Closed LPX55 closed 5 years ago
Is it possible the binary kernels have power limit setting over riding your settings?
Add this to bypass binary kernels
--cl-only Use opencl kernel. Don't attempt to load binary kernel
Is it possible the binary kernels have power limit setting over riding your settings?
Add this to bypass binary kernels
--cl-only Use opencl kernel. Don't attempt to load binary kernel
Thanks for the swift response. Unfortunately, still the same results. Throttled at 48~ Watts.
Looks like I may have to try re-flashing my VBIOS for each card, unless you have any other suggestions :/
How are you over clocking them? If using Wattman I have seen a GPU from a group gets set to a weird stock 0 setting. Check there and reapply any settings to see if it gets up to regular speed.
I have had to restart my miner, reset wattman to sometimes get correct power settings not be throttled(or lower that I expected).
use this to specify compatible gpus:
--opencl-devices omit the smaller gpu with too little memory avoid it being a potential source of problems here.
How are you over clocking them? If using Wattman I have seen a GPU from a group gets set to a weird stock 0 setting. Check there and reapply any settings to see if it gets up to regular speed.
I have had to restart my miner, reset wattman to sometimes get correct power settings not be throttled(or lower that I expected).
I used MSI Afterburner, and I think that may be the root cause of the problem. I underclocked with Afterburner, then set "Apply settings on reboot" or something, and now that I think about it the problem was happening after that. I reset the settings from Afterburner many reboots ago, but for some reason it's still causing problems? I tested this out by replacing the 2GB RX460 with a 4GB RX480 and the swapped out card is the only card that is close ~90W.
Re-flashed all my cards. Still having the same problem... power draw throttled at ~48W. I guess next step is to DDU and then reinstall the drivers.
edit: One thing I noticed in Wattman was that the "max" core speed was set to 300MHz. I tried changing it with TRIXX and Afterburner, and of course Wattman, but it doesn't budge.
possibly custom bios in the cards?
possibly custom bios in the cards?
They were all stock BIOS with PBE one-click patches. Only thing I changed was the max temp and the target temp (70 deg).
Fixed by downgrading drivers to 18.5.2. But hashrates are horrid, ~15 Mh/s for a RX580 8GB. Sigh, gonna be a long night.
I just noticed this
i 10:28:28 main Found suitable OpenCL device [Baffin] with 2.000 GB of GPU memory
2 GB memory are not enough for Ethereum mining. You probably should remove that GPU from the used devices.
As I can see he excluded it from mining(seeing 6 of 7 all loading DAG). The throttling can come as a reason of wrong OC or no OC at all, besides full cmd line is preferable without user/wallet of course.
No further feed backs. Closing
I've done everything in my limited knowledge to try and figure out what's going on. I have a 1300W Platinum rated PSU, so I know it's not my PSU, and my power meter is reading a power draw of 400W. I made sure that it wasn't my undervolting (in fact, I've overvolted/overclocked to make sure) that was limiting my GPU's wattage usage (around ~50W per card). Did anyone have any similar experiences with issues like this in the past?
Notice how all cards are at 48W +/- 1, and 42W +/- 1... Which is bizarre, since they are all underclocked with different values.