Multithreading difference in Windows and Linux

Hello, dear experts.
Is it possible that the same project will work with different calculation speed at the same computer? I accidentally find this on my work task. Usually i perform calculations on linux virtual machine. The host system is win10 64 bit, and guest system is linux mint 20.02.
On the host system i successfully install geant4.10.07.p02, as on a guest operating system.
But problem is that on guest system my project working about few times faster than on windows system. I found that this actually happen when I’m working with HP physics model.
My project based on Activation example, but i found the speed differences in other simple examples too.
This confused me a bit, because i planned to move from linux to win.
Also i don’t know exactly what category fit this problem precisely.
Any ideas?

I’ve checked this on g4.11 and yes, there is also slow down when modeling neutrons with HP model (QGSP_BERT_HP in particular). I’ve modified B1 example by changing projectile particles from gamma to neutron, and run model with 1e+6 events in beamOn.
So in Linux system i get average calculation time about 10 seconds, while in Win10 it was about 100-110 seconds.
In both cases CPU and number of worker threads was the same: 16 worker threads, and CPU with 8 cores with HyperThreading, each core with 4.7 GHz.
If anyone know what i’m doing wrong or how to explaine this, tell me please, cause i still planing to migrate from linux to Win10 system on my work, and this is only one that stopping me.
Thank you.