Python tests used to compare different OSes

This is a very poor way to compare operating systems. Each has different optimizations by default. That, and many many other factors make this a terrible project. It's unfortunate that so much work was put into testing so few things when many more things need to be considered. The set of data (machine variations, software tuning, etc) are far too small. It doesn't matter how many times one runs the same test on one machine if it's not the machine he's really testing, but the operating system.

Now if it was only a matter of what operating system without tuning works best on this one machine without tuning, then this might just legitimate.

Great. Wonderful.

So what does it mean to "win" the test? What do any of those numbers even mean?

Doesn't it test a Python implementation on different OS rather than the OS itself?

python is a platform independent language. What tests does this run?

Very interesting, but these OSes are actually just 2 branches of the "unix" tree of OSes. Too bad he didn't include Windows, or OpenVMS, very different operating systems.

The only experiment using Windows NT vs NetBSD that I'm aware of is this: "The Measured Performance of Personal Computer Operating Systems" (http://www.eecs.harvard.edu/~margo/papers/sosp95/). On the same hardware, for cross-platform tasks, NetBSD outperformed Windows NT. This was in 1995, so it was probably Windows NT 3.51 and NetBSD 1.0 or 1.1