[TECHNICAL PROBLEM] 32GB 3200 SODIMM only failed on pattern6?

[Minghung Chen]

Recently I used SK Hynix 16Gb 3200 A-die to made 32GB ECC SODIMM test by Memest86 V8.4,
(Original part number H5ANAG8NAJR-XNC)
I found only failed on pattern6, other pattern can test PASS, only pattern6 will fail

I tried SAMSUNG 16Gb 2666 M-die to made 32GB ECC SODIMM over clocking to 3200,
have the same chance fail only on pattern6,
do you have any same case?

Platform info : MSI MPG Z390 GAMING EDGE AC + i9-9900K

[David (PassMark)]

The screen shot you posted is tiny, so hard to see the details.

This doesn't look like ECC RAM as single bit errors should have been corrected.

Maybe you shouldn't be overlocking the RAM?

[Minghung Chen]

Because my platform not support ECC function, so memtest does not detect ECC, but it doesn't matter.
The problem is why 3200 have the chance fail only on pattern6, other pattern can test PASS.
(this DRAM IC uesd SK hynix Original IC, not ETT or UTT)
and i try modify frequency from 3200 to 2666, it will be PASS!

What function or program tested in pattern6?
have you heard a similar case?

Many thanks.

[David (PassMark)]

You can find a description of the tests here
https://www.memtest86.com/tech_indiv...est-descr.html

Test #6 does seem to be more effective at finding errors than some of the other tests. We don't know the deep technical reason for this.
But it must depend on the type of fault in the RAM chip. Some faults do pass test #6 and report errors on other tests however.