5700XT Teamred miner GPU Detected Dead Fix

[coatercup]

For many users including myself every 1 or 2 days there would be a GPU Detected Dead for my 5700XTs. No Stales or Rejects, temps are fine. HiveOS reboots and everything resumes as per normal. Although it does not really impact the daily performance but I could not find a solution until now. Switched the config from B mode to A mode (add “–eth_config=A” in Miner Config → Extra config ) and the annoying GPU Detected Dead is gone. You would need to increase the core clock to ~1400 to match the hashrates though. Although no increase in power from wall for my case.

[tgdolemite]

I must be missing something… I have this problem across 80% of my cards. I got into the flight sheet and added “-eth_config=A” then went and changed my core clocks from 1250 to 1400, ran the flight sheet and none of my cards were hashing. Waiting 2-3 minutes… got rid of the config line, changed my clocks to 1250 and it started right back up again. Definitely interested in making this “GPU detected dead” thing go away… did I miss a step?

[bs.victor]

the exact config that worked with me is:

–eth_config=A (two “-”) this is important to say, i think this forum is changing the characters

i just changed here and is working fine, just need to see if will show more “gpu dead” messages

EDIT: It showed, so didn’t work for me

[Kanuki]

Thanks for the fix. I’ve been scratching my head trying to
fix the problem for a few day now.

Problem became more serious after I added 8 more cards
(total of 13 cards) into the rig. It’s been quite stable when
I was using only 5 cards. (I tried to add 11 more cards (total of 16 cards)
but my motherboard won’t POST. Max only 13 cards.)

GPU3 was always detected dead every 1-2 hour(s).
GPU2 was the 2nd frequent dead.
Only sometime GPU7.

I’m testing it now. Looking good so far.

Noticed that I’m going 2-3MH/s lower per card
and vRAM is hotter compare to B mode.

a few of my cards fan are running at 80-85%
whereby B mode is around 70-80%.

It’s still better than using nanominer because I’m getting
high rejection rate in one of the GPU (GPU5. but GPU5 was
never detected dead with B mode).

[Kanuki]

You need to put “- -eth_config=A” as first thing in “Extra config arguments:”.

If you put “- -eth_config=A” as the last thing in “Extra config arguments:”
then the miner will ignore the script.

You can verify by remote in your rig check the miner during the start.

If you’re seeing something like “tuning using Axxx” then you’ve done it correctly.

But if you’re seeing somehting like “tuning using Bxxx” which mean you did something wrong somewhere.