Micro553

I understand and I’m sorry to be in the wrong place. I will post in vm part. Thanks for the help with everything. Have a good one!

Hi. I did look and I have not bind the gpu and gpu audio in vfio. Will try that now to see if it helps. I just want to use my card in my win11 Vm. I have updated to RC3 and all works well as far as I can see. Except my problem with driver update. I have this in syslinux video=efifb:off is this not needed anymore? I did put it in a long time ago because I had trouble with my last gpu a Rx 5700 xt

I have 6.9.2 yes. I did run the script thing to run win11 without TPM etc. I use and want to use the gpu in my win11 VM. Just to be clear. My win11 works but im unable to uninstall or install newer adrenaline softeware. Im not sure if I have bonded it to vfio. Will have to check. Now I’m running memtest and it looks good so far. my plan is to update to rc3 and then boot VM without gpu and try to uninstall the current adrenaline. Then pass the gpu and see if newest version will install. thabks for all help. so if I understand - I can’t run your plugin and a vm at the same time with the same gpu?

Thanks. I’m having some issues with my win11 vm and my Amd adrenaline software / driver. Basically I’m on a version that does not allow u to check updates from the gui. So in hope to fix this I have tried a lot of adrenaline versions. Even Amd cleanup tool BUT every time when the “checking your graphics hardaware screen” comes up it freezes the server. Sometimes it reboots just the VM sometimes it reboots the whole server. I’m now running a memtest but doubt that is the problem. you say that my driver gets loaded and then unloaded. Could that be the cause for my problem with not being able to update my Amd adrenaline software and drivers? So maybe newest RC will help me?

Here are my diagnostics from yesterday Thanks. unraid-diagnostics-20220315-1307.zip

Did try to install Radeon Top and GPU statistics todag and missed the part about setting model vendor. So after deleing both apps from the usb im able to access the GUI again. The problem was that I could not access the webgui at all. Then I read the following from the text about the plugin "no editis to the 'go' file or creation of other files are necessary (please not that this plugin only enables the 'amdgpu' Kernel module and not the 'radeon' Kernel module)." so this got me thinking because I have this (have been there long time) in my syslinux append initrd=/bzroot video=efifb:off So how would I do to make this work? 1. Install GPU statistics 2. Install Radeon TOP 3 Remove enything from syslinux? 4. Chooce model vendor where? 5 Reboot server I have a AMD RX 6900 XT GPU passed to my win11 VM if that matters.

Thanks. Did work. But do u think that me not choosing the model vendor did this? Maybe try again so is this right 1. install Radeon top 2. install gpu statistics 3. Chooce vendor etc where? 4 reboot I have Rx 6900 xt

Will try that. Thanks.

Did install the Radeontop and GPu statistics and missed the part that I should have selected vendor etc in settings. so just installed the 2 plugins and did a reboot. Now I can’t access the web gui for Unraid. but the server starts etc. so tried the safe boot option and got access to gui. But could not uninstall the plugin because no plugins are loaded in safe mode. so I’m kind of stuck. My quess is that because I did not chooce vendor the plugin somehow blocks the webgui or similar. Did read that if a scrips is not run to finish it blocks the gui. so how would I go ahead to just uninstall them without gui.

Hi Got a tip from FB about Radeon Top so I installed the app and also installed the GPU Statistics. After that I restarted the server and now I can´t access the webGui. I did access via SSH and ran diagnostics and did get the latest zip file from USB. Somehow the timestamp is wrong. I did not run diagonsics this morning but just 20 mins ago. Not sure why this is. GPU is Red Devil RX 6900 XT. Happy if someone can help. Can the gui be restarted from SSH perhaps? Did google but did not get a clear answer. Thanks unraid-diagnostics-20220315-1307.zip

So been running your proposal for a while. I mean pinning 8 -15 for a 8 core VM. It seems stable but the thing thats bugging me (not a big deal but still) is that with the old pinning I got alot better FPS playing warzone. With 8 core (2,3,67,10,11,14,15) and my Rx 6900 xt I got over 200 fps. Now I´m at 150-160 fps. Not the end of the world but been reading other having problems with the RX 6900 xt where the cpu was the bottleneck. I will try to give my VM 4 more cores (non pinned) and see if that helps.

Thank u so much for the great explanation. I have now changed and pinned 8 - 15 as u said. So 8 cores / 16 threds on the same chiplet pinned for win10. Are there any perfomece benefits to assign 4 more cores like non pinned core 4,5,6,7? As I´m running now is this the same as running a 5800x BM?

I just went from 3700x to 5950x. I runa daily driver/gaming win10 VM with a passed red devil rx 6900 xt. When I was on the 3700x I isolated cores 2,3,6,7 because of how the cpu is builed. I got tips from here that this was the way to go. Now with the 5950x I´m kind of lost. I think i undersland that 5950x is builed diffrent. Right now I have cores 2,3,6,7,10,11,14,15 isolated and choosen in the gui for win10. But maybe thats not the best way to have it? Maybe isolate 8-15 instead. Or 8-15 plus 4 more that are shared with the rest for a 12 core vm? I want to have a fast stable win10 but also some headroom for the rest. I think I have a cpu that can manage Other then win10 I run about 20 dockers, Plex, shinobi, nextcloud, glances, the *arr ones etc I also run Home Assistant as a vm and have given it 8 cores (non of the isloated ones)

Same here. Did manage it to plot to ssd (second cache) but I have not manage it to actually move the plots to the array. Plots directory appears to be empty and you have not specified another, try mounting a plot directory with the docker -v command Added plot directory "/plots".

Why is my /plots dir not working Getting ths in log: Plots directory appears to be empty and you have not specified another, try mounting a plot directory with the docker -v command Added plot directory "/plots". The chia_stor is a shared folder on the array set to not use the cache. And chia_plots is a folder in the folder. I quess I´m stuupid but can´t understand why its not moving plots from /plotting to /plots