Asus rtx3080 tuf gaming oc - settings & vbios?

[H2miner]

Hello Fellow miner,

I have 2 ASUS RTX3080 TUF GAMING OC that seem to be LHR, but i get very different performance from both of them :

I have tried multiple settings and overclocks, but nothing changes, GPU 0 is still 20 MH below GPU 1, and GPU 1 is 20MH below what other rtx3080 owners get. (I am on NBminer)

When I look in the shel it seems that GPU0 locks frequently, and its temp is 20°C over GPU1 which doesn’t lock.

Any ideas to how I could increase the hashrates on both cards, and stabilize GPU0 ?

Any firmwares I might try ?

Thanks,
H

[mathelm]

I don’t have a 3080 (have 3080ti’s, 3090, etc…) but off hand I would say up the fan on the first one to 100%. If you see a increase in hashrate with a decrease in temp, then it’s memory throttling (because memory is too hot). Try lowering the mem clock to start, and maybe look into replacing the thermal pads on the backplate.

[keaton_hiveon]

Start with changing to locked core clocks, 1200 to 1500 is a good place for 3080 lhrs. Increase fan to max and if unstable lower the memory oc. Trex is best for lhr at the moment, should get you 75-76mh

[H2miner]

Thanks @mathelm, already tried putting the fan to 100% but it didn’t change anything.

I would rather not touch the thermal pads for now as the card is under guarantee, this would be my last resort

Thanks for your reply !

[H2miner]

Thanks @keaton_hiveon I will try these and get back to you

Do you have any bios I could try ?
Can a gigabyte bios be flashed to an asus card as long as it is a 3080 ?

Thanks for your reply,

[keaton_hiveon]

bios isnt your issue. how does it run with better clocks and 100% fan?

[H2miner]

Yep tried many memory settings with trex miner, with fan @100%, it starts around 70MH and drops to 30 MH

Here are some screenshots of different overclock settings. None are stable and keep changing the MH rate.

Best results (and stable) I get are with NBminer 40.1, see below :

Any other ideas to what can be tested to improve this ?

Also I am on EtHASH ? maybe I should try a different algorithm ? @keaton_hiveon

Thanks,
HB

[keaton_hiveon]

make sure to restart the miner after any clock changes. is your card thermal throttling? you can check by running nvidia-smi -q | egrep "Thermal Slowdown"

[H2miner]

thanks @keaton_hiveon
Yes I do restart the miner everytime, and no the card isn’t thermal throttling.
Any other troubleshooting ideas ?

[keaton_hiveon]

What lhr % is the miner stepping down to? Should show inside the miner after running

[H2miner]

Here are screenshots of the miner screen:

Here is a screenshot of the flightsheet (mode 1 makes the card unstable, at least mode 2 is stableish), I put “lhr”=80 but when it starts it start at 71 for GPU 0 and 74 for GPU 0 :

@keaton_hiveon, any other insight ?
Thanks

[keaton_hiveon]

try trex 24.8 with no extra config. 63c on core is pretty warm, i would add some fans to lower that too

[psc69352]

How is your gpu pcie line setup, the power value is so strange, you can take out the normal one, to check again.

[DediZones]

Try Trex:

CLOCK="1185"
MEM="2000"
FAN="0" (autofan min 65% at 45° max)
PLIMIT="250"

TREX ARGUMENT:

"worker": "%WORKER_NAME%"
"gpu-init-mode": 1
"dag-build-mode": 2
"validate-shares": false
"kernel": 4
"lhr-autotune-step-size": 0.1
"lhr-autotune-interval": 30
"lhr-autotune-mode": "full"