What to do in case of a sharp drop in the hash rate on a video card during mining?

What to do in case of a sharp drop in the hash rate on a video card during mining?

Sometimes, after starting mining, the user will experience an unpleasant surprise, which is manifested by a sharp drop in the hash rate, followed by the dumping of the video card.

At the same time, the available video memory is more than sufficient for the current DAG, the overclocking settings are minimized, but the miner produces a sad picture similar to the one below.

Screenshot of the nanominer miner when mining ETC with DAG equal to 2.591 Gb on a Radeon RX5500XT video card with 8 GB of memory (hash rate is only 1.5 mh / s, during normal operation it is about 28-29 mh / s):

What to do in case of a sharp drop in the hash rate on a video card during mining?

Such a problem can occur in any operating system, both Windows and Linux (including HiveOS) and is associated with resetting the frequencies of the video card to the base ones. Checking the current settings of the video card, which can be done in different ways, shows that the video card is operating at very low memory / video processor frequencies.

Screenshot of AMD Radeon RX5500XT video card, which refuses to work not only in overclocked, but also in basic mode:

What to do in case of a sharp drop in the hash rate on a video card during mining?

Displaying information about the state of the video card in the amdcovc program (you can see that the memory frequency has been reset to 500 MHz):

What to do in case of a sharp drop in the hash rate on a video card during mining?

Switching to an algorithm that heavily uses video memory has no effect, the problem remains open.

On the kawpow algorithm, the problem video card continues to "strike", sabotaging the calculations:

What to do in case of a sharp drop in the hash rate on a video card during mining?

It would seem that a (temporary) salvation could be switching to the mining of cryptocurrencies that do not require high memory performance, for example, on Equihash 192, but here, too, errors occur, up to the "departure" of the card.

Blade of the video card in the lolminer program with the error Stuck device detected, invoking emergency script:

What to do in case of a sharp drop in the hash rate on a video card during mining?

Problems that are expressed in the strange operation of a video card, as a rule, are associated with incorrect operation of drivers, or violations in the data transmission chain.

There are the following ways to solve the problem of resetting memory frequencies on a problem video card:

  • Check the virtual memory limit set on the computer. Problems can appear if the swap file is small on Windows, or if the vm.swappiness value is not set correctly on Linux.
  • Include a monitor / emulator in the AMD card.
  • Reset the overclocking and downvolting settings followed by a reboot. Before that, you need to disable saving the session. In the xfce environment, this is done with the command:
rm ~ / .cache / sessions / xfce4-session- *
  • Remove / disable overclocking programs and check if the card is back to normal operation.
  • Perform a complete removal and clean installation of the drivers.
  • Reinstall the video card into another PCI-E slot or riser (replace the USB data cable).
  • Update the system (Linux) with the command:
sudo apt update && sudo apt -y upgrade && sudo apt -y dist-upgrade && sudo apt -y autoremove

and reboot.

  • Transfer the video card to another rig / reinstall the operating system / switch to a different kernel.
  • Flash the correct BIOS of the video card.

Sometimes the simplest solution to problems with incorrect operation of video cards is a clean installation of the system. Based on practical experience, this helps after upgrading a running xubuntu 18.04 system to version 20.04. After such an update, the video cards began to behave abnormally, dropping frequencies, while installing the drivers, there were errors associated with the installation of the amdgpu-dkms package:

Error! Bad return status for module build on kernel: 5.4.0-56-generic (x86_64) Consult /var/lib/dkms/amdgpu/5.6.0.15-1098277/build/make.log for more information. dpkg: error processing package amdgpu-dkms (--configure): installed amdgpu-dkms package post-installation script subprocess returned error exit status 10 ... Errors occurred while processing the following packages: amdgpu-dkms E: Sub-process / usr / bin / dpkg returned an error code (1)
What to do in case of a sharp drop in the hash rate on a video card during mining?

Typical ways to fix this error

sudo dpkg --configure -a sudo dpkg -P error-packages apt install -f sudo apt --fix-broken install

did not help, the rig resumed normal operation only after a clean installation of the system.