that was just for brief periods. if anyone has any tips or knows why my score is always slower when i overclock the CPU, please let me know. it's not getting hot or anything, by the way. everything is running smoothly and i've never seen this computer lock up, even when overclocking it a lot. thanks for your help guys.
Posted: Thu Sep 30, 2004 7:24 am
by Illuminati
Run memtest86 with you comp overclocked ... I bet you are throwing errors every-so-often which causes 3dmark to have to resend some info occasionally... this causes slower performance times... thus lower score...
That's my guess anyway.
Posted: Thu Sep 30, 2004 9:02 am
by eric m.
Illuminati wrote:Run memtest86 with you comp overclocked ... I bet you are throwing errors every-so-often which causes 3dmark to have to resend some info occasionally... this causes slower performance times... thus lower score...
That's my guess anyway.
thanks. i will try that for sure. so if i am throwing errors, should i lower my front side bus or change the timing on my memory?
Posted: Thu Sep 30, 2004 2:02 pm
by Bio-Hazard
bump the ram volts up a notch first and then run the test again.
Posted: Thu Sep 30, 2004 2:06 pm
by eric m.
Bio-Hazard wrote:bump the ram volts up a notch first and then run the test again.
i can't. the RAM voltage is already as high as it can go on my board.
Posted: Thu Sep 30, 2004 2:28 pm
by Bio-Hazard
Try loosing your timings or the ram ratio.
Posted: Fri Oct 01, 2004 9:38 am
by Apoptosis
I just got this e-mail from my contact at Futuremark and thought I'd share it with the forums as some of you might have interest in what it says.
Hi all,
Thanks to the enthusiasts of our community, we found a way to produce artificially inflated results in 3DMark05. This had to be fixed ASAP, so we made a hotfix for 3DMark05. This hotfix updates 3DMark05 to version 1.1.0.
The list of mirrors for this hotfix is here:
http://www.futuremark.com/download/?3dm ... x110.shtml
Here's the official fixlist which you may post:
* A possible exploit to inflate results in 3DMark05 version 1.0.0 has been fixed
* Updated Entech DLL
* One incorrect unit in the MS Excel importer has been fixed
Please note that in order to submit any 3DMark05 results to the ORB, the
3DMark05 version needs to be 1.1.0.
It is of course recommended that you notify about the hotfix on your full install v1.0.0 download page too!
Sorry for this hassle (or as we Finns call it "säätö"), but this could have been a major issue if we hadn't made & released this hotfix so soon!
Have a great weekend guys! :D
So if you have V1.0.0 they have blocked you from sending in results and only allow V1.1.0 to submit results. My download just finished and the hotfix is only 2.80MB in size, so no biggie!