Minix 3.2 Released, Uses LLVM/Clang, SMP, ELF

[cl333r]

Year 2069, Minix 14.2 released, a whooping 314 full-time users, they all keep showing up academic papers "proving" how it's the best designed OS in the world, and certainly better than Linux.

[vertexSymphony]

Yes, I would like to see some benchmarks of Minix and a monolithic OS like FreeBSD or GNU/Linux just to have some reference numbers in the impact of IPC on performance.
Proper benchmark please, same hardware if possible, NO VMs and no GUI (you could actually make a benchmark with GUI and without) ...

I would love to see that benchmark.

[sylock]

Here:

https://lkml.org/lkml/2012/2/26/69



Tanenbaum, go home with that crap.

The benchmarks your are linking is really outdated (2007). It would be worthwhile to see up-to-date benchmarks to see where is MINIX in regards of performance.

[curaga]

Yes, lets bench a research system :P

Willing to bet it loses head-first to even Hurd, simply due to its different goals.

[renox]

Yes, lets bench a research system :P

MINIX 3 is not only for research, its website says "an open-source operating [system sic] that can be used as a base for research projects but also for commercial (embedded) systems where microkernel systems dominate the market. Much of the focus on the project is on achieving high reliability though fault tolerance and self-healing techniques."

That said when the FAQ was written, the disk driver didn't have DMA(!)..

[markg85]

Benchmarks of minix compared to linux would certainly be interesting.

[Wingfeather]

+1 for MINIX benchmarks.

[Nobu]

I tried to download the iso, but the connection kept being interrupted mid-download, and wget didn't want to --continue from where it left off.

Is there a mirror or a good torrent somewhere? I searched, but only found a torrent on TPB, and the file size was wrong.

[vertexSymphony]

Yes, lets bench a research system :P

Willing to bet it loses head-first to even Hurd, simply due to its different goals.

We assume that is going to be slower, but it's interesting to know "how much" ... we also know that here we trade off performance for a theoretical stability (that's the goal, not performance)
Just, pointing out the obvious, Regards.

[Nobu]

Actually, rather than "how much slower", I'd be more interested in "how much more stable". Has anyone tried writing (for instance) a real network driver that causes havoc on purpose?