Locating faulty stick/sticks using MemTest V7.1 PRO

[FrugalHorn]

Hi!

Apologies for a long post ahead.

Since I have assembled my machine, I was facing memory management BSODs on Windows-10 64 bit. I experienced about 3 per day when the machine was used for 4-5 hours. After all the usual troubleshooting, I concluded that this is a memory issue and started using MemTest86 7.1 free version.

The list of system components:
Motherboard: ASUS X99 Deluxe
Processor: Intel I7 5820K
RAM: 2 of Crucial Ballistix Sport 32GB Kit 2400Mhz(8GB x 4). Total 64 GB
Storage:
- Samsung 850 Evo 500GB
- WD Black 4TB
Graphics card: Gigabyte GeForce GTX 980Ti
Processor Cooling: NZXT Kraken X61 All-in-one
Power supply unit: SEASONIC 860W POWER SUPPLY 80+ PLATINUM
OS: MS Windows 10

Misc. info about the current state of the machine:
1. I don't overclock. Only basic BIOS parameters are touched like boot sequence, enabling/disabling LAN controller, wifi, bluetooth etc.
2. I have up to date BIOS/UEFI ie V 3301
3. As far as possible, I did investigate and updated drivers for all the devices.
4. Initially, I used the free version to confirm that it's a memory problem. Later, for efficient, automated and quick troubleshooting I purchased the Pro version and it's helping me a lot.
5. Temperatures are good. Some tests were deliberately run using sequential/round robin to corner the possibility of a bad CPU.
6. Well known here that ASUS X99 deluxe can't run parallel tests and I am executing them with 1 CPU.

What did I do during troubleshooting?
1. I went about reducing the sticks and isolated the stick that was problematic. This stick gave errors even when used alone or in combination with another stick.
2. Now problem reduced to 7 sticks and feeling lucky, I straight plugged in 7 sticks, leaving a hole in one slot. Fired up all the 13 tests for 8 passes. Landed in trouble once again with errors during 4th test in 1st pass. Log file for this test run is at: http://www1.zippyshare.com/v/TTzy7WJI/file.html
3. Aborted the idea of working with 7 sticks and went back to 2 sticks at a time.

Where, am I currently?
1. I've managed to put together group of 4 sticks that worked well for 8 passes, in the slots mentioned by ASUS(A1, B1, C1, D1)
2. I'm working on remaining 3 sticks in isolation, to locate the 2nd villain.

What are my questions:
1. Is it a good idea to put seven sticks on a X99 motherboard for testing? Trying to confirm if it's capable of giving false positives.
2. Now that I think I've isolated 4 good sticks. How should I proceed with remaining three? Combination of "4-good + 1-to-be-tested" OR "3-good + 1-to-be-tested"?
3. Just to get through a combination quickly, can I use only the tests that have failed previously.
4. Just as a smoke test what combination of tests will work well if my last failure was Test 4(Moving inversions, 8 bit pattern)?

Thanks for a very useful and easy-to-use tool. Any help, tips, pointers is deeply appreciated.

Regards!
FH

[keith]

1. Is it a good idea to put seven sticks on a X99 motherboard for testing? Trying to confirm if it's capable of giving false positives.

Ideally, you want to reduce as many variables as possible by isolating and testing each module separately. If not, at least ensure each stick is running in single-channel mode.

2. Now that I think I've isolated 4 good sticks. How should I proceed with remaining three? Combination of "4-good + 1-to-be-tested" OR "3-good + 1-to-be-tested"?

See above. Is there a reason for not testing the "1-to-be-tested" separately?

3. Just to get through a combination quickly, can I use only the tests that have failed previously.

It is a good idea to go through at least one pass of all tests to catch the most obvious errors. The tests are generally ordered from quickest to longest. You can only save time if the previously failed tests identify errors; no errors does not mean the sticks are not faulty.

4. Just as a smoke test what combination of tests will work well if my last failure was Test 4(Moving inversions, 8 bit pattern)?

See above.

[David (PassMark)]

To have 1 bad stick is just unlucky. To have two bad sticks makes me more suspicious of a design flaw or motherboard / RAM compatibility issue. You didn't mention the speed of the RAM, but maybe running it slightly slower might bring some stability.

These lines from the log file are indicative of a real RAM error. In each case there is a 1 bit error.

[MEM ERROR - Data] Test: 4,Address: 3502A4D8, Expected: 10101010, Actual: 11101010
[MEM ERROR - Data] Test: 4,Address: 4003B770C, Expected: 02020202, Actual: 02020242
[MEM ERROR - Data] Test: 4,Address: 48272A5D8, Expected: 10101010, Actual: 11101010
[MEM ERROR - Data] Test: 4,Address: 4D2457E5C, Expected: 04040404, Actual: 04040444
[MEM ERROR - Data] Test: 4,Address: 57FB2A498, Expected: 10101010, Actual: 11101010

It seems there is more than 1 fault as well. As different addresses and bits are corrupted.

Also what is sometimes reported is that using a lot of high speed RAM (8 sticks in your case) causes additional power draw and a corresponding small voltage drop. DDR4 is normally 1.2V. So maybe try 1.25V

[FrugalHorn]

Thanks Keith. Thanks David.

The RAM speed is 2400 Mhz. (will update the original post too)

I will continue troubleshooting and will be back in a day or two to post the status.