Hello there, i am having headaches trying to figure it out why MATS is not running properly.

So, first things first. I have 2x units of Zotac RTX 2080ti (ZT-T20810D-10P) with different problems, but maybe the same reason, idk for sure.

1 - The first one: When i install drivers i got space invaders all over the screen.

2 - The second one: The drivers are installed correctly, but when i push something on it i got different type of artifacts, triangular ones. If i lower the mem/gpu freq, sometimes it works for a few minutes at kombustor and games, but mostly it just close programs and games instantly.

*important, both cards install drivers and i do not get error 43.

After i saw this post from Assasin (viewtopic.php?f=48&t=103075) I've been trying using mats in a lot of different ways to track witch memory banks have issues, but the thing is, except from MATS 400.184 fully automated with -e 20 test, all other testing i got in the first line: RM Unavailable - Won't Test TULIT2 FrameBuffer upper memory.

- with -e 1~100 i got 0 errors.

- with -c 10 or -c 100 i got "Read Error's" in all memory chips.

Then, i saw this post from "anton127": viewtopic.php?f=48&t=106368&hilit=%2Arm+unavailable%2A

He tested MATS in a lot of different ways, and some of his testing he also got read errors in all banks, so i guess this is not an actual problem with those bank's, but with the testing method.

My mainly questions here are:

1- Why do i get RM Unavailable - Won't Test TULIT2 FrameBuffer upper memory. And how much this is important for the testing.
*(tested with 3 different motherboards already)

2- Mats only work when you get error 43?

3- Does anyone fixed "space invaders" artifacts on RTX 2xxx by switching those 3 memory banks near the PCIE slot ?

I just give up trying to do MATS show me those memory banks errors and now i am moving on to the old fashion way, but i need to know for sure if the problem is related to memory at this point.

Will appreciate any help!

"space invaders" artifacts on RTX 2xxx

if memory is micron (8X) A77 all 11 module memory must be replaced

Yes, all memory chips on both cards are micron 8RA77 - D9WCW.

Swapping all chips will be very expensive, are you sure about this? I really don't understand why all chips must be replaced, my guess was mechanical damage under those 3 near the PCIE slot.

all memory chips (micron (8X) A77 - D9WCW.)
factory defect

Update:

So, i tried reflux with simple flux in both cards on that 3 memory chips under the GPU.

1 - Card, with "space invaders" artifacts

Now i got different artifacts and error 43.

MATS show's me this:

=== MEMORY ERRORS BY SUBPARTITION ===
SUBPART READ ERRORS WRITE ERRORS UNKNOWN ERRS
------- ----------- ------------ ------------
FBIOA0 0 0 0
FBIOA1 0 718091 0
FBIOB0 0 0 0
FBIOB1 0 544187 0
FBIOC0 0 0 0
FBIOC1 0 0 0
FBIOD0 0 0 0
FBIOD1 0 0 0
FBIOE0 0 0 0
FBIOE1 0 0 0
FBIOF0 0 0 0
FBIOF1 0 0 0


2 - Card with spikes and closing programs when stressing GPU:

After reflux on that 3 memory banks the card goes on but i got no video.

Is it possible that i did other kind of damage on this second card? Or those chips are now just stopping the card to boot up?

I do not want to stress with any more heat this card anymore until the new memory chips arrive, but i don't know if i damage more or no at this point.

My 2070 doesn't have any artifacts, nor does it give any errors in MATS, but freezes whenever the driver tries to load.

When you say the defective memory is (8X) A77, do you mean xxA77, or 8xA77? The memory on this 2070 is all 9BA77 D9WCW.



So did replacing the memory chips fix the problems?