Interesting that it needs extra juice to be stable, the frequencies stay the same as before.
I think that I will switch to Monero soon, Monero doesn't need blockchain drivers and all my problems will just go away.
Topic: ASUS DUAL RX580 O4G OC - GPU frequency drop (Read 985 times)
I think that writing high clocks in all states in BIOS will brick your card. I haven't tried it but someone in this thread mentioned it.
Thanks.
I have modded and flashed bioses with high GPU clocks in all states, the initial problem is solved.Thanks.
But some cards need 10-30mV higher GPU voltage for stable work in this case, so power consumption is a bit higher.
99.9% of gpu frequency, hashrate and any other drop has to do with power, 0.01% has to do with memory errors. So check your voltage correct and memory errors and more importantly your psu.
Thanks.I checked, my PSU is fine. These cards are really tricky to setup, they are behaving strange, everything I do through AB to reduce power consum. will result in instability. Don't really know since this can happen 2 hours from running miner or one week after.
Regards.
I've found cause of my MHz(hashrate) drop. I have water pump in the house that occasionally turns on to fill water in(house water system) when certain level of pressure is reached and it does it couple of times a day. I noticed that pump is using much power(draws pretty much electric current) in the moment of ignition which caused the instability of my PSU in a way not keeping correct voltage on GPUs. So i decided to turn it of for a couple of days and voila, got stable mining, no problem at all.
Thanks for advice to check power, you said that it is 99.9% power issue, and it was.
Best regards.
I have faced the same problem with new blockchain drivers,
GPUs are stuck in P0, P1 or P2 states (300-600-900 MHZ core clock) after reconnection to pool.
This problem does not happen with old drivers.
So maybe claymore can fix it.
Simple setting -cclock in claymore config doesn't help, maybe writing high clocks in all states in BIOS can help, I have not checked it yet.
GPUs are stuck in P0, P1 or P2 states (300-600-900 MHZ core clock) after reconnection to pool.
This problem does not happen with old drivers.
So maybe claymore can fix it.
Simple setting -cclock in claymore config doesn't help, maybe writing high clocks in all states in BIOS can help, I have not checked it yet.
I think that writing high clocks in all states in BIOS will brick your card. I haven't tried it but someone in this thread mentioned it.
Thanks.
I have faced the same problem with new blockchain drivers,
GPUs are stuck in P0, P1 or P2 states (300-600-900 MHZ core clock) after reconnection to pool.
This problem does not happen with old drivers.
So maybe claymore can fix it.
Simple setting -cclock in claymore config doesn't help, maybe writing high clocks in all states in BIOS can help, I have not checked it yet.
GPUs are stuck in P0, P1 or P2 states (300-600-900 MHZ core clock) after reconnection to pool.
This problem does not happen with old drivers.
So maybe claymore can fix it.
Simple setting -cclock in claymore config doesn't help, maybe writing high clocks in all states in BIOS can help, I have not checked it yet.
Core voltages are probably too low for your cards. Set them to higher like 950mV and you will see.
I'll try with higher voltages and see how they behave. Can you recommend optimal mem and gpu MHz to match with higher voltages?
Thanks a lot.
https://bitcointalksearch.org/topic/guidetutorialxmrethhow-to-mod-bios-rx470570480580-vega-mining-1954245
Core voltages are probably too low for your cards. Set them to higher like 950mV and you will see.
I'll try with higher voltages and see how they behave. Can you recommend optimal mem and gpu MHz to match with higher voltages?
Thanks a lot.
Core voltages are probably too low for your cards. Set them to higher like 950mV and you will see.
99.9% of gpu frequency, hashrate and any other drop has to do with power, 0.01% has to do with memory errors. So check your voltage correct and memory errors and more importantly your psu.
Thanks.I checked, my PSU is fine. These cards are really tricky to setup, they are behaving strange, everything I do through AB to reduce power consum. will result in instability. Don't really know since this can happen 2 hours from running miner or one week after.
Regards.
99.9% of gpu frequency, hashrate and any other drop has to do with power, 0.01% has to do with memory errors. So check your voltage correct and memory errors and more importantly your psu.
No, you cannot do that. You'll brick your card! Which drivers are you using? Try to decrease your core and memory under-volt and see if it makes any difference. Moreover, use the latest blockchain drivers and claymore 10.
Currently I'm using r0binhood's drivers from https://bitcointalksearch.org/topic/ethash-dag-epoch-rx470480570580-fix-2068446. My core is operating on 1162MHz, and mem volt is 1000mv, so can you please suggest me what values to try for them?
Tnx.
Well, Uninstall your previous drivers and install this one instead, Driver Link use claymore 10.0 with this driver and it should permanently fix your issue.
No, you cannot do that. You'll brick your card! Which drivers are you using? Try to decrease your core and memory under-volt and see if it makes any difference. Moreover, use the latest blockchain drivers and claymore 10.
Currently I'm using r0binhood's drivers from https://bitcointalksearch.org/topic/ethash-dag-epoch-rx470480570580-fix-2068446. My core is operating on 1162MHz, and mem volt is 1000mv, so can you please suggest me what values to try for them?
Tnx.
Use latest driver , it'll be fix that .
No, you cannot do that. You'll brick your card! Which drivers are you using? Try to decrease your core and memory under-volt and see if it makes any difference. Moreover, use the latest blockchain drivers and claymore 10.
Hi,
I've new rig testing with 2 Asus Dual RX580 O4G OC cards. By now, only modification was on bios itself. Copied mem straps from 1500 to 1750 and 2000, changed memory freq from 1750 to 2000MHz, and GPU freq after 900MHz to fixed 1162MHz(and same mV to 65282). Now, I'm getting between 29 - 29.8MHz with acceptable power consumption as you can see from image below:
https://pasteboard.co/GINcu5c.jpg
The problem is that in no any specific pattern both of the cards drop in hashrate to 15MH/s or 7MH/s, and when I check gpu MHz via Afterburner, they dropped to 300MHz. This can occur ie. 6 hours from running claymore miner or after 5 days, no rule.
Can I modify bios settings by putting gpu value 1162MHz to values where is 300MHz to 900MHz(fields shown in image below)? I'm afraid to do that and brick my card. Can you please advise me what would be the safest way to eliminate frequency drop?
https://pasteboard.co/GINfwNP.jpg
Best regards.
I've new rig testing with 2 Asus Dual RX580 O4G OC cards. By now, only modification was on bios itself. Copied mem straps from 1500 to 1750 and 2000, changed memory freq from 1750 to 2000MHz, and GPU freq after 900MHz to fixed 1162MHz(and same mV to 65282). Now, I'm getting between 29 - 29.8MHz with acceptable power consumption as you can see from image below:
https://pasteboard.co/GINcu5c.jpg
The problem is that in no any specific pattern both of the cards drop in hashrate to 15MH/s or 7MH/s, and when I check gpu MHz via Afterburner, they dropped to 300MHz. This can occur ie. 6 hours from running claymore miner or after 5 days, no rule.
Can I modify bios settings by putting gpu value 1162MHz to values where is 300MHz to 900MHz(fields shown in image below)? I'm afraid to do that and brick my card. Can you please advise me what would be the safest way to eliminate frequency drop?
https://pasteboard.co/GINfwNP.jpg
Best regards.
Jump to: