running out of memory, is it possible?

midi_amp

Distinguished
Sep 28, 2002
12
0
18,510
I want to stress test my system by running simultaneously prime95, super pi and 3dmark2003, although the system runs flawlessly... after 30 minutes of running, 3dmark starts to load slower for each scene

I'm fairly new to overclocking scene, so I didn't know what to make out of this phenomena
 

Schryver

Distinguished
Jan 26, 2003
45
0
18,530
What kind of system are you running? If it's P4, I'd venture a guess that you're looking at something called thermal throttling.

But you said just the load times are increasing... is the actual framerate dropping? If only the load times are going up, there's something a bit more esoteric going on...
 

midi_amp

Distinguished
Sep 28, 2002
12
0
18,510
oops... sorry, I didn't post my spec

AXP 1700+@ 175X11 (1925 Mhz) @1.66v /JIUHB batch
MSI KT3-Ultra
Apacer PC3200 256
onboard sound
PixelView GF4 ti4200
all other core setting default/auto

the load time and the actual framerate is dropping, 1 fps and slowing... when it happen, I even consider my machine has failed, but I can still press escape and abort 3Dmark03

what's esoteric means?
 

Schryver

Distinguished
Jan 26, 2003
45
0
18,530
Alrighty... not thermal throttling...

Esoteric: Of or relating to that which is known by a restricted number of people.

Translation: it's not instantly obvious to everybody (including myself) what your problem is.

My next guess is that there's a memory leak somewhere. That basically means that one of the programs you've got running (OS included) is sectioning off memory to use, but is not marking it unused when it's done with it. I'm pretty positive that 3dmark or prime95 isn't the source of the leak, since those are pretty commonplace for stress testing. I'm not familiar with superpi, though...

Regardless, I'd look into getting another stick of 256 for your system. Your system could surely benefit from it when running multiple apps like you are.
 

midi_amp

Distinguished
Sep 28, 2002
12
0
18,510
I see... okay then, I'll borrow another stick of memory, and try the stress test again, if in a set of time the problem become apparent again but in longer interval, I should get clearer data of what to make of this

by the way, just today, I ran memtest86 for 7 hours straight, 1 extended test, and 35 standard test, the system is stable as rock, no errors whatsoever