Memtest86+ (Pass,Fail,Pass...) Help.

[scumm4]

My recent build is giving random blue screens. I have run memtest on all four 1GB sticks of RAM individually and thought I found the culprit. Now I am running 2x1GB (while I wait for replacements) which passed multiple runs in memtest. But, the other day I run into blue screens again! After this I ran memtest and it showed 4000+ errors. I have been busy recently so the machine has sat for a while. Well, today I decided to run memtest again to verify the problem and NO errors were reported. I don't understand how the same setup can be fine one minute, broken the next and then return to fine. My voltages and timing have been set to the manufacturer's recommendation the whole time(2.2V 4-4-4-12). Can someone help me with this?

 

[MadHacker]

some details to help solve teh problem
what Motherboard? video card, powersuply
overloaded powersuply?
posibley compatibilyt issues with the motherboard and ram... check MB manufactures site if ram is suported...

also ram may be fine but the data could get corrupted on teh way to teh ram or back... northbridge issues?
does your system run hot?

myself I may not know the answers... but these questions may help others solve your problem....

 

[night_wolf_in]

thats the problem with memtest, it will tell u. " u have a problem". but it wont tell what is the problem. hence the trial n error techniques. i suggest lower the timings (make them looser). and just give it another rest.

just for the info, many mobo's goes into trouble running 4 sticks of ram.

yet again. u have to give more info on your trials. it could help locate the issue.

 

[3Ball]

My recent build is giving random blue screens. I have run memtest on all four 1GB sticks of RAM individually and thought I found the culprit. Now I am running 2x1GB (while I wait for replacements) which passed multiple runs in memtest. But, the other day I run into blue screens again! After this I ran memtest and it showed 4000+ errors. I have been busy recently so the machine has sat for a while. Well, today I decided to run memtest again to verify the problem and NO errors were reported. I don't understand how the same setup can be fine one minute, broken the next and then return to fine. My voltages and timing have been set to the manufacturer's recommendation the whole time(2.2V 4-4-4-12). Can someone help me with this?

I would personally RMA all of the sticks of memory and if that doesnt fix the problem it could be your motherboard is acting a bit finicky. If you could list the rest of your system specs it could help us out.

Best,

3Ball

 

[Zorg]

If your RAM is this then I have 2 sticks in three machines. I actually had one stick fail and then come back to life and then fail again 2 weeks later. I switched both sticks with another machine so it was not machine related. I suggest that you run Memtest86+ on each stick for several hours, to see if you can get a failure. After you get a good stick, then use that and then the second during the day and memtest at night. It is unlikely that you got 4 bad sticks. I have a DQ6 mobo and that RAM works well with the GA-P35. Even if you find no errors I would run only 2 sticks for a while to check for stability. Run the sticks at 2.2V.

 

[ZOldDude]

Ultra X-2 750W Power Supply

Shame to run all those other good parts with one of the worst PSU brands made.

Two years ago I had a case of Ultra 500's and had one fail in under 24 hours running a system 100% stock durring burn in...saw the issues other had online and sent the whole case back and will never use that brand ever again.

Anyhow if you end up sending the RAM back (see bad reports on Newegg) get a refund and grab some 4GB sticks of G.Skill...they also come in 8GB or 16GB sets and still run @ 1.8 volts @ 5.5.5.15 as well.
http://tkc-community.net/forum/index.php?topic=7408.0

I have no idea how well they OC but if thats what you want to do other G.Skill ram is always a good bet and a good price.
The Dutch seem to know what they are doing more so than Corsair or OCZ and have far better prices.

The seven systems on my game LAN in my profile are all using G.Skill and running 50% OC with stock volts (MB/RAM/CPU/Chipset) on air and low temps (same as H2O systems).

 

[Zorg]

I agree that PSU is crap. I haven't seen the reviews on Newegg. I'll have to look, even though you have to take them with a grain of salt. I did get one bad stick out of six, so that's not really a good percentage. I just passed it off as the luck of the draw.

 

[scumm4]

As far as the PSU goes, would a crap PSU be messing with voltage regulations (causing random BSOD's one minute and working fine the next?). All the reviews I read of this particular PSU were favorable and I also wanted some extra headroom for future expansion (a case of you get what you pay for?). This is my first build so I had trouble deciding on quality parts. What PSU would you recommend? Finally, I ran two sticks in memtest for 5 full passes without error, though this still doesn't convince me. All temps are normal? (CPU 27-30 at idle, 40 max stress) Memory is that which is recommended by Gigabyte at recommended settings. Nothing overclocked.

P.S. The last time it BSOD'd the video driver was corrupted so I uninstalled it and installed the most current off of the nvidia website. Is this a sign of the problem?

 

[Zorg]

The power supply could be your problem and it is certainly junk, but I don't want to jump to that conclusion immediately. 5 passes is a good indication that the RAM is OK. The RAM voltage is not dictated by Gigabyte it is dictated by Crucial. They make the RAM and they determine the voltage that it should run at. I assume you have the RAM that I linked to. That RAM should run at 2.2V or a setting in MIT of +.4V. That is because the setting is added to the default of 1.8V, so 1.8V + .4V = 2.2V. Do you know how to manually set the RAM voltage or do you need assistance?

 

[scumm4]

Yes the RAM you linked to is what I have. I had already manually set the ram voltage and timings in BIOS at the onset of build completion. I also had flashed the motherboard BIOS (to F2) because out of the box it was causing all sticks to fail memtest. What power supply would you recommend if at some point I would need to replace it? Thanks.

 

[MadHacker]

myself I use CORSAIR CMPSU-620HX
http://www.newegg.com/Product/Product.aspx?Item=N82E16817139002

easy to work with cables... Modular..
I like it and would buy another...

 

[RetiredChief]

To bad CPUID does not provide a log file. You might try running CPUID, verify initial voltages are real (Not showing 8 V for 12V). Leave it running for 5 min at idle and not noninal/min/max.

Load up prime95 (I used TAT) and load the cpu Note changes in voltages.
stop Prime95. My Vcore dropped from 1.20 to 1.18 or 1.8%

Load up a program to task the GPU. I ran 3d view in ATI Tools. This time my +12 V dropped to 12.10 to 11.85, or 2%

No effect (from TAT nor 3dview) on DDR V 2.11V, 3.3 V and 5V
DDr V @ idle 2.11 -> 2.10
3.3 V 3.23 solid - 2% low
5 V 4.92 -> 4.95

PSU supplies should be within +/- 3 %
What you need to what out for is (Especially on lower quaility PSUs) that the nominal is close to the 3% low, ie +12 @ 11.75, and then under load it may periodically spike down to 11.5 V

Edit Added -- If a voltage changes to way out of range and computer did not turn to jello - IGNORE it ( ie my vcore dropped to 0.02 V ) If a voltage goes way out of range, your computer is going to Blue screen/shut down and you will NOT see this drastic of a change.
END edit

NOTE - Note all monitoring software is accurate, but only other option is buy a DVM from Radio Shack (maybe Newegg) with computer interface and set it up as a data logger. Bought 2 of them from Radio Shack and they work great even had one calibrated while working @ NASA