First experience with MemTest86

[Andrew Bauer]

Initially one has no idea of how MemTest86 should run ... it would be helpful if you could describe what should appear on the screen, i.e. normal behavior.

My hardware: ASUS G73JW notebook (purchased Dec 2010) with new Samsung 850 Evo SSD. Latest UEFI BIOS for Windows 7, reportedly Win 8 compatible. 8 GB RAM. Intel Core i7 Q 740 @ 1.73GHz
Symptoms: strange errors in Windows 10 after upgrade from Windows 7, e.g. "The instruction at 0x00000000DB4E75FF referenced memory at 0x00000000DB4E7FF" (same address); occasional hangs during boot; unexpected errors; unexpected behavior.

With UEFI boot initially disabled, I ran v4.3.7 and saw it stepping through each test with a rotating State character. No errors.

With UEFI boot enabled, I see two lines in the boot options, both "UEFI: LG USB Drive 1100". Why two? It takes ages to reach the initial splash screen. Then it freezes almost immediately on Pass 0% Test 36% on Test 0 Pass 1/4 and neither ESC nor (c)onfiguration react... even after waiting 5+ minutes.

I'm puzzled why v4.3.7 runs happily with no errors (beyond the first iteration) and v6.2.0 Free hangs almost immediately. Could this be a sign of faulty RAM; a bug in v6.2.0; or have some other cause?

[David (PassMark)]

I see two lines in the boot options, both "UEFI: LG USB Drive 1100". Why two?

Both options are the same. It is an artifact of getting the boot drive to boot correctly across a wide range of machines.

The other UEFI boot issue could be a BIOS bug. Can you check if there is any newer motherboard firmware available for the machine.

[Andrew Bauer]

Hi David,

Thanks for your response. The BIOS is the latest and greatest v205 maybe from 2012. I bought the notebook in December 2010, so it's almost 5 years old. There will never be a BIOS update for this old model.

I just disassembled my notebook and gave it a good clean to let it breathe properly. I can't believe this ... but now the tests in v6.2.0 actually run!. CPU temp is 62C (before I remember it was 75C). So maybe v6.2.0 checks the CPU temperature and simply stops the test at some critical point? In which case a warning message "your CPU is about to melt down" or similar would be practical ... to let the user know that MemTest86 decided to stop running it. I assume this because I was able to run all tests using v4.3.7 before cleaning, and both vents were really blocked.

Greetings to Australia!

By the way .. the ISO image can't be burned to CD. I insert a CD-ROM and Nero claims I should insert a DVD! (which I didn't try)

[David (PassMark)]

MemTest86 doesn't have any feature to stop testing at high temperatures.

But maybe the hardware itself was failing due to high temperatures. (Not that 75C is all that extreme).

[Andrew Bauer]

Today I find that it depends which test I run in v6.2.0: as before Test #0 stops at 36%. So does Test #3 though it runs a little longer. Test #8 too (it runs for 0:02:00). Test #9 runs the longest (0:08:14) and *also* hangs at 36%. Test #13 seemed to hang at 0% but later I saw "Settings: (1) Skip Current Test ..." though I wasn't able to select any option. For every test I have to reboot MemTest... this takes over 5 minutes.

Most tests just hang/freeze once "Test 36%" is reached (and Pass 0%).
This fooled me yesterday ... I apologize ... I ran Test #9 and thought all's okay now. Have you ever seen these symptoms?

The RAM benchmark runs to completion. Multiple CPUs can't be selected due to limitations in the UEFI firmware. The CPU temperature I mentioned was as displayed in any test within v6.2.0 ... it doesn't seem to update during tests.

As before, all tests in v4.3.7 run to completion in one Pass. I'm puzzled as to what can cause the same tests in v6.2.0 to fail. To me it suggests a possible fault in v6.2.0 as it's 100% reproducible (not a random memory fail). v4.3.7 gives me confidence my RAM's okay.

I'll just use the notebook as it is and wait for the 6th gen Intel Skylake processors to come on the market.