Problems with cpu tests in 3dmark03

[Tomay 0]

Just got a new comp. So far everything went good. q3, 3dmark2001SE, sisoft sandra...

 

The problem occurs when running cpu-tests in 3dmark03. Sometimes the screen just goes blank and you can hear the (error) sound. I can't see the mesage but it will probably look like this 3dmark03.exe has generated an error ...

 

The only thing I can do then is press the power button. The windows shuts down correctly (no checkdisk in next boot)

 

I have dx9.0a, 3dmark 03 build 330 (latest), latest bios, cats, drivers...

 

XP pro sp1

ASUS P4P800

Celeron 2.4

2x transcendent DDR (not running in dual channel mode)

Maxtor 6Y080L0

Radeon 9600 128MB

LG burner

[akiss 0]

I HAVE SEE THIS PROBLEM AND I FIXIT WITH INTEL CHIPSET UPDATE...

HOPPE THIS WORKS FOR YOU TOO....

[Tomay 0]

I think I have the latest chipset inf. 5.0 build 1012.

[Mr.Guvernment 0]

aren't you overlcocking that 2.4 to 3.2....

 

if so means your not stable.

 

use SETI@HOME or Prime95 and run your overclocked cpu through it for @ least 24 hours to make sure your CPU is infact stable....

[Tomay 0]

It's stable alright. The same thing happens at stock speed. Beside if the cpu is unstable you usualy can't shut down the windows, you have to pull the plug

 

Have tested with seti@home 3 units in a row (10,5 hours)

[Mr.Guvernment 0]

^^^

 

actually not true, your system can be "unstable" and u can operate fine to some degree, pending on what you run.

 

it is once u run that certain program(s) or something that executes a thread that your cpu can not handle and BAM crash.

 

this is why it is best to run a CPU application like SETI or Prime95 to fully burn in your cpu.

[Tomay 0]

Could be true but not in my case, cos it also happens when not OCed.

 

anyway

 

I'm burning in with seti@home for almost a week (24/7). Case and CPU temps are almost 50°C, and the computer seems stable. So far so good, but I'm not gonna atempt to go beyond 3,2GHz without some aditional cooling. I've read http://developer.intel.com/support/processors/pentium4/thermal.htm that intel recomends temperatures around 40°C.

I also read somewhere that if you take your heatsink off your P4 while operating it'll lower the clock when the cpu gets too hot so you can't burn the cpu.

 

BTW I'm getting higher sandra and 3dmark scores if the cpu has been idle for a few minutes (the temp drops to 30°C). Could it be that my cpu is lowering the clock because it's too hot?

[ViolentGreen 0]

Sounds like a memory problem.... Have you tried running memtest?

[Tomay 0]

No problems with memtest. It's a brand new comp. Memory is not Oc-ed and running inside specs.

[Mr.Guvernment 0]

u tried flashing the bios to a newer one?

 

also u do know when u O/C your chip u do O/C your ram unless u got like pc3200 or 3500 and u have not yet hit the ram max speed limits.

 

do u have 2 sticks of ram in, if so, take one out and try it, just to be sure.

have you check the Event Viewer to look for anything ?

[Tomay 0]

Flashing the bios was the first thing to do. Nothing special in the event wiever. Tried prime95 and it gave me errors with the torture tests. Now I'm running it stock for a few hours (9 or so). Should I increase the voltage to burn in?

 

I'm not overclocking the ram cos it's a 2700. Celeron normaly runs at 2100 wen I OC it runs at 2700 and this ram supports CL2 up to 133 CL2.5 up to 167. Don't see any problem here.

 

How long sould I keep the prime95 running? Do these tests go in a loop or are they random?

[Mr.Guvernment 0]

run any cpu or any burn in test for @ least 24 hours to be safe, the longer u run it, the better, some people go for 48-72 hours.

[Tomay 0]

Found out what the problem was. I ran too tight timings. My ram can't handle 2-5-2-2 and pifast gives me errors. But 2-5-3-2 or 2-5-2-3 run ok. But the ram was very cheap so I can live with that.