[vfio-users] Poor performance with nvidia GTX 980

[Alex Williamson]

On Sun, Nov 1, 2015 at 10:40 AM, Georgios Kourachanis <
geo.kourachanis at gmail.com> wrote:

> Intel Core i7 3770 ( I use the igpu for the archlinux)
>
[snip]

>   <memoryBacking>
>     <hugepages/>
>   </memoryBacking>
>   <vcpu placement='static'>4</vcpu>
>   <cputune>
>     <vcpupin vcpu='0' cpuset='2'/>
>     <vcpupin vcpu='1' cpuset='3'/>
>     <vcpupin vcpu='2' cpuset='6'/>
>     <vcpupin vcpu='3' cpuset='7'/>
>   </cputune>
>
[snip]

>   <cpu mode='host-passthrough'>
>     <topology sockets='1' cores='4' threads='1'/>
>   </cpu>
>
[snip]

>
> 1) Using "<topology sockets='1' cores='4' threads='1'/>" instead of "<topology
> sockets='1' cores='2' threads='2'/>" provided about 2% boost in GPU
> performance. No change in RAM or CPU tests. I've tested with the passmark.
>
> 2) I tried using the emulatorpin method Alex says on a mail here on
> vfio-users, but I didn't notice any changed in GPU performance. I didn't
> test it on the CPU side though.
>

So now that you're using virtio networking and you're in the ballpark of
performance, why are you comparing bare metal, where the game has 4 cores +
threads to a config with only 2 cores + threads?  Have you tried giving it
more cores?  Less threads?  Have you tried using the sibling threads only
for emulatorpin?  Have you tried isolcpus and nohz_full to dedicate as much
CPU resources as you can bear to the VM to test how close you can get to
bare metal?  In my experience, threads are somewhat useful in the host, but
you really need to be careful how you make use of them for a VM.  A
quad-core may really only be marginal for running a high performance gaming
VM.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/vfio-users/attachments/20151103/7f5c2114/attachment.htm>