It seems that my mail client butchered my response to you so that it was hard to see my in-line comments. I thought I would re-post a cleaned-up version:
> It was discussed several time in the past, last I remember the issue is
> mainly with Windows guests,
> There are some hints to solve it in this
> page:
> http://serverfault.com/questions/624690/windows-guest-on-kvm-qemu-suffers-horrible-key-lag
Yes I saw this too when researching. However my guest is Fedora 21, not Windows.
> (iirc,
> using a usb keyboard would solve it too)
I did see this also and played around with changing the input type of the keyboard to USB in my xml file. However it appears that QEMU creates the PS/2 keyboard regardless and thus I end up with two keyboards and not sure which it is using. I saw the same lag with this approach.
> I am not convinced at this point that the
> keypress-delay is problematic. Do you have a particular use case, or simply text
> typing?
Use case is any sort of regular typing in a VM, whether in gedit, vi in a terminal window, or typing an email in a browser window. It is a slight but noticeable delay when typing in the VM vs. on the physical host. The VM is running locally so I don't need there to be any delay to compensate for network conditions.
In the link from 2014 I previously posted, the person who recompiled with 10ms value for keypress-delay saw a huge improvement in the overall feel. My hope for a solution is that the keypress-delay can be auto-tuned based on network conditions or user configurable. Thank you for considering this.
Thanks,
Jared Kwek
> It was discussed several time in the past, last I remember the issue is
> mainly with Windows guests,
> There are some hints to solve it in this
> page:
> http://serverfault.com/questions/624690/windows-guest-on-kvm-qemu-suffers-horrible-key-lag
Yes I saw this too when researching. However my guest is Fedora 21, not Windows.
> (iirc,
> using a usb keyboard would solve it too)
I did see this also and played around with changing the input type of the keyboard to USB in my xml file. However it appears that QEMU creates the PS/2 keyboard regardless and thus I end up with two keyboards and not sure which it is using. I saw the same lag with this approach.
> I am not convinced at this point that the
> keypress-delay is problematic. Do you have a particular use case, or simply text
> typing?
Use case is any sort of regular typing in a VM, whether in gedit, vi in a terminal window, or typing an email in a browser window. It is a slight but noticeable delay when typing in the VM vs. on the physical host. The VM is running locally so I don't need there to be any delay to compensate for network conditions.
In the link from 2014 I previously posted, the person who recompiled with 10ms value for keypress-delay saw a huge improvement in the overall feel. My hope for a solution is that the keypress-delay can be auto-tuned based on network conditions or user configurable. Thank you for considering this.
Thanks,
Jared Kwek
_______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel