RX 6800 XT randomly dropping hashrate

[umansia]

Hi, I have two RX 6800 XT Sapphire that i normally get 63 MH/s, but after a couple of days their hashrate randomly drops (They dont drop to a certain number sometimes 30 MH/s sometimes MH/s 53 etc). And they get their normal hashrate after 3-4 reboot. Most of the time their startup hashrate would be low too. So i have to reboot multiple times. When i catch 63 MH/s boot, they are farming for days and then they drop their hashrate again…

I am using gminer and their settings are here; (For example in this screenshot GPU 2 started with 56 MH/s)

GPU’s with this problem are GPU 2 and GPU 3. Does anyone know how to approach to this proble?

[tom.cnd]

Having exact same issue, replaced riser, pcie slot, same. Rebooting ten times until get correct numbers. I tried a few times to change overclocking while was low hashrate and every time I change core or memory number, watts dropping on that card, it doesn’t matter going higher or lower, like would be not enough power for that gpu on boot, randomly, sometimes gets enough, using evga psu, good cables, weird

[themisters]

same thing here, I have an asus TUF 6800xt, one day it just randomly dropped, a reboot fixed it, after a couple more weeks, now I can’t get it to hold a steady hashrate, it’s bouncing between 55-62. Before I was able to get a solid 63-64 out of it for days on end. This is gonna drive me nuts, so when I come up with a solution I’ll update.

[tom.cnd]

I think I found solution for this random issue, it cause after hiveos or just miner update, use older version of miner, for example, for me stable team black miner 1.27 for 6800cards or 1.23 for rx6600cards

[themisters]

I down rev’d TRM to 8.3 and 8.4 still no dice. This is seriously annoying

[tom.cnd]

Try another miner, like teamblackminer

[themisters]

tried virtually everything, after troubleshooting, I’m starting to feel it’s a powertable setting issue. After using amd-oc reset amd-co then reboot I’m now back to being able to get 58-62 fluctuating at 145 watts. I was at 130 watts and a solid 63-64 before. I’ve gone through the amd-oc.sh code and amd-oc.navi.sh code and there have been some changes over the last few OS versions, but I can’t see anything that specifically would be doing this yet.

[x_Checkpoint_x]

i have the same issue with a Asus TUF RX 6800 XT GAMING OC , getting 61MH/s then after few days drops to 58~55 MH/s. already tried a lot of OC settings no luck so far…

[orionpax]

I have the same problem as well with several different rigs. Some mixed and some just AMD cards.
Also: when i change the OC settings of one card, several cards are affected and i need to reboot several times to get the rig on normal hashrate again.

Seriously annoying

[x_Checkpoint_x]

@orionpax

so: when i change the OC settings of one card, several cards are affected and i need to reboot several times to get the rig on normal hashrate again.

yes same here… only AMD cards and every time i need to change something or do something i need to restart multiple times my rig…

[bloodbathnbeyond]

If you haven’t found a solution to this problem… I got you.

The 6000 series AMD cards have a setting in the feature controls that is disabled by default: TEMP_DEPENDENT_VMIN

So, with that disabled your voltage is will always drop to min spec (around 881mv) when your GFX hits ~50’C or the SoC hits ~60’C. (These temps are rough estimates and are dependent on make/model/brand).

When the temps come down, the mV creeps up, hash goes up, etc etc.

Now, if you’re mV are pegged at <881 for efficiency, it won’t matter. Again, once the temp creeps up, the card neuters itself.

Only way around it is to enable TEMP_DEPENDENT_VMIN and change the Vmin values from 881mv (high and low … yes they are the same) to something higher. 1050mv or 1075mv.

Good luck.