AMD driver package and AMD drivers builtin in Hive OS not the same things
Builtins AMD driver is mix of 3 things
So why do i then have better results with official newesr amd driver and how can i reach same results in hiveos…must be an optimization to the driver after latest firmware release of my 5700 card i assume…
I finally got it…updated to 20.20 and got 0.5 mhash per card at same watts.
Can you tell us how did you manage to update to 20.20 ?
guide to update amd drivers to 20.20 which is the latest driver supported by phoenixminer. you can update to 20.30 or 20.40 if you have another miner that supports a newer driver.
boot to maintenance mode with loading drivers
open remote access
cd …
cd hive-drivers-pack
wget http://download.hiveos.farm/drivers/amdgpu-pro-20.20-1089974-ubuntu-18.04.tar.xz
tar -Jxvf amdgpu-pro-20.20-1089974-ubuntu-18.04.tar.xz
cd amdgpu-pro-20.20-1089974-ubuntu-18.04
amdgpu-uninstall
y
y
after uninstall check dkms status:
dkms status
-> if you have Error, use command:
for i in /var/lib/dkms//[^k]/source; do [ -e “$i” ] || echo “$i”;done
copy the output line directory and remove those directory (or directories if more than one is listed in outut)
rm -r “directory”
reboot
check “dkms status”
should be without errors and without amd driver now
cd …
cd hive-drivers-pack
cd amdgpu-pro-20.20-1089974-ubuntu-18.04
./amdgpu-pro-install -y --opencl=pal --headless
reboot without maintenance mode - done
Great ! Thanks!
When I run “dkms status”, I get the error “Error! Could not locate dkms.conf file.”
And when I browse
/var/lib/dkms/amdgpu
I do see 5.6/ 5.6.5/ 5.6.5.24-1109583/ kernel-5.0.21-201105-hiveos-x86_64@ kernel-5.0.21-hiveos-x86_64@ kernel-5.4.67-hiveos-beta-x86_64@ kernel-5.6.0-hiveos-beta±x86_64@
And withing each directory, I have something like
5.6.0-hiveos-beta+/ source@
Any idea how I can move forward?
Thank you in advance,
Silat
it´s one post above. you need to delete the directories that are shown after typing the command. should be clear
Thank you. I saw that, but when I do “for i in /var/lib/dkms//[^k]/source; do [ -e “$i” ] || echo “$i”;done”
I get :
/var/lib/dkms//[^k]/source
And I am sorry, but I do not know what to do from here…
There’s a typo in the command it should be:
for i in /var/lib/dkms/*/[^k]*/source; do [ -e “$i” ] || echo “$i”;done
damn…didnt see that…copy pasted it into this forum…sorry for that
i am getting an error where it won’t let me remove the dkms directories i have installed. I have three and I assume I should leave the amd one alone. Please some help would be great
Good stuff! Thanks for taking the time to post this.
hello, thanks for your guide.
i tried to do all steps, had to change “amdgpu-uninstall” command to “amdgpu-pro-uninstall”
and install gets stucked on “Building initial module for 5.4.80-hiveos” and then error.
after restart it says A: 20.40 but can I trust this? my 6900xt still does not be recognized
any help?
just a quick self reply, I made a new boot hdd with fresh install and the problem continued, even if the new amd drivers 20.40 are already in the latest build.
then I made a clean BETA install, and its working…
so… users that have this problem, use the BETA version for now…
did you tried install 20.50??? i am stuck at “WARNING: amdgpu dkms failed for running kernel”
no, i did not. on Hive GPU Drivers Collection only 20.40
regarding dkms error, see the procedure CryptoMining talked about. I tried all the process but always got an error on kernel.
anyone here update to driver 20.50?
https://www.amd.com/en/support/kb/release-notes/rn-amdgpu-unified-linux-20-50
not sure which one is the correct for hiveos
there was a typo in my commands…use the corrected one which deliveres a result that you can delete.
if you got a result and deleted the packages correctly, then I dont how to fix that kernel error 
