What are the key differences between mvisor and QEMU?

[fpemud]

I have a few questions regarding the key differences between mvisor and QEMU.

I think it is unlikely that mvisor would outperform QEMU in terms of speed, since both of them are using KVM.
Does mvisor have a significantly smaller memory footprint or lower CPU usage compared to QEMU?
Does mvisor offers any additional functionality than QEMU?

Specifically, I would like to know if mvisor can be used as a gtk-widget or qt-widget, allowing it to be seamlessly embedded in GTK or QT applications. QEMU + spice-gtk-widget solution is heavy and complex, since QEMU needs to be managed as a separate process, and QMP protocol and SPICE protocol must be used to interact with it.

[nooodles2023]

Thanks for your attention.
Just as the name suggests, mvisor is a mini-hypervisor, which means it's simpler and lighter than qemu and easy to extend. We use mvisor in our production environment. In our use case, many VMs need to be resumed quickly at the same time on a server. Therefore, we have done a lot of work to improve the speed of snapshot resuming and to reduce memory usage. It has been proven that its performance and compatibility are now better than those of qemu.

We have implemented opengl-vgpu device for windows vm, which qemu has not implemented yet. https://github.com/tenclass/mvisor-win-vgpu-driver

Of course, it can be used in other graphic interfaces, just like SDL.