GTX 1060 MATS results?

[didi79]

MATS diagnosed faulty memory on banks A0 and A1 on a MSI GTX 1060 Armor 6G. I replaced them (from another 1060 donor).
No more errors on A0 and A1, but now MATS gives me write errors on all the other banks:

mats version 400.184. Testing GP106 with 20 MB of memory starting with 0 MB.

Read Error Count: 0
Write Error Count: 2327
Unknown Error Count: 0

=== MEMORY ERRORS BY SUBPARTITION ===
SUBPART READ ERRORS WRITE ERRORS UNKNOWN ERRS
------- ----------- ------------ ------------
FBIOA0 0 0 0
FBIOA1 0 0 0
FBIOB0 0 283 0
FBIOB1 0 829 0
FBIOC0 0 648 0
FBIOC1 0 567 0

Failing Bits:
B016 B017 B018 B019 B020 B021 B022 B023 B024 B025 B026 B027 B028 B029 B030 B031
B032 B033 B034 B035 B036 B037 B039 B040 B041 B042 B043 B044 B045 B046 B047 B048
B049 B050 B051 B052 B053 B054 B055 B056 B057 B058 B059 B060 B061 B062 B063 C000
C001 C002 C003 C004 C005 C006 C007 C008 C009 C010 C011 C012 C013 C014 C015 C016
C017 C018 C019 C020 C021 C022 C023 C024 C025 C026 C027 C028 C029 C030 C031 C032
C033 C034 C035 C036 C037 C039 C040 C041 C042 C043 C044 C045 C046 C047 C048 C049
C050 C051 C052 C053 C054 C055 C056 C058 C059 C060 C061 C062 C063
...



The error count is quite low in comparison with the count I had earlier on A0 and A1 (in the 700- and 800-thousands).


What could that mean?

[didi79]

Re: GTX 1060 MATS results?

While replacing the memory chips, I seem to have blown away a cap near A0. Found the cap by doing another visual inspection, it was stuck in some residue flux...

It's on of many FBVDD decloupling caps, so it really should not make a big difference. Last report still doesn't make any sense in that regard.
But I put the cap back anyway. Ran a new test and now report is different again

=== MEMORY ERRORS BY SUBPARTITION ===
SUBPART READ ERRORS WRITE ERRORS UNKNOWN ERRS
------- ----------- ------------ ------------
FBIOA0 0 227660 0
FBIOA1 0 0 0
FBIOB0 0 0 0
FBIOB1 0 251986 0
FBIOC0 0 0 0
FBIOC1 0 0 0

Failing Bits:
A000 A001 A002 A003 A004 A005 A006 A007 A008 A009 A010 A011 A012 A013 A014 A015
B048 B049 B050 B051 B052 B053 B054 B055 B056 B057 B058 B059 B060 B061 B062 B063


Guess next thing to try is reballing those chips. Maybe I have some stray solder balls that chose to camp there...

[benwaterson]

Re: GTX 1060 MATS results?

Maybe try reflowing them first as I had similar issues on 1080ti.

Heating the chips until the solder balls melt and the chip can be nudged and springs back with the surface tension.

Maybe check the FVBVDD resistance from the coils first to check its not low?

[didi79]

Re: GTX 1060 MATS results?

Resistance is ok. Card probably would not start if there was a short.

Thank you for suggestion to reflow first, takes less time than a reball. I will post back when I get to it.

[didi79]

Re: GTX 1060 MATS results?

So, I reflowed the ram chips. On the modules I had not touched earlier, solder balls escaped from under the modules when I nudged them. That's not good.
And obviously I now have shorts on ram. Clearly I will not power the board ;-)

So I will go ahead and reball all the chips anyway when I find more time (remove 6 chips, clean them and the pcb, reball the chips and put them back... with my novice skills and tools that may take a few hours...)

Still, this was not entirely useless I tend to think: I may have identified some dodgy factory soldering. Surface tension should keep the balls in place, I reckon.
May also explain the erratic test results: some flex on the pcb may cause solder joints to make or lose contact.