3dMark2001 quitting early

[supraway]

When I run 3DMark2001, the program quits early -- in the first test. I can run ANY test except the first car chase test. I originally figured it was because I am running all my stuff on a generic 300 watt power supply, but my friend with almost this exact setup (except he has a 431 watt power supply and Iwill KK266 mobo) has his computer quit early too. I tried this with Win98 and Win2k, but both quit. I am using the latest drivers from the Nvidia homepage (12.90 I think). The only thing it could possibly be is the video card, the Detonator drivers, or the sound card. Has anyone else had this problem?

[Subzer0]

YES me too and im also running the IWILL kk266,1 week old ,first installed 3dmark 2001 and it worked fine then maybe the next day or so tried to run it again and as you explained it would do the 1st car chase and stop in the second. it would just close up and go back to my desktop.
we should all get paid for the testing we do!

[Paul.K]

Hi guys I'm getting the same problems you've described. 3DMarks2001 quits after the jeep chase and return to windows. However 3DMarks2000 has no problems with benchmarking. Since we all have IWill KK266R boards maybe its a bug with the board....

[supraway]

I was thinking the same thing. I get my new 431 watt power supply on Monday, I will try it then... it may be a power issue, as I have read that some Iwill KK266 boards have that problem. I guess I'll investigate it :)

[Burning Phoenix]

I have the same problem you guys have except it only happens when i overclock. I have a P4 1.7 GHZ whichs will go through the 3dmark2001 test fine until i oveclock to 1869+ then it runs the test approx. 10 secs. then goes to the desktop as you mentioned. I'm thinking it is a PSU problem and i'm using a Genermax 350 Watt unit which may not be enough!

[Monaco]

Looks like I'm not the only one with this problem after all! My 3dmark2001 would lock at the same part in the 2nd jeep scene, no matter what I clocked at. I nuked it and forgot about it till a few weeks ago; reinstalled and it work ok now.

God knows why. I don't.

[Sohryu Asuka Langley]

May be a currupt install or too high DMA bus.

[redduc900]

This might help...it's taken from the KK266 faq (http://24.147.235.112:8082/kk266faqx.html)

3D Mark 2001 Crashes to Desktop or Locks Up -
cjinsomniac gives us his experience and solution:
I thought I would pass this along, since I have seen a number of people having issues with 3DMark 2001 on certain KT 133A boards. Anyway, I couldn't get 3DMark to run on this thing.
It whipped my A7V in 3DMark 2k benches by 600 points or so, bringing me to 10,439 marks with a 3D Prophet II 64. But no 2001. It would just crash to the desktop or lock. Even at low memory timings and default clockspeeds, I could only get it to run about 60% of the time with WHQL drivers only...And no core overclocking at all, just a bit for the memory at times.

I finally discovered that the January BIOS would run it stable at fairly high clocks if I used WPCRSET to set the PCI latency at 64...96 for really tough drivers.

I also used WPCRSET to set my DRAM timings up, and now have reached 96% of full memory speed available in the 0307 and Turk OC BIOS images.

Even setting the 52nd register at EB and latency did no good with any other BIOS.
Something about the January BIOS, along with the PCI latency hack, did the trick.

On my board, the CPU idle changes the 5v value from 4.6 to 5v, which has got to be better than 4.5 or 4.6...Also cools it down a bit of course, which is never a bad thing.

Simple setup for those with similar problems on the KK 266 :

With WPCRSET, set register 75 to A2 for 64 clocks PCI latency or A3 for 96.
Set 64,65, and 66 to 12 for 5-2-2 CAS 2 Dram timings. Actually, 64 is for DIMM 0, and empty slots can be set to 11 (reserved) or
set to 12 (no harm).

64=DIMM 0
65=DIMM 1
66=DIMM 2
67=DIMM 3 (Nonexistant on KK 266, leave it as it is).

Mine are

64-data=12
65-data=11
66-data=11

And 3DMark 2001 will rock and roll.

Another user, Cathar, reported the following for NVidia Detonator users:
A lot of people have had problems with the 10.xx, 11.xx and 12.xx series of NVidia Detonator drivers on their KK266 boards when using the 0307 or 0426 BIOSes.
With the help of cjinsomniac and mskreis I managed to track this down to register 55, bit 7 in
WPCREdit. If this bit is set to 1, then 2001 fails, if it is set to 0, 2001 works to at least being able to run the benchmark numerous times without lockup/failure. This has been verified by 5 separate people so far. If you're still having troubles, it might be due to other reasons (AGP driving strength, RAM or CPU overclocked too far, bad power supply, etc). For a large number of us who reported the problem, the reg 55 bit 7 set to 0 trick seems to have solved it for us.