Memtest86 Hammer Test Errors

[RexFury117]

I have recently upgraded my system and after I got everything setup I ran memtest86 version 6.0 to check my new memory. I was fine till I got to the “Hammer Test,” I ended up with about 21 errors just from the hammer test. I decided to isolate each memory stick and see if I could figure out which one was bad. The following is a breakdown of all the tests I ran, also these are all 1 pass through the Hammer Test only:

1 memory stick of 8GB RAM with xmp settings @1.5v: No Errors
The same setting on the other memory stick: No Errors
Both memory sticks installed using xmp setting @1.5v: 103 Errors
Both sticks xmp settings @1.55v: 60 Errors
Both sticks xmp settings @1.65v: 70 Errors
Moved stick in A2 slot to B2 and stick in B2 slot to A2 with xmp settings @1.65v: 15 Errors

My concern now is the memory passes the hammer test when I use 1 stick, but when they are both in there I get errors. Could this possibly mean that there is a problem with my motherboards memory slot or is it still most likely bad RAM. I am not sure what else to try to figure out if it is a RAM issue or a motherboard issue.

My parts list is:
I7 4790K @ 4.0gHz
Asus Maximus VII Hero
Crucial Ballistix Elite 16GB DDR3-1866 RAM
GTX 980 2 way SLI
Ultra 1200watt Modular PSU

 

[Calculagator]

You have to check the documentation, but I believe that "hammer" errors don't exactly indicate that your memory is faulty. A hammer error happens when an entire row of memory is repeatedly written to with the max voltage and it causes a bit in a neighboring row to flip. It's not a good thing, but won't cause problems unless you run a malicious program.

The fact that you only observe it with both sticks installed is probably related to the way that affects voltages. Just a small change could be enough to allow the bit flips.

I could be wrong-I need to read up on it more.

 

[RexFury117]

You have to check the documentation, but I believe that "hammer" errors don't exactly indicate that your memory is faulty. A hammer error happens when an entire row of memory is repeatedly written to with the max voltage and it causes a bit in a neighboring row to flip. It's not a good thing, but won't cause problems unless you run a malicious program.

The fact that you only observe it with both sticks installed is probably related to the way that affects voltages. Just a small change could be enough to allow the bit flips.

I could be wrong-I need to read up on it more.

Thanks for the info. From the limited knowledge I have and what I read, I thought even single bit errors could cause relatively significant issue during everyday use. My main concern for right now is trying to rule out the motherboard as the culprit of this issue, if possible. I have played with the voltages a bunch and would prefer not to mess with the RAM timing. I would hope that it could run error free at it's native setting. When you say check the documentation do you mean for the RAM, for memtest86, or something else?

 

[Calculagator]

The row hammer test is different. RAM that fails the hammer test may still conform to manufactures specs. The hammer test is more about determining whether your RAM sticks are vulnerable to a theoretical exploit. RAM that fails may run just fine. Here are the relevant notes from the memtest86 technical info:

Test 13 [Hammer Test]

The row hammer test exposes a fundamental defect with RAM modules 2010 or later. This defect can lead to disturbance errors when repeatedly accessing addresses in the same memory bank but different rows in a short period of time. The repeated opening/closing of rows causes charge leakage in adjacent rows, potentially causing bits to flip. This test 'hammers' rows by alternatively reading two addresses in a repeated fashion, then verifying the contents of other addresses for disturbance errors. For more details on DRAM disturbance errors, see Flipping Bits in Memory Without Accessing Them: An Experimental Study of DRAM Disturbance Errors by Yoongu Kim et al.

 

[RexFury117]

The row hammer test is different. RAM that fails the hammer test may still conform to manufactures specs. The hammer test is more about determining whether your RAM sticks are vulnerable to a theoretical exploit. RAM that fails may run just fine. Here are the relevant notes from the memtest86 technical info:

Test 13 [Hammer Test]

The row hammer test exposes a fundamental defect with RAM modules 2010 or later. This defect can lead to disturbance errors when repeatedly accessing addresses in the same memory bank but different rows in a short period of time. The repeated opening/closing of rows causes charge leakage in adjacent rows, potentially causing bits to flip. This test 'hammers' rows by alternatively reading two addresses in a repeated fashion, then verifying the contents of other addresses for disturbance errors. For more details on DRAM disturbance errors, see Flipping Bits in Memory Without Accessing Them: An Experimental Study of DRAM Disturbance Errors by Yoongu Kim et al.

Yea I read that on their site and actually started reading through the paper by Yoongu Kim regarding disturbance errors. I just find it strange that I can run 1 stick without any issues, but not both. There is also this bit of info in their troubleshooting section:

http://www.memtest86.com/troubleshooting.htm

I plan on moving a single stick around to each memory slot on my motherboard just to help make sure that it is not the motherboard.

 

[Tradesman1]

You can get false errors on basically any test when running multiple sticks, alwasys best to run individually

 

[pigsun]

maybe mutiple sticks creats more load for the IO or CMD/ADR pins for motherboard, which may causes signal integraty issue, leading the err...

BTW, it the err always find at just harmmer pattern or still can find on other pattern when mutipu stick?