It worked perfectly for weeks or even months, with "shitty" Afterburner and settings described above. Nothing has been changed on my settings since months ago.
Mine settings are not "default usage", regarding voltage and/or frequency.
Every card has been tuned rigurous (voltage and frequency), one by one and all at once.
I can go much higher as 1910MHz for GPU and 1120MHz for memory with around 1055mV GPU (still undervolted regarding stock 1126mV most of cards), achieving 97-99 MH for each card, but in the past I had 2 cards dying at 1 month interval, on simple manual system restart, they gave Code 43 error.
Those cards have been replaced at warranty. Since then I dropped all to 1831MHz GPU/1071MHz memory at 1000mV GPU, and everything was perfect for months. No other card died since then. I heard at friends of mine more Radeon VII cards died when exceeding 1890MHz GPU and 1100 MHz HBM 2, even they not exceed, ever, 70 celcius degrees.
So, back to issue.
This problem started couple days ago, I think 2 or 3 days ago.
Here is Nicenash Miner with Phoenix plugin, starting and running OK, for 39 minutes:
https://postimg.cc/PLSG6Pch
After 39 minutes it founds "new job" aand after allocating DAG and tunning -GT parameter, is dropping to 83,xxMH and never rise up to 93,xxMH, until miner fully restart.
https://postimg.cc/xJ8wB9Dm
Here is a part of the log when "staring new job" and allocating DAG, hashing is dropped to 83,xxMH for all cards and stays there, even after -GT has auto tunned:
https://postimg.cc/KKxkPHXw
Mine settings are not "default usage", regarding voltage and/or frequency.
Every card has been tuned rigurous (voltage and frequency), one by one and all at once.
I can go much higher as 1910MHz for GPU and 1120MHz for memory with around 1055mV GPU (still undervolted regarding stock 1126mV most of cards), achieving 97-99 MH for each card, but in the past I had 2 cards dying at 1 month interval, on simple manual system restart, they gave Code 43 error.
Those cards have been replaced at warranty. Since then I dropped all to 1831MHz GPU/1071MHz memory at 1000mV GPU, and everything was perfect for months. No other card died since then. I heard at friends of mine more Radeon VII cards died when exceeding 1890MHz GPU and 1100 MHz HBM 2, even they not exceed, ever, 70 celcius degrees.
So, back to issue.
This problem started couple days ago, I think 2 or 3 days ago.
Here is Nicenash Miner with Phoenix plugin, starting and running OK, for 39 minutes:
https://postimg.cc/PLSG6Pch
After 39 minutes it founds "new job" aand after allocating DAG and tunning -GT parameter, is dropping to 83,xxMH and never rise up to 93,xxMH, until miner fully restart.
https://postimg.cc/xJ8wB9Dm
Here is a part of the log when "staring new job" and allocating DAG, hashing is dropped to 83,xxMH for all cards and stays there, even after -GT has auto tunned:
https://postimg.cc/KKxkPHXw
your volts seem to be way too high. never touched Radeon VII, but amd 14 nm should be fine on 20-30% lower core frequencies (less needed for ethash anyways) at below 900.
all polarises originally set to 975-1000(at 1150-1180 core), but they do just fine at 850, vegas can do lower yet about 817 or so, 5700 can go below 800
Try 864-875mv and lower the core... I guess
