98% pass on the beta software... is this a false positive?

[goathens]

I recently bought some new ram, memtest86+ ran all night without errors, but memtest86 failed 1 out of 56 of the tests. I'm attaching the report (great feature!).

A few things don't quite sit well with me:

1: The ram is pc-1600 instead of pc-1333 which is the motherboard spec. The bios reports that the ram is clocked down to 1333. Would this affect the test? Memtest still reports the ram at it's rated speed, but I suppose that may be getting read from the dimms not from the bios.

2: It's weird to me that one single subtest has so many failures and everything else succeeds. I'm not an expert in ram failures, but I would expect some failures on each test at the same address if there were something wrong with one of the chips, or errors all over the place on all tests if the ram were defective. This result makes me think that there could be a problem with the test. I am using the beta.

Your thoughts on the issue? Suggestions? I'm already going out to Microcenter to buy the same part and test that (my current parts purchased from amazon, and I'm hoping that picking another merchant will mean I get ram from a different batch).

Summary

Report Date	2016-06-07 06:19:43
Generated by	MemTest86 V7.0.0 Beta 1 (64-bit)
Result	FAIL

System Information

EFI Specifications	2.31
CPU Type	Intel Core i5-2400 @ 3.10GHz
CPU Clock	3100 MHz [Turbo: 3400.0 MHz]
# Logical Processors	4
L1 Cache	4 x 64K (89391 MB/s)
L2 Cache	4 x 256K (46326 MB/s)
L3 Cache	6144K (26290 MB/s)
Memory	16439M (16127 MB/s)
DIMM Slot #0	8GB DDR3 XMP PC3-12800
	Crucial Technology / BLS8G3D1609DS1S00. / E0AE2B53
	9-9-9-24 / 800 MHz / 1.500V
DIMM Slot #1	8GB DDR3 XMP PC3-12800
	Crucial Technology / BLS8G3D1609DS1S00. / E0AE2B49
	9-9-9-24 / 800 MHz / 1.500V

Result summary

Test Start Time	2016-06-06 23:39:02
Elapsed Time	5:34:28
Memory Range Tested	0x0 - 41F000000 (16880MB)
CPU Selection Mode	Parallel (All CPUs)
# Tests Passed	55/56 (98%)

Lowest Error Address	0x13D7E17C (317MB)
Highest Error Address	0x150FFFFC (336MB)
Bits in Error Mask	00000000FFFFFFFF
Bits in Error	32
Max Contiguous Errors	2

Test	# Tests Passed	Errors
Test 0 [Address test, walking ones, 1 CPU]	4/4 (100%)	0
Test 1 [Address test, own address, 1 CPU]	4/4 (100%)	0
Test 2 [Address test, own address]	4/4 (100%)	0
Test 3 [Moving inversions, ones & zeroes]	4/4 (100%)	0
Test 4 [Moving inversions, 8-bit pattern]	4/4 (100%)	0
Test 5 [Moving inversions, random pattern]	4/4 (100%)	0
Test 6 [Block move, 64-byte blocks]	4/4 (100%)	0
Test 7 [Moving inversions, 32-bit pattern]	4/4 (100%)	0
Test 8 [Random number sequence]	3/4 (75%)	5019598
Test 9 [Modulo 20, ones & zeros]	4/4 (100%)	0
Test 10 [Bit fade test, 2 patterns, 1 CPU]	4/4 (100%)	0
Test 11 [Random number sequence, 64-bit]	4/4 (100%)	0
Test 12 [Random number sequence, 128-bit]	4/4 (100%)	0
Test 13 [Hammer test]	4/4 (100%)	0

Last 10 Errors

[Data Error] Test: 8, CPU: 1, Address: 13DDA1C0, Expected: 0CCA7C9C, Actual: B1BC62AD

[Data Error] Test: 8, CPU: 1, Address: 13DDA1BC, Expected: F66B76DF, Actual: 2EA91B84

[Data Error] Test: 8, CPU: 1, Address: 13DDA1B8, Expected: C09E621E, Actual: 95F8A1E7

[Data Error] Test: 8, CPU: 1, Address: 13DDA1B4, Expected: A69A8D09, Actual: 9E5233C6

[Data Error] Test: 8, CPU: 1, Address: 13DDA1B0, Expected: 1CDCC810, Actual: AEE851D1

[Data Error] Test: 8, CPU: 1, Address: 13DDA1AC, Expected: 630BB863, Actual: E1316678

[Data Error] Test: 8, CPU: 1, Address: 13DDA1A8, Expected: B1AA66F2, Actual: B02E38EB

[Data Error] Test: 8, CPU: 1, Address: 13DDA1A4, Expected: 90C0BB6D, Actual: 0FD11C1A

[Data Error] Test: 8, CPU: 1, Address: 13DDA1A0, Expected: DE575344, Actual: CF3F89B5

[Data Error] Test: 8, CPU: 1, Address: 13DDA19C, Expected: FEBCE4A7, Actual: 2845992C

[keith]

1: The ram is pc-1600 instead of pc-1333 which is the motherboard spec. The bios reports that the ram is clocked down to 1333. Would this affect the test? Memtest still reports the ram at it's rated speed, but I suppose that may be getting read from the dimms not from the bios.

You are correct. MemTest86 reads the rated specifications from the RAM SPD, and not what is configured in the BIOS.

2: It's weird to me that one single subtest has so many failures and everything else succeeds. I'm not an expert in ram failures, but I would expect some failures on each test at the same address if there were something wrong with one of the chips, or errors all over the place on all tests if the ram were defective. This result makes me think that there could be a problem with the test. I am using the beta.

Yes, the results do look a bit weird. It doesn't appear to be a problem with the RAM.

Can you send or upload a copy of the MemTest86.log under the EFI\BOOT\ directory of the USB drive.

[goathens]

Yes, the results do look a bit weird. It doesn't appear to be a problem with the RAM.

Can you send or upload a copy of the MemTest86.log under the EFI\BOOT\ directory of the USB drive.

My log is posted here:

https://www.dropbox.com/s/ptrvuvqu68...est86.log?dl=0

I should also note that I ran this test a few nights ago with an 8 gig kit of pc1333 ram in the same pc (the pre-upgrade ram), and got no failures.

[keith]

Thanks for the logs.

It seems to happen at one specific point in time, for a particular memory range (317MB - 336MB) and CPU (1) where nothing in memory seems to match what was supposedly written. The fact that there were no errors outside of this specific occurrence makes it unlikely that there are problems with the RAM. It could have been a one-off corruption of the memory contents which could potentially be caused by a BIOS bug.

I would try re-running the tests again with just Test 8 enabled and see if the errors re-appear.

[goathens]

I've tried running only that test for a couple passes with no errors.

I did a similar test with another batch of RAM I purchased to replace this batch, and that batch failed test 8 and 12 in a higher memory range. This is different from the original run, which indicated failures at the lower end of the memory range. Running the tests individually yielded no failures. Strange stuff.

I've been working through the ram sticks individually (plugging only a single stick into slot 1 and running testing overnight). So far the tests on both replacement ram individually threw no errors, and I'm working on the original ram sticks starting now.

Is this a recognized behavior? Finding failures in ram as a set but not individual slots?

Maybe I should be using the stable 6.3.0 version instead?

[keith]

Test 8 and 12 both use random numbers so I'm suspecting it may have something to do with the random number generator. There haven't been any changes in v7.0 beta and we aren't aware of any issues but would like to get to the bottom of it. As I mentioned earlier, it is possible that a BIOS bug may be causing the error.

Do the errors appear in single CPU mode as well?

It would be good to get the logs for all runs that have errors.