[Request] Setup guide update for AMD V-Cache CPUs #14
Comments
|
Hi @ggeorgo, that's a bit difficult, since I do not have access to a 79X0X3D at the moment. From what I heard, the VCache is always on the first die, which would mean that if there's no way to tell the bios to boot from the second one, you'll end up with core 0 being used by Linux. A description of the problems that brings is here on reddit. It is still worth a try to just isolate the first die and just live with the flaw that the host schedules some work there. I have no data on how this affects things. You'd basically do something like this to isolate host execution to the second die, cores 1-7. Good luck, and report back with results! |
|
Thank you Martin for your prompt response.
Actually, that thread on Reddit is mine :D
Will update both, as I am setting up my new box next week.
Regards,
G
…On Sat, 1 Apr 2023 at 12:22, Martin Schrodt ***@***.***> wrote:
Hi @ggeorgo <https://github.com/ggeorgo>,
that's a bit difficult, since I do not have access to a 79X0X3D at the
moment.
From what I heard, the VCache is always on the first die, which would mean
that if there's no way to tell the bios to boot from the second one, you'll
end up with core 0 being used by Linux. A description of the problems that
brings is here on reddit
<https://www.reddit.com/r/VFIO/comments/11ni06e/pinning_and_isolation_of_7950x3d/jbx6a7g/>
.
It is still worth a try to just isolate the first die and just live with
the flaw that the host schedules some work there. I have no data on how
this affects things.
You'd basically do something like this
sudo vfio-isolate \
cpuset-modify --cpus C9-15,25-31 /system.slice \
cpuset-modify --cpus C9-15,25-31 /user.slice
to isolate host execution to the second die, cores 1-7.
Then configure the VM to use the complete first die for the VM, and
emulation on second die, core 0.
<cputune>
<vcpupin vcpu='0' cpuset='0'/>
<vcpupin vcpu='1' cpuset='16'/>
<vcpupin vcpu='2' cpuset='1'/>
<vcpupin vcpu='3' cpuset='17'/>
<vcpupin vcpu='4' cpuset='2'/>
<vcpupin vcpu='5' cpuset='18'/>
<vcpupin vcpu='6' cpuset='3'/>
<vcpupin vcpu='7' cpuset='19'/>
<vcpupin vcpu='8' cpuset='4'/>
<vcpupin vcpu='9' cpuset='20'/>
<vcpupin vcpu='10' cpuset='5'/>
<vcpupin vcpu='11' cpuset='21'/>
<vcpupin vcpu='12' cpuset='6'/>
<vcpupin vcpu='13' cpuset='22'/>
<vcpupin vcpu='14' cpuset='7'/>
<vcpupin vcpu='15' cpuset='23'/>
<emulatorpin cpuset='8,24'/>
<iothreadpin iothread='1' cpuset='8,24'/>
...
</cputune>
Good luck, and report back with results!
—
Reply to this email directly, view it on GitHub
<#14 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ALVA5R33XY7HCWU7WPCICNLW7AFW5ANCNFSM6AAAAAAWPQYDFY>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Any updates? We're on the edges of our seats here ;-) |
Sorry, forgot to update, however this is the 1st weekend I am using it, since I was battling with a bug for 10 days... |
|
Hi @spheenik ! ` enable_isolation () { and I am getting below errors: when I omit "cpu-governor performance" I do not get the errors |
|
I am not sure what changed in the last month or so, however, isolation stopped working for me. I have tested this before, and was working fine, but seems that lately it is not, as I run “stress --cpu 64” while my VM was running and to my surprise, all 32 threads went up to 100%, while previously this was the case for half of them and isolation would prevent VM cores to be used by the host. My system (I can post more info if needed): My XML file: and my /etc/libvirt/hooks/qemu file (which is marked executable) Maybe an incompatibility with latest Kernels? |
|
Unsure what the problem is. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
With all the 79X0X3D processors out there, is it possible to update the setup guide with some examples on how to:
The text was updated successfully, but these errors were encountered: