Cap of performance ratio of VM/dual boot setup
Posted: 6. Jun 2017, 20:13
Hi, I am thinking about dual booting Windows 10 with Arch or running it in a VM (Virtualbox). I'm looking for some numeric data that could help me decide and some speed tweaks if there are any.
So far I could come up with a calculation-test (That is mainly a numpy calculation, about 220,000 matrix product calculation and a few times 100,000 condition evaluation. I can give the link to the program but I guess it's irrelevant.). I run the program with the exact same parameters and number of calculations.
The results I got so far:
Windows 10 "clean install" on a computer with 4 GHz processor with 4 GB RAM: 17.5 seconds.
Windows 10 VM on my computer with 4x2,7GHz processor and 8 GB RAM, the VM has 4 GB RAM and 1 virtual CPUs (setting more CPUs decreased performance): 37.5 seconds
Arch Linux (the host OS of the previous VM): 5 seconds.
The ratio that I'm currently investigating is: time(VM)/time(non-VM install) that is currently around 0.47.
My question is: is it possible to get a VM closer to the non-VM install with some trick? Do you know some good resources on this topic? Have I already reached the top of my VM's capabilities?
More info: dinamically allocated virtual hard disk,
Guest additions installed
VT-x/AMD-V is enabled
Nested Paging is enabled.
Video memory: 128 MB (capped).
I wish to use my Windows for simulation purposes that seem to be quite resource intensive. Do you think that I had better do that on a dual boot setup?
So far I could come up with a calculation-test (That is mainly a numpy calculation, about 220,000 matrix product calculation and a few times 100,000 condition evaluation. I can give the link to the program but I guess it's irrelevant.). I run the program with the exact same parameters and number of calculations.
The results I got so far:
Windows 10 "clean install" on a computer with 4 GHz processor with 4 GB RAM: 17.5 seconds.
Windows 10 VM on my computer with 4x2,7GHz processor and 8 GB RAM, the VM has 4 GB RAM and 1 virtual CPUs (setting more CPUs decreased performance): 37.5 seconds
Arch Linux (the host OS of the previous VM): 5 seconds.
The ratio that I'm currently investigating is: time(VM)/time(non-VM install) that is currently around 0.47.
My question is: is it possible to get a VM closer to the non-VM install with some trick? Do you know some good resources on this topic? Have I already reached the top of my VM's capabilities?
More info: dinamically allocated virtual hard disk,
Guest additions installed
VT-x/AMD-V is enabled
Nested Paging is enabled.
Video memory: 128 MB (capped).
I wish to use my Windows for simulation purposes that seem to be quite resource intensive. Do you think that I had better do that on a dual boot setup?