RAM issues? DIMM module issues? Memtest86+ (version 4.20) showing errors on test 6 -

[Green1224]

As of January 2014 I built my first computer consisting of the following:

* Asus Intel Z87-Plus - ATX, Socket LGA1150
* Cooler Master RS850-AMBAJ3-US Silent Pro M 850W Power Supply - ATX, Modular, 850 Watt,
* Intel Core i7-4770 Haswell 3.4GHz LGA 1150 84W Quad-Core Desktop Processor Intel HD Graphics
* Kingston SH103S3/120G HyperX 120GB Solid State Drive - 120GB, SATA III, 2.5"

I purchased 2 memory kits from G.Skill (G.SKILL Ares Series 32GB (4 x 8GB) 240-Pin DDR3 SDRAM DDR3 2400 (PC3 19200)). Windows 7.0 Home edition is the OS. Since January 1st the computer had been running top notch. The Asus BIOS was flashed to vers. 1802. The RAM profile chosen was XMP (as was recommended for this faster RAM).

The compuer had been running fantastic up until the second week of April. The only changes I made to the computer up to that 2nd week was installing 3 more chassis fans (Coolermaster 120mm). Then I started to get BSOD mini dumps happening multiple times a day. After each BSOD, at best, my computer would restart; at the worst the screen would be frozen with what looked like "interference" lines running horizontally ( maybe 3 -4 of them) and my computer would be locked up.

Hence I started to test the computers hardware. I uninstalled everything except the SDD, HDD, and of course the monitor, mouse, and keyboard. I downloaded Memtest86+ 4.20. I started testing the RAM sticks in the following order:

1) in slot A2 I'd test every stick with 3 passes of the Memtest86+ ver 4.20 test cycle, 3 passes each stick
2) no errors were found or recorded (again, just using 3 passes and in A2 DIMM slot)

Next I decided to install all 4 sticks in the proper sequence (ie from one kit I'd install into DIMM slots A1 / B1; the other kit I'd install into slots A2/B2) and re-ran Memtest86+ Vers 4.20 last night for 12 hours with the following results:

3) test #6 recorded the following: (pass 0) 22 errors - error confidence 163
(pass 1) 44 more errors - error confidence 227
(pass 2) 178 more errors - error confidence 213
(also test #9 revealed 8 errors)
(pass 3) 0 errors - error confidence 213
(also test #9 revealed 7 more errors)

I have included 3 screen shots as seen below... I have a fourth but might have to load that later...

At this point I am wondering if this is truly a RAM issue or a mechanical/hardware issue (ie DIMM slots)?

Please let me know what you feel I must do next in order to flush out this problem. Any help would be greatly appreciated....

Sincerely,
Darren

[David (PassMark)]

You say you were using MemTest86+ V4.2, but the screen shots show MemTest V4.3.6.

It looks to be a real error, so I think it is just a matter of narrowing the problem down.

I would initially assume it is a bad slot or bad ram, (bad ram that only fails in dual channel mode).

It is easy to test for a bad slot, just take one stick and try it in each slot, one at a time.

Then if there are no errors trying just 1 pair of sticks at a time.

[Green1224]

Thank you for your suggestions, David. Funny you should mention memtest86+ 4.3 because I had just figured out tonight, after work, how to get version 5.0 up and running through UEFI. I am doing exactly what you suggested and am testing each slot, choosing to run tests 6 and 9 for 6 passes each slot. I'll report back tomorrow with results.....

[Green1224]

Updates: last night I ran Memtest 86+ v 5.0 with a single RAM stick to test each slot. I chose test 6 (5 passes each DIMM). This produced no errors. I then tested the sticks in pairs - again test 6 with 5 passes) again no errors. So I installed both sets of G.Skill RAM, test 6 only; CPU in parallel, then ran the test again with Random and then again with single CPU. The errors were in the multiple hundreds.

I emailed G.Skill with the results and the rep emailed me back immediately offering two suggestions. Either (1) drop the DRAM frequency down to a lower amount (the RAM are 2400hz, 8Gbs each) and if the system is stable, try increasing the frequency a little at a time; or, (2) request a RMA.

i indeed entered my BIOS and dropped the frequency to 2133hz and reran Memtest86+ (ver 5.0) the same way (test 6 only - parallel, random, and single CPU - 5 passes). There were NO errors this time!

being a bit cautiously optimistic about it all, I replied to G. skill and asked why this error took 3 months and 14 days to manifest itself and the rep stated that the random access might have caused physical errors (does that sound like a reasonable reason?).

so, any thoughts on the results of the test and perhaps any more suggestions for me?

[David (PassMark)]

Sounds like it is fixed. So good news there. The small frequency change will barely be noticed i think.


The second law of thermodynamics, as famously enunciated by Rudolf Clausius in 1865, states that: “The entropy of the universe tends to a maximum.”
In other words, everything fails eventually.

There is an interesting article here about how silicon devices age.
http://spectrum.ieee.org/semiconduct...ansistor-aging
Having said that, a RAM stick should last more than 3 months.