PDA

View Full Version : RAM trouble... maybe?


TierMann
11-05-05, 11:03 PM
The past couple days I've been getting the wonderful "Machine Check Error (106 corrected)" errors when playing FEAR (1.02). It makes pretty much no sense to me because ram has basicly been running at stock since I got it (mostly running 2mhz under stock). I've never raised the voltage higher than 2.7 on it because this board won't even let me :P. Anyway, each module checks out fine in memtest86 and no driver or software change seems to stop it. I also noticed while checking though the Event Viewer that this same problem (it's close cousin , 107 fatal) might actually be causing the freezes I get with 8x.xx drivers.

For those who know what these errors mean, I've checked the 00010 offset in each case and it sometimes comes from cpu0 and sometimes cpu1.

Should I just RMA the ram? I've got a 3 year warranty through Monarch.
Anyone else have this problem before or ideas for anything I can try before I send it back?

Some might remember that I had random freezes back when I was using WinXP32 SP2 when I first got this setup.
They magicly disappeared with WinXP64 so maybe 64-bit is just better at handling this error in most cases and my ram really is borked?

macatak
11-05-05, 11:41 PM
There was a few posts on the FEAR Forums with people stating they couldn't run FEAR while thier ram was run in Dual Channel, did you run MemTest with both sticks installed or individually ?

TierMann
11-05-05, 11:58 PM
There was a few posts on the FEAR Forums with people stating they couldn't run FEAR while thier ram was run in Dual Channel, did you run MemTest with both sticks installed or individually ?

Both in dual-channel, and then again one at a time each for 6 passes.
I can try in single channel I suppose, but that's not really a fix since I'll lose performance in other apps as well and also I have no bios option for dual-channel. I'd have to open the box each time to switch back and forth, heh. It seems to be only FEAR which makes me think it's a software issue, but then again it's the biggest memory hog I have.

macatak
11-06-05, 01:02 AM
(3-3-3-8) Are those settings the recommended ones for your ram ?

TierMann
11-06-05, 01:06 AM
(3-3-3-8) Are those settings the recommended ones for your ram ?
Yup. That's stock for this version of OCZ4002048PFDC-K (performance series kit). Newer sets run 3-3-3-7 and the only way I could tell them apart is the sticker that actually says 3-3-3-8 on the heat spreader.

Some other info too.. This happens at stock cpu as well. I don't OC for the tests. I keep voltage down as much as possible. 1.375 when OC'd to 2.4ghz and 1.35 at stock 2.2ghz. I clock the ram down to 333 when I OC which brings it back up to ~398 (2/1.66). I don't use CnQ at all and I've got the latest AMD driver which still fixed timer issues between the two cores and the need to set affinity.

macatak
11-06-05, 01:52 AM
Maybe try testing with MemTest86+ if you haven'y allready

http://www.memtest.org/#downiso

TierMann
11-06-05, 03:02 AM
Maybe try testing with MemTest86+ if you haven'y allready

http://www.memtest.org/#downiso
Thanks. I was still using Memtest86 3.0. I'm running +v1.65 now, dual channel, cpu OC'd to 2.4. So far it's on 69% of the first pass.

Updates:
12% of pass 4 now, for those of you on the edge of your seats over my progress.
That's ~3 passes every 2hrs 10mins and I'm going to be now.
It'll probably check out fine just to annoy me further, heh.

And of course 16 passes, 0 errors today. That either means it should be good enough to not be my problem or it just really likes memtest ;). Anyway, I just hope it doesn't pull the same thing when I get the new mobo and 7800gtx sometime in the next couple months.