Return to BSD News archive
Newsgroups: comp.unix.bsd.freebsd.misc Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.EDU.AU!munnari.OZ.AU!news.ecn.uoknor.edu!news.wildstar.net!newsfeed.direct.ca!imci2!news.internetMCI.com!newsfeed.internetmci.com!in1.uu.net!news.thepoint.net!news1!not-for-mail From: root@dyson.iquest.net (John S. Dyson) Subject: Re: Linux vs. FreeBSD ... X-Nntp-Posting-Host: dyson.iquest.net Message-ID: <4n2q6b$5v9@dyson.iquest.net> Sender: news@iquest.net (News Admin) Organization: John S. Dyson's Machine References: <3188C1E2.45AE@onramp.net> <4mnsc5$6qo@sundial.sundial.net> <4mr1pk$cdi@dyson.iquest.net> <4n0dhd$cff@agate.berkeley.edu> Date: Sat, 11 May 1996 19:35:39 GMT Lines: 145 In article <4n0dhd$cff@agate.berkeley.edu>, Nick Kralevich <nickkral@america.CS.Berkeley.EDU> wrote: >Oh, by the way, I removed comp.os.linux.misc from the newsgroup line, >since your flamebait is off topic there. > >In article <4mr1pk$cdi@dyson.iquest.net>, >John S. Dyson <root@dyson.iquest.net> wrote: >>This is getting wierder and wierder. FreeBSD is a very open consortium >>of developers. We have 3-5 developers actively committing to the kernel, >>for example. Linux is the closed (controlled) development. > >Yep, Linux is closed, which is why you see hundreds of people >actively hacking on the kernel, vs FreeBSD, which has much >fewer. > Nope, they are not actively hacking on THE kernel -- Linus filters everything. We have many people hacking on the FreeBSD kernel (CVS tree.) There might be alot of contributors getting their modifications included, but Linus reserves the right. > >The difference is that, for good or bad, FreeBSD is more centralized, >where Linux is more distributed. To many people, a more distributed >development environment provides more freedom. Both systems have >their advantages and disadvantages. > Depends on what you define the terms to be, I guess. Linus controls *everything* in Linux and has the final say-so. The distributions are more diverse though. FreeBSD is a full OS -- Linux is a kernel with various distributions that might or might not be compatible. FreeBSD has over 50 people who can commit *directly* to the CVS-tree. How many can commit to the Linux source code control tree??? Hmmm.... Oh, it doesn't have one? I guess that is what you mean by free and open... Sure doesn't sound like a professional development either... >>> My Adaptec AHA-2842VL (an older VLB SCSI-2 Fast Host Adapter) craps >>Sounds like a broken Linux driver to me!!! > >Unless you have some evidence to back that up, it would be wise of >you not to post such FUD (fear, uncertanty, doubt). However, >if it's a broken Linux driver, than prove it. > Remember, the Linux driver tracks the FreeBSD driver now... You know, it is in catch-up mode all of the time... > >>FreeBSD is primarily a U**X clone. That means that we want to support as >>many U**X platforms as possible. > >Hmmm, is FreeBSD available on the Alpha, 680x0, power PC, Sparc, etc? >Linux is. > NetBSD is available on many more than Linux, so Linux doesn't have that yet either. > >It might be better if you said that FreeBSD wants to support as many >U**X platforms on the Intel architecture as possible. > Right now that is true, and I stand corrected on that. FreeBSD will probably never support more than the 2-3 predominant platforms. > >The reason why no one on the Linux side has written FreeBSD support >is because, quite frankly, no one is that interested in it. There has >been support for the BSD UFS to be added, but until someone finds a need >to run *BSD binaries, it's not likely to happen. > That is sad for Linux users, because the original simplication of FFS/UFS into EXT2FS dropped fragments... >>Actually, my latest current vs. a very recent Linux shows that we perform >>about 10% faster at one of our previously worst performance numbers: fork > >Can you please provide the actual numbers, the testing methodology, >software and hardware used, etc... so we can make an informed decision. > lmbench -- lat_proc... P5-166, ASUS TP4N, Linux 1.3.9x, FreeBSD-current. Note that Linux is probably much closer to release than FreeBSD-current, so I expect the FreeBSD release to be faster than the -current version. >>Yep, our IDE/ATAPI support is a little rough. That is one of the legit >>problems with FreeBSD right now. > >Linux IDE/ATAPI support has been called one of the best around. > So? You are not disagreeing with me, are you -- Are you trying to sell Linux here??? Wrong newsgroup... > >Again, try posting some facts, like numbers, that show your position >to be correct. > Do iozone -- I get 13-15MBytes/sec max cached on Linux 1.3.9x ... -current on FreeBSD gives me about 35MBytes/sec and our experimental code is approaching 40-60MBytes/sec. Other benchmarks like Bonnie, etc show similar results. Nick, if you don't believe it, and since you know so much about FreeBSD, please try running and benchmarking it, before you try to question my credibility. The differences are clear... > >>If you want to compare the system you must >>mount the FreeBSD disks using the dangerous (like Linux) -o async option. > >Dangerous, huh? LOTS of people run Linux daily, and never have the >problems you are describing. You are, quite frankly, full of it. > Not true, you WILL loose more data -- period. If you don't write the data, IT WILL BE LOST!!! You are, truely silly if you don't understand that. >>(BTW, if you try to upgrade your >>a.out Linux system to a nice (but wasteful) blocksize for ext2fs -- many of >>your binaries won't run.) Of course, FreeBSD with it's efficient 4K/8K/16K >>blocksize can handle the binaries just fine. > >Again, more FUD. If you think this is true, provide examples! >Otherwise, stop spreading your misinformation. > Do the upgrade to 4K or 8K ext2fs -- you simply don't know. 1K offset a.outs haven't worked on various versions of Linux. I have noted that complaint recently on a Linux newsgroup. >I did the upgrade myself, from a Linux a.out system to a Linux ELF >system, without reformating the filesystem, and had none of the problems >you've described. I've never even heard of ANYONE having this problem, >and I follow the Linux newsgroups pretty closely. Again, prove it. > That is not the upgrade that I am talking about -- non-sequiter... You must be confused. The ELF binaries have the correct offset, and don't cause the problems above. Those binaries work well on FreeBSD also. In fact, I run the Linux-ELF lmbench suite. Kind-of eliminates variables in my benchmark runs. You accuse me of FUD -- you are the pot calling the kettle black. Your ignorance of FreeBSD is so complete and you make so many statements without running or testing FreeBSD, that you can only be accused of disinformation. I do read the Linux newsgroups and run Linux on the same machine that I run or benchmark FreeBSD on, so how much space have you dedicated to FreeBSD to become such an expert? John