Welcome to MilkyWay@home

Problem switching from Windows to Ubuntu: ROCm question

Questions and Answers : Unix/Linux : Problem switching from Windows to Ubuntu: ROCm question
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Joseph Stateson
Avatar

Send message
Joined: 18 Nov 08
Posts: 291
Credit: 2,461,693,501
RAC: 0
Message 75163 - Posted: 18 Mar 2023, 18:35:46 UTC
Last modified: 18 Mar 2023, 18:41:31 UTC

System with AMD VII and S91x0 boards that work fine in Windows 10. Want to switch to Ubuntu.
I installed 22.04 and AMD drivers using "=rocr,legacy". I disabled CSM in bios and the motherboard bios is UEFI
Only one slot is Gen3. The AMD VII board is in that one. The S91x0 boards are all in Gen2 slots. Gen 2 does not support atomics I read somewhere long ago.

Only the AMD VII board works
clinfo reports the other gfx9 boards (s9150, s9100) have no platform.
sudo lshw -c video mis-identifies the s9150 and s9100 as w9100 and w8100
However, I have seen windows do the same and they work anyway.

According to some ROCm doc I cannot find at the moment quote:

GFX9 GPUs (such as Vega 10) no longer require PCIe atomics


That Gen3 slot has atomics since it is Gen3. The Gen2 do not but supposedly ROCm does not need it for gfx9 boards. I specified legacy as the s91x0 boards are legacy but I might be mistaken.

Is anyone running ROCm in Gen 2 slots?
ID: 75163 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Joseph Stateson
Avatar

Send message
Joined: 18 Nov 08
Posts: 291
Credit: 2,461,693,501
RAC: 0
Message 75169 - Posted: 20 Mar 2023, 16:01:43 UTC - in response to Message 75163.  

I have "almost" got this to work. I forced Gen3 on the x16 slot and now both the VII and the Mi25 boards work in that slot.
Unfortunately, the VII will not work in any of the x1 slots and none of my x9150 boards work. The drivers for the mi25 and VII seem to be missing what is needed for those older S boards.
ID: 75169 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Keith Myers
Avatar

Send message
Joined: 24 Jan 11
Posts: 708
Credit: 543,284,486
RAC: 140,345
Message 75171 - Posted: 20 Mar 2023, 18:51:45 UTC - in response to Message 75169.  

I haven't heard of anybody being able to use those mining motherboards efficiently for BOINC crunching. The 1X slots are just too restrictive.
ID: 75171 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Joseph Stateson
Avatar

Send message
Joined: 18 Nov 08
Posts: 291
Credit: 2,461,693,501
RAC: 0
Message 75175 - Posted: 21 Mar 2023, 4:22:00 UTC

Think I found the problem. The Mi25 is recognized in 22.04 with ROCm 5.x but is not supported
AMD Instinct MI25 End of Life
ROCm release v4.5 is the final release to support AMD Instinct MI25


When I tried amdgpu-install --rocmrelease=4.5.2
The error "the package was not found"
The 4.5.2 was last used in 20.04. I cannot find a repo that has the 4.5.2 rocm release.
Messing around with and updated repo from AMD I ended up with broken packages I cannot fix.
Will try 20.04 and see if it can find a 4.5 rocm version.
ID: 75175 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 18 Nov 22
Posts: 81
Credit: 637,705,956
RAC: 42,679
Message 75177 - Posted: 21 Mar 2023, 15:36:48 UTC - in response to Message 75175.  

isn't this exactly what I said in the other thread?

Vega only supported by ROCm 4.5
ROCm 4.5 only supported on 20.04 with 5.11 kernel or 18.04 with 5.4 kernel.

i wasnt so specific, but I said you might need an older OS and to check the install docs, and the above is what the install docs say.

ID: 75177 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Joseph Stateson
Avatar

Send message
Joined: 18 Nov 08
Posts: 291
Credit: 2,461,693,501
RAC: 0
Message 75178 - Posted: 21 Mar 2023, 15:55:48 UTC - in response to Message 75177.  
Last modified: 21 Mar 2023, 16:23:22 UTC

isn't this exactly what I said in the other thread?

Vega only supported by ROCm 4.5
ROCm 4.5 only supported on 20.04 with 5.11 kernel or 18.04 with 5.4 kernel.

i wasnt so specific, but I said you might need an older OS and to check the install docs, and the above is what the install docs say.


I did not see that (rather did not think about it). got too involved getting the Mi25 to work

It turned out that windows was easier to work with. Going to post in the GPUUG over at Einstein
ID: 75178 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 18 Nov 22
Posts: 81
Credit: 637,705,956
RAC: 42,679
Message 75179 - Posted: 21 Mar 2023, 16:34:16 UTC - in response to Message 75178.  

the AMD linux driver situation is certainly all over the place. thankfully nvidia's driver install is pretty painless on both windows and Linux. sucks that AMD's linux experience isn't similar.

ID: 75179 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3319
Credit: 520,332,827
RAC: 21,341
Message 75180 - Posted: 21 Mar 2023, 21:57:15 UTC - in response to Message 75179.  

the AMD linux driver situation is certainly all over the place. thankfully nvidia's driver install is pretty painless on both windows and Linux. sucks that AMD's linux experience isn't similar.


YES IT IS!!!
ID: 75180 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
.clair.

Send message
Joined: 3 Mar 13
Posts: 84
Credit: 779,527,671
RAC: 3,042
Message 75181 - Posted: 22 Mar 2023, 2:43:37 UTC

I have recently played the match the drivers kernel to the Linux Mint version so I could get my `old` amd7970 running
it worked , eventualy ,finding out then remembering that "Legacy" commandline switch had to be used so that a working open cl would be installed
What a total pain in the #&%$" it was
Two installs of mint , nuked the first 20.2
went back to mint 19.0 as it had the 4.15.0 kernel the driver needed
Then its make shure the card don't cook while crunching coz the fans are far too slow
the back of that card was instant burnt fingers hot
Read a heap more on the web , try it , it don't work , BUT find , dirty hack the file `pwm1` with 255 and its full fan speed for cool crunching .
Though I have an almost identical system that runs win7 ultimate and amd7970s all that OS / drivers / fan speed control stuff is so easy ..
And yet another system that has Linux Mint but with a nv 1060 in it , install driver / install "coolbits" = get work done .
AMD shure duz need to get its act together
ID: 75181 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Keith Myers
Avatar

Send message
Joined: 24 Jan 11
Posts: 708
Credit: 543,284,486
RAC: 140,345
Message 75182 - Posted: 22 Mar 2023, 6:34:25 UTC - in response to Message 75181.  

I know that AMD cards are a pain to work with, but are aware of this set of gpu utilities focused primarily on AMD cards? Would give you fan and clock control for one thing and some great displays of card info and graphs.

https://github.com/Ricks-Lab/gpu-utils

I've helped the developer out extensively on its ability to work on Nvidia cards also.
ID: 75182 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
.clair.

Send message
Joined: 3 Mar 13
Posts: 84
Credit: 779,527,671
RAC: 3,042
Message 75190 - Posted: 22 Mar 2023, 23:54:22 UTC - in response to Message 75182.  

I know that AMD cards are a pain to work with, but are aware of this set of gpu utilities focused primarily on AMD cards? Would give you fan and clock control for one thing and some great displays of card info and graphs.
https://github.com/Ricks-Lab/gpu-utils
I've helped the developer out extensively on its ability to work on Nvidia cards also.

I have had a read of that page and will give it a go , many Thanks .
ID: 75190 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : Problem switching from Windows to Ubuntu: ROCm question

©2024 Astroinformatics Group