A VM can have very different characteristics from a real hardware when it comes performance of nested virtualization, so I would suggest testing on a real hardware and see you observe the same performance degradation if performance on a real hardware is actual concern. On my testing, I did not see that.

Also, you can measure which functions are slow with HYPERPLATFORM_PERFORMANCE_MEASURE_THIS_SCOPE macros in the scope granularity. I suggest you to take numbers on both 2 and 4 logical core machines, and compare them.

While I wish to make this project as performant as possible, I do not plan to actively work on this issue as performance is not a priority. If you can take data, I can perhaps assist you to find the root cause and a fix, however.

I tested on real hardware, ddimon works fine! so I assume that there are some bugs in vmware