Interesting!fb wrote:Without kvm-enable it's ~850k exits, with kvm-enable it's ~550k exits. Funnily its considerably faster with more exits
One could spend days or weeks with further experiments to investigate the background reason, but I'm not sure if it's worth the effort, if you have a working solution. One example for an experiment to start with: You could run the VirtualBox VM twice, with the QEMU VM spending some considerable effort, once with and once without KVM, doing the same actions at approximately the same time for approximately the same duration, and compare the VM statistics in the VBox.log files, to see which components show considerable differences. But that would possibly be only the beginning of a really deep dive ...
I read about it, and other users trying HAXM inside VirtualBox VMs had similar problems like with KVM, when using VirtualBox 6.1.18 and older. The KVM problem has been solved within VirtualBox 6.1.20 and newer, and there have not been any reports yet regarding HAXM (except yours ).fb wrote:Do you know about Intel HAXM (https://github.com/intel/haxm)?
Googling for only "Host CPU does not support APM" will lead you to the HAXM source code, and the comments therein will lead you to the Intel SDM part 3B chapter 18.2 Architectural Performance Monitoring. I'd assume that VirtualBox does not provide the APM functionality to the VM. But it seems to be only a warning ...fb wrote:haxm_warning: Host CPU does not support APM