Memory allocation error when looping q3 timedemos, memory problem?

[mcbiff]

I was gonna loop q3 to see if my comp is stable at 868 Mhz. All seemed to be fine, but suddenly it crashed. It said something like: Failed on allocation of 1845742 bytes. Does this highlight a RAM problem? Cause I was a bit worried that my noname PC100 could handle 124 Mhz.

The entire system didn't go down, I'm using it to post this. Without a reboot. Q3 just crashed.

 

[NOS440]

Thats a very common error that almost everyone gets after 20 minutes of looping. I don't worry about it myself. I'ved use all kind of different top quality Ram on Athlon and P-3 systems and get the same thing with all of them.

 

[rickn]

if you get the z_malloc error, that is some sort of memory leak with nvidia drivers and quake3 when looping demos. I have talked to Nick Triantos at nvidia about it, I doubt it will ever be addressed, but if you can run the demos for the 35 mins or so, then your card is pretty much stable

 

[garrison]

i've got the z_malloc error msg too (always in the same bytes) either with my CuMine 933/133 and 700/100 and my Celeron at default and OCed speeds.
either w/ my micron pc100 and my old 64 oem module.

so its clear is not a memory or cpu problem.

i was thinking was a motherboard stability issue,
but rikn seem to know the real cause of the problem.

 

[StarCropper]

I get the same error. I'd like to be able to loop q3 all night long to test my system stability; however, that error always crashes quake (usually within an hour when I had 256 MB, now within 30 mins at 128 MB). I was wondering if there was a command that could be added to the demo loop that would clear the memory after every demo run. I tried using the command \clear, but that had no effect.

sc

 

[dman]

If you think it's (nvidia) video drivers then perhaps you can use the software opengl--that is--if quake III even has that anymore. But, if it does, then you won't be using the nvidia video drivers AND you'll be stressing the CPU that much more.

Of course, If it's a bug in the quake code, it could show up anyway.

But, for burining in your machine, you could always use QuakeII to test. I use SETI@Home myself to burnin the CPU's. If it locks up after a few hours, I'll bump the clock back and test again. Eventually I'll find the stable spot.