VirtualBox 4.0 vs VirtualBox 3.2 benchmarking - a non-scientific approach
As promised in a previous post I finally got around to running a very simple test that can be a good indicator of performance change.
I have compiled OpenSSL from port on both VirtualBoxes. Everything else is the same, including all options in VirtualBox.
Just to be clear though, this isn't really a scientific way to do this. I should have run more tests and tried different things too, this is basic but it will give us a rough idea.
Also to make sure that download times did not affect this benchmark I made sure to do a 'make fetch-recursive' before the 'make install clean'.
What the 'make fetch-recursive' does is, it downloads the source code for the port and all of it's depending ports prior to compiling. No real processing takes place here. It's mostly dependent on your Internet connection.
Also, to make sure that no delays are added by the time needed to configure port options, I also run 'make config-recursive' - What this does is to display all the options for the port and all of it's depending ones.
After all that, I went on and did a 'make install clean' as we have seen on 'Ports 101'.
To summarize the steps:
1. cd /usr/ports/security/openssl
2. make fetch-recursive
You will see something like this:
3. make config-recursive
Typical options screen for a port: (you navigate through it with the TAB button)
4. make install clean
RESULTS:
VirtualBox 3.2
VirtualBox 4.0
From these results we can see that the difference is about 12 seconds improvement when doing the compiling under VirtualBox 4.0