Date: Wed, 5 May 1999 10:58:17 -0700 (PDT) From: <unknown@riverstyx.net> To: "James A. Mutter" <jmutter@netwalk.com> Cc: David Schwartz <davids@webmaster.com>, John Reynolds~ <jreynold@sedona.ch.intel.com>, freebsd-chat@FreeBSD.ORG Subject: RE: Mindcruft ... Message-ID: <Pine.LNX.4.04.9905051049470.14073-100000@hades.riverstyx.net> In-Reply-To: <Pine.BSF.4.05.9905051442001.9655-100000@insomnia.local.net>
next in thread | previous in thread | raw e-mail | index | archive | help
I don't think you can really do that -- every OS has different hardware that works well with it. An extremely obvious example in the Mindcraft tests was the choice of RAID controller. Worked great with NT, flopped on Linux. Then Mindcraft tried to compare that with a different benchmark using a different RAID controller under Linux, claiming that the other benchmark (I think it was PCWeek?) was faulty in some way, or that the people who did that benchmark weren't giving out enough info on how they conducted the test. However, the RAID controller driver that Mindcraft used was still in beta development, and was single-threaded, as compared to the heavily tested and fully functional multi-threaded driver used in the PCWeek test. I've heard from other people that the RAID controller used in the PCWeek test doesn't perform so hot under NT. Where's the middle ground? Find some card that performs the same under both, then test the network infrastructure? Then people can claim that the poor design of the RAID driver was using too much CPU, and adversely affecting the rest of the system. Or that "in a perfect situation" the Linux driver could be made to work better than a perfectly tuned NT driver, meaning the actual operating system was better, and just lacking in drivers? Maybe a better way would be to set an amount of money, then let each team choose the hardware in the budget, based on list prices from the manufacturers. Each team gets a $15000 server and then they go head to head on performance. --- tani hosokawa river styx internet On Wed, 5 May 1999, James A. Mutter wrote: > :> Generally, a fair test on a level playing field. > : > : Same hardware is not a level playing field. Sensible people optimize the > :hardware for the operating system they plan to use. I could show you Linux > :and NT beating the stuffing out of FreeBSD by using my MX98713 network card > :in 100Mbps half duplex mode. For some reason FreeBSD chokes on it. > : > : DS > Well then, you need to find hardware that is equally agreeable, or as > close to equally agreeable as you can get, to _all_ operating systems. > Running the tests on different hardware just gives the critics > one more thing to be critical of. > > I don't belive that you can have a "level playing field" without the > same hardware. > > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-chat" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.LNX.4.04.9905051049470.14073-100000>