MemTest86 fails at Test 3

[BostonT]

I keep getting failures on Test 3 (v4.2.0 & v4.3.0 Beta). I have tried new RAM. Different brand of RAM. A couple new motherboard. Different processors. Also, memory transfer rate fluctuates. Rates drop down to ~4000 MB/s (Same RAM. Just after a reboot and restart of test). Rate will not go back to 20K+ MB/s unless the power is unplugged from the system.

Memory will FAIL when testing memory modules individually
Memory will PASS MemTest86 v4.2.0 if the CPU was set to Round Robbin & Sequential
Memory PASSED MemTest86+ v4.2.0

System Configuration:
Intel Xeon E3-1275 V2 (Brought in for testing - Core I3-3250 / Pentium G860)
Intel S1200BTLR (3 of these motherboard) - Latest BIOS
32GB (4 x 8GB) DDR-1600 ECC RAM (Crucial / Kingston - Intel Certified)

Same RAM PASSED all Test on a completely different machine.

Please advise.

[David (PassMark)]

So just to confirm my understanding, you see the same bad behaviour with the Intel S1200BTLR motherboard, regardless of the CPU used and the RAM used. Is that correct?

Would it be possible to try the test again with just 1 stick of 8GB?

We have had similar (but not identical) issues reported with the multi-threading option. But we have never been able to reproduce the problem. We were sure if the problem was due to the Xeon chip, the ECC RAM, the Qty of RAM or the MB.

The S1200BTLR is a relative low end (cheap) server board, and if you have narrowed the problem down to just the motherboard, then maybe we can purchase one of these boards for testing.

[BostonT]

Same behavior regardless of CPU and RAM.

Tried running the test with 1, 2 & 4 8GB modules. Tried swapping slots. Tried 4GB DDR3-1333 ECC Modules. All with the same result. Errors on Test 3.

[BostonT]

Very similar situation as xerces8 thread "multi CPU test fails / single CPU mode passes" with different hardware.

[David (PassMark)]

I agree, it is similar to xerces8's problem as well.

I have ordered an Intel S1200BTLR motherboard, Xeon E3 CPU and 16GB ECC RAM today. Should have it next Monday. We'll do some testing on it when it arrives. We needed a new web server machine in any case and also wanted to better support ECC RAM in MemTest86 in the future.

[bdon]

I agree, it is similar to xerces8's problem as well.

I have ordered an Intel S1200BTLR motherboard, Xeon E3 CPU and 16GB ECC RAM today.

This is an old thread. However, I've also encountered this problem on a SuperMicro X9SRH-7TF w/ single E5-2690 v2 processor, and 128G of Hynix HMT84GL7AMR4C-RDMCAD PC3-14900 LR DIMMS.

[David (PassMark)]

bdon, you need to give a better description of the problem.
Start by telling us what software version you are using, what the exact errors where, what settings you selected in MemTest86, if the problem is reproducible, and what you have already tried to narrow down the problem (e.g. testing with 1 stick of RAM at a time).

[bdon]

bdon, you need to give a better description of the problem.
Start by telling us what software version you are using, what the exact errors where, what settings you selected in MemTest86, if the problem is reproducible, and what you have already tried to narrow down the problem (e.g. testing with 1 stick of RAM at a time).

David, my apologies for the lack of clarity. I'm using Memtest86 version 5.01. The sequence that occurred was something like this.
a) booted memtest with no options
b) after running for nearly an hour, it caught my attention that memtest was reporting that SMP mode was disabled. So I did a bit of browsing and saw that F2 could be used to enable SMP mode.
c) rebooted, and as soon as memtest started, I hit 'F2'. Immediately, I was able to see that the program identified 20 cores (actually 10 cores, 2 threads-per-core). Things ran ok for the first 30 minutes, and then there was a burst of 48 errors. From memory, I think all of the errors were from a single processor, and the expect vs actual showed that almost no bits were correct in any of the errors. Finding this thread, I thought that the SMP mode was broken for my processor and
d) I rebooted and ran again using default settings. Ran for 26+ hours and 2 complete passes with no errors. (strangely, the first pass completed in about 9 hours, whereas the second pass took longer). CPU temps were constant at between 29 and 33 degrees.See the attached screen shot showing the completed passes.
e) today, I thought I'd try SMP mode again, but this time, I'm not seeing any errors - 4+ hours in. However, there's still some strangeness. It looked like the program got to the end of pass 1, where both the pass and test were at 100% for a long time (i.e. more than an hour) Now tests are progressing again, even though the pass progress is locked at 100%. The second screen shot shows the current state.

[David (PassMark)]

bdon,
The screen shots you have posted are from MemTest86+. Which is not our software, even though they copied the name our of product, which makes it confusing. I suggest you post to the MemTest86+ forum if you want help with their product.

Our product is the original and the best, MemTest86 (without the plus), which is currently at release V5.1