Errors found in Memtest of 4 modules of ram

LowlySkeleton

Reputable
Aug 5, 2015
235
1
4,860
Hello,
I've been originally have difficulty with OCing past 3.5 GHz on my phenom X6 1055T. However, I realized that most (almost all) of my BSOD's had been attributed to some sort of error involving my ram. A brief look of my system:

Gigabyte GA-880GM-UD2H motherboard

4 x 2GB of DDR3 Kingston 1337 ram, rated for 1.5 V (P/N: 9905471-001.A00LF)

Phenom X6 1055T (125W) OC'ed to 3.5 GHz @ 1.375 V

I had made no changes to any of the settings for my ram, except for the multiplier to maintain a ~1336 MHz frequency. The motherboard only allows for a minimal of 1.6 V for ram and can only be increased from there. The timings were set to auto and resulted in 7-7-7-20-27-1.

After 22 minutes of memtest, I saw ~4200 errors during test 6. I adjusted my timings manually to be inline with the suggested manufacturer timings of 9-9-9-24-33-2. There were no real changes to the results. These tests also seem to say that the lowest error address is at 1.6MB and the highest error address is at 8959MB. For the most part, the system performs fine underload, but does occassionally have BSOD once in a while ramdomly. Also, the blended test from Prime95 showed no errors after an hour of testing.

I know I have overheated my CPU for extended periods of time at one point (unknowingly, as I was ignorant 6 years back) and I know that the phenom II series has the memory controller integrated onto the chip itself. Is it possible that some part of the memory controller is damaged from the overheating? I know at least 2 modules of ram has been tested to be error free, so the chances that all 4 modules of ram are actually exhibiting hardware issues are slim. Or is it possible that the extra 0.1V on the ram is causing more problems? Should I attempt to search for ram modules rated at 1.6+ volts? I would really appreciate some help with making my system a bit more stable. Thanks in advance!
 
Solution
I would suggest your overclocking is the cause of your problems, to prove your ram is not at fault but is unable to handle the overclocking I would make proof by taking off the overclocking applied, and set it back to OEM specs. then runs the memtest again. if the ram fails again, then replace it with ram that is known to be overclocked without issues
I would suggest your overclocking is the cause of your problems, to prove your ram is not at fault but is unable to handle the overclocking I would make proof by taking off the overclocking applied, and set it back to OEM specs. then runs the memtest again. if the ram fails again, then replace it with ram that is known to be overclocked without issues
 
Solution


Thanks for the advice! I ran some tests this morning, after reading your post, and tried some different settings in BIOS.

At all default stock settings (other than Core voltage, because I knew it was stable when I was initially trying to under-voltage before OC), a full run tests 1-10 on memtest returned no errors!
At 230 bus speed x14 multiplier at normal Core voltage, the results from memtest were also negative for errors. I did have to adjust the multiplier for the ram. The ram was running at 1225 MHz. The NorthBridge & HT Link both were at 2070 MHz.
At 240x14 at +0.025V, the memtest was still showing up errorless. The ram was now at 1279 MHz and NB&HT Link both were at 2160 MHz.

At this point, I am certain that nothing is physically wrong with my 4 modules of ram and that it works find at a higher 1.6V (which still bugs me that I can't lower it to 1.5V). I also realized that I had my NB & HT Link was set to 2000MHz in original scenerio, as I read that it was fine (or would actually be better) to keep it at 2000-2200 MHz. However, this article tells me that I actually want to match my OC'ed speed with a certain range of NB speeds to optimally OC.

I tried my 3.5 GHz at +0.050V, and it started to show 78 errors on test 6. So I increased the NB & HT Link from x9 (2250 MHz) to x10 (2500 MHz) and ended up with 172 errors.

Now that I found the problem has something to do with NorthBridge and HT Link frequencies, I think I will try a few more tests with their voltages and multipliers. Thanks again for the advice!