errors found in memtest86+

Caitlyn

Reputable
Jun 23, 2014
2
0
4,510
I got a new motherboard (MSI X79 GD45 Plus), CPU (i7 4820), and RAM (2X 8GB HyperX DDR3 1866 CL10). Memtest86+ reported timings of 19-15-15-31. Nothing overclocked. Seasonic 850W power supply. Testing was by Memtest86+ 5.01 booted from CD.

I actually started with Ballistix, but I had 7 errors in the first 18 minutes. The Ballistix had no errors in a different motherboard. In this motherboard, ValueRAM had no errors so I know the DIMM slots are ok.

In this motherboard with they HyperX, the test started ok. CPU temp 39-41. But upon the 42nd hour of testing (inside test 7 of pass 17), 2 errors 1MB apart were reported. The CPU temp was 44. Does this indicate bad RAM or a thermal issue, or can I just write this off as a possible software bug?
 
You vastly overtested. Memtest only recommends a maximum of 7 passes. If it took until pass 17 it's probably not a major issue, even though there shouldn't ever be errors with 7 or less passes unless the RAM has problems. Even healthy RAM starts having faults at some point. It is hard to take that into account to plug it into statistics formulas, though.

Also, every stick of RAM will fail eventually, regardless of its state before initial testing. On Large server farms they get gamma ray failures on a weekly or even daily basis. So if you sample too many times, your RAM will fail.
 
Also, as mentioned elsewhere by another member which reminded me of it, memtest should be run on modules individually, not together. Running memtest on multiple modules, whether in dual or triple channel mode or not, can cause false positives. Run memtest on single modules at a time so that you don't get false positives and so that if there are hard errors you can identify the module that is at fault. If you run it on multiple modules you will just need to run it again later to determine which module is to blame.