Hello everyone! Im right now dealing with 3 graphics cards. All same model and brand (GALAX RTX 2060 6GB). When i do run Furmark and Mining software the GPU goes black or shows artifacts. After hours struggling with them i came to the conclusion the issue is located inside the GPU. The mods output showed B0 and B1 channel with error(sometimes B0 sometimes B1). I replaced RAM chips with no success. Also removed the GPU chip and reballed it with no success. The only thing i can think about is disabling the B ram bank and solder the RAM chips to the A bank (A bank is factory empty). Unfortunately im not good with bios editing. Ram chips are SK Hynix H56C8H24AIR This seems to be a common issue regarding GALAX 2060 6GB. Thank you!

if hynix has "s2c" marking - with high possibility that chip's have a factory defect.

described symptoms is have coincidence with this type of factory defect.

try to change all memory chips to samsung.

Thanks for the reply. What do you mean by "s2c" marking? I have Samsung k4z80325bc-hc14 to use as replacement. After replacing them, is there anything else i need to do? I was thinking about flashing bios with Samsung compatibiliy. Also i saw some users here move or modify a resistor to make them work. I don't know if this is the case. Thanks

EDIT: Nevermind, just found all my Hynix ram chips are S2C

BIOS already have samsung support. only straps resistors replacing is needed

Have these GPUs been undervolted?

Your GPUs are dead from mining while undervolted. To prove this, lower the GPU clock speed by 300-500 MHz and test again. Once you lower GPU clocks enough to run stable, try overclocking the VRAM by 10-15%. A slight memory overclock shouldn't change anything if it's a GPU problem.

Samsung memory won't help you. I've got a 2060 with Samsung memory sitting right here with the exact same symptoms. Crashes, black screen, and occasional single errors on B channel in MATS. When reducing GPU frequency by 300 MHz, this card will run stable even with a 15% memory overclock. Card had been mined with at 0.85V core voltage. Raising the voltage back up to normal 1.05V does nothing to fix the damage.

Trying to achieve Bank A with BIOS is ridiculous because Bank A is fused off in the GPU itself (and that's what "KA" in the GPU part number means - bank A is permanently killed). BIOS firmware has nothing to do with it.

Hello Joe24. No, these GPU's were not undervolted. They are all new, and come from a customer who has a contract with Galax to sell them new in his business. Three of these new 2060 failed in the same location, B ram bank. After some time swapping ram chips and doing reballing to gpu the problem persist. I took notice of what you said about KA marking and eventually the gpu has A channel fused off. The only solution is to replace the GPU chip or use the frequency stuck at 300mhz.

Thanks intell2000 and Joe24 for your help. I learned a lot about this. Thank you!

Original vBios extracted
https://drive.google.com/file/d/1yBm_V4 ... sp=sharing