Screaming Penguin

The evil empire and its hoards of marketers are further devising schemes to attack the rebel outpost of Linuxsia. But the rebels will not accept defeat and vow to fight and conquer the goliath entity. Well, OK maybe its not that dramatic, but Microsoft has again devised a benchmarking test that is heavily swayed in their favor, and the opening skirmishes have already begun. Microsoft and Mindcraft are clearly upset at the outcry their earlier "independent" testing has evoked. The new test comes at the PC Week labs in CA. The lines are drawn and a team of Microsoft marketers (including the cons who bought the first game) are pitted against 2 RedHat engineers and 1 Penguin Computing engineer. At least Linux has some real representation this time. However, Linux is not as tried on high end SMP machines as it is on normal PCs and single CPU PC Servers. The mega 8x+SMP world is relatively new to Linux and this is a class where optimization truly counts. This should be very interesting, but it may still be a short time (for further configuration and kernel tweaking) before Linux wins the upper echelon battle. Once the Linux community has had a chance to look at the testing criteria (apparently still designed or suggested by the NT crowd) and formulate a strategy, it wont be a contest. Linux, tweaked and configured will kick some NT arse. Its just a matter of time (and nowhere near the time it takes NT to adapt to new challenges and requirements.) Linux will win all of this in the end, and Microsoft only have themselves to blame for the wrath of the "Linux fanatics", they started the pissing contest. I am looking forward to the results of this bout. The animosity is enormous. The original "MindLessCraft" president was on hand to have the Penguin Computing chap call the original test "benchmarketing" and poke fun at the "independent test lab" quotes throughout the original report. He responded and stated that this was "challenging his integrity." Ha Ha Ha. Regardless of the outcome, I am glad to see a true independent test. Frankly I am past the Linux-NT battle, I use them both and there is no real comparison (yeah if you can manage to hold your mouth right for long enough and stand on the right foot while you figure out the service pack install order and then find the secret unpublished-they-might-blemish-our-reputation hotfixes, maybe you can make NT faster at a few tasks on mega servers, but so what, that certainly is not worth the cost, maintenance and usability overhead that comes along with it.) I would like to see more benchmarks in the area of Linux vs. Linux, distribution vs distribution (out of the box tests, regardless of tweaked perf) and SMB(samba) vs NFS vs NCP(ncpfs.) I would also like to see other packages such as MySQL and MiniSQL compared for speed (yeah MySQL is better, cause its free, but I am still curious.) In the future we may plan some of these tests at TOTSP using real world single processor "out of the box" configurations (if we can find time.) The latest flap on this issue comes from this story in the Register. Checkit.