A few years ago, I was an avid QNX user. I used the non-commercial desktop version of this wonderfully clean and elegant pure-microkernel operating system for a long period of time, as a desktop operating system. I liked the whole style of this operating system, its Photon user interface, and its excellent package management system. I even wrote a three-page article about it. Sadly, QSS, the company behind QNX, lost all interest in the non-commercial desktop version, and ditched it, leaving only a hard-to-find 30-day evaluation version alive. Community interest dwindled, and so did mine. Despite my lost interest, it saddened me today to learn that QNXZone.com, a community portal for QNX, has been shut down. Read on for a few short thoughts.
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:2005-07-24
Well, off the top of my head, how about this:
http://lwn.net/Articles/220255/
Member since:2007-07-27
well, they aren't just comparing the performance of the kernel but of the whole os. it seems to be relatively simple to write a fast second generation microkernel, but it's a huge endeavor to build an usefull new os around this kernel. in most cases, an old os is bolted on top of the microkernel, and the additional layer costs about 5% performance in case of l4linux:
http://os.inf.tu-dresden.de/pubs/sosp97/
even the l4darwin/darbat project, which started as an attempt to swap mach with l4 in the xnu-kernel stopped trying and implements xnu on top of l4:
http://www.ertos.nicta.com.au/publications/papers/Lee_Gray_06.pdf
as such, microkernels are probably more usefull as hardware-abstraction layer than to build a full fledged os.
Member since:
2005-06-29
Except for Minix which, despite what Tanenbaum claims during his combination dog and pony show / microkernel pep rallies, performs abysmally.
Proof? Benchmarks? Wow me, sbergman27.