Random RAM errors and game crashes

T

TheForumTroll

Weaksauce
Joined
Oct 11, 2009
Messages
105
Hello experts :)


I have some kind of error that I can't seem to find the cause of. I'll try to explain in detail what is wrong and what I have tried to correct it.

The error that started it all was some games crashing to desktop with a message that "The driver stopped responding" (and the name of the driver for my video card). Some games only freeze a bit while the driver is recovered. One of the problems is that I don't know when the problem started since I haven't been gaming in a long time and I have been running Linux until one day I wanted to play some games again and that's when the problem showed it's ugly little face :mad:

So I started to see if I could track down the problem....
First I was sure it must be a driver problem so I updated to a new Catalyst driver for my video card that was just released. Didn't help. Then I tried the older drivers (Catalyst 9.8 and 9.7 i think) after cleaning up my system with some driver cleaner apps. Still no luck. Then I tried to run some benchmark apps to stress my system (Furmark, 3Dmark, etc) but they didn't cause any errors or heating problems at all and I even tried a new install of Windows (went from 32bit to 64).

I then ran Memtest86 overnight. I got a bunch of errors. I then tried some conservative RAM settings but still got errors. Then I tested with one block of RAM to find the faulty one. No errors. Then another. Still no errors. After I tested all RAM blocks i tried with 2 and 2. Last with all 4 in again. No errors! My guess then was that I by accident didn't put the 2 blocks together that was tested to run in a pair and I was happy the problem was gone.

The strange thing is that an hour later I ran the test again just for kicks. And guess what? The errors in memtest86 is back...

I have tried everything I can think of by now. My RAM was running at the specifications they need (1066/5-5-5-15/2.1v) fine and suddenly I get errors again. I'm starting to think it could be lack of power from my PSU since I guess that could also be the reason of my video card driver crashing(??). But as far as I can tell in the BIOS it is fine. Software apps like Speedfan get the voltage all wrong and I don't have a multimeter to test it. Without parts to swap with what do I do next? Buy a new PSU?

Here is my system specs:
Intel Core 2 Duo E8500
Corsair Dominator Twin2X2048-8500C5D v.1.1
Corsair Dominator Twin2X2048-8500C5D v2.2
ATI HD4870
Gigabyte GA-EP45-DS4
NorthQ 650Watt PSU

Oh and by the way I *only* see errors in games. No BSOD, no problems in Linux or in Windows. Nothing..

EDIT: I forgot to say that everything is updated. OS, drivers, BIOS..
 
Last edited:
Ok one - you aren't overclocking and everything is stock correct? I would probably say to bump your NB since you are running 4 sticks...maybe VTT as well. Still seeing any errors, ever, in Memtest is a big cause for concern for me.
 
That is a bad review of the PSU indeed. I knew it sucked when I got it cheap after my old one died but I didn't know it was that bad :eek:


And no, no overclocking with any of the components.
 
Doesn't your motherboard have onboard diagnostic LEDs to tell you what is wrong?
 
It got lots of LEDs but when should I look at them? As far as I can tell nothing is happening that would show up on the LEDs.
 
When it crashes, read your motherboard's manual. I'm too lazy to get mine out ;).
 
So, it looks like a bad PSU? Or does a bad PSU not do stuff like this?

Anyways, I don't want that bad a PSU near my system so I'm looking to get a new one ASAP either way. Which would you recommend of these:

Antec TruePower New TP-750
Corsair CMPSU-750HXEU
Corsair CMPSU-750TXEU
Thermaltake ToughPower 750Watt

:confused:
 
The randomness of the errors you have described suggests that your problems could be power related. If changing the PSU does not resolve it, let us know.
 
Have a look at those readings (I have no idea how accurate they are):
BIOS:
+3,3v: 3,296v
+5v: 5,187v
+12v: 12,429v

SensorView Pro (The first is idle in Windows, the second is when running Furmark):
+3,3v: 3,30 | 3,28
+5,5v: 5,19 | 5,19
+12v: 12,20 | 10,60

:rolleyes:
 
Isn't a 2V drop horrible? Use a multimeter, onboard voltage measurment tools seem to never get accurate readings.
 
A 2.0v drop would likely cause a complete system crash or the PSU would puke as the current rises. I don't think those readings are accurate. Software voltage readings cannot be trusted. Let us know what the new PSU does and, as noted, if you need to make any measurements of voltage. use a meter.
 
I'll second that - whatever OCCT and HWMonitor use, they both claim my TX750 is putting out either 5 vor 16V on my 12v rail. Scary? Sure if you trust the software. However BIOS and EasyTune show a healthy 12.1v. So yeah, there is massive variation based on how software is polling your rails.
 
Now with a Corsair CMPSU-750HXEU, and so far everything seems fine, though I have yet to run a complete over-night memtest86 test with only one RAM stick in. It seems like the error I got when running test #5 with 4 sticks is gone though :)

I tried contacting NorthQ by mail, as they are from the same country as me I thought i might give it a try, and point them to the jonnyguru.com review, but so far no reply :confused:

Thanks for the help and input :)
 
Yeah I wanted something good this time ;)

Even my Server/Gateway got a Corsair PSU (CX400W) even though it might be a bit overkill since it is only a head-less Intel Atom with 2 disks :D
 
You must log in or register to reply here.
Back
Top