After much development and four previous releases, FreeBSD-5 has finally gone stable with version 5.3. But don’t mistake a stable codebase with stable software. While the development team will no longer accept major changes to the base system, FreeBSD 5.3 still has bugs and problems. NewsForge has the review.
The FreeBSD website clearly marks it as their production release and ever since the 5.3-RELEASE, my servers and desktops have been running it, only going down when I purposefully rebooted them to update to 5.3-STABLE. I trust the FreeBSD team’s ability to tell us what their product is and yes it has bugs and problems but what doesnt?
yeah! I’ve been using it too for my webserver. The 5 branch went through a considerable number versions before. I think I trust the FreeBSD team more re their discretion of what they tag as stable or not, rather than this author.
I’m loving FBSD 5.3.
This is truely pathetic. Newsforge.com is just running this junk article to get ppl to read it for various reasons.
The person making these claims does not even bother to support the claims. They just hyper link to pages that proove little especially with a bit of investigation.
Having moved from 4stable to 5stable, I feel quite safe in saying it is ready for desktop and server work. Yes, there are a few problems, but for the most part nothing that is going to be a problem to most ppl. Still is is unrealistic to expect perfection from any OS. It is something to be strived for, but does not and will never exist.
Any ways I think this sums it up nicely…
http://os.newsforge.com/comments.pl?sid=42861&cid=103724
FreeBSD 5.3 is KickAss. I’ve just migrated a server from SuSE 9.0 to FreeBSD 5.3-Release-p2 and I’ve had no problems. It’s running on a P4, Promise RAID , NVidia (AGP), and running MySQL, PostgreSQL, Java with Tomcat running Web-based applications, Apache with PHP. I’m about to setup Bacula to manage backups on some workstations.
Totally satisfied!!!
I have a dual Opteron colocation server which was thoroughly unstable with 5.3-RELEASE… I’d experience kernel panics, although I didn’t ever have dumpdev configured so I never managed to get a kernel core file to examine. After building a kernel from RELENG_5 sources, I continued to experience stability problems when using the 4BSD scheduler, possibly a resource exhaustion issue… the system would respond to pings but all services would go unresponsive under heavy load. I’ve backported the ULE scheduler along with kern_sig.c from HEAD and haven’t had stability problems since, but I’d agree that there are a number of bugs in RELENG_5_3, some of which have been fixed in RELENG_5 but others I’ve only been able to resolve using code from HEAD.
I agree about this being a troll article.
There is really no substance to prove its points in the article. ULE was never officially release, so how can it be a failure. Its development is taking place in the 6.x branch. What has been omitted from the article is the fact that several ULE improvments have been backported to the orginal bsd schedualar.
I have been used FreeSBIE 1.1 (based on FreeBSD 5.3) without trouble on my hardware. It seems to me that the author is mostly p*ssed that FreeBSD is not working on his systems. On the one hand, I agree that if in fact FreeBSD is severely unstable on some of the hardware it claims to support, then more testing should have been done before going to a RELEASE branch (perhaps wait till 5.4 to iron out the bugs.) On the other hand, the article just sounded like a long rant. FreeBSD seems to be perfectly stable on most configurations.
We’re running 5.3-STABLE with a number of jails, java, postgresql and everything runs fine and dandy. Using intel Pro/1000 nics
“I’ve backported the ULE scheduler along with kern_sig.c from HEAD and haven’t had stability problems since, but I’d agree that there are a number of bugs in RELENG_5_3, some of which have been fixed in RELENG_5 but others I’ve only been able to resolve using code from HEAD.”
Hello Bascule. Do you mean that you ported it back from head to 5.3? Or from head to 4.x? I believe that you are reference is in regards to the 5.x series.
Just curious.
Yes, I’m a FreeBSD nut. Yes, I like it a lot. No, FreeBSD 5.3 is not without it’s faults BUT that article is bad.
A FreeBSD STABLE branch is just that, API stable. Usually with API stability also follows system stability but that is not always the case. FreeBSD 5-STABLE is a very stable system. There are cases where it is a not so stable system and that is usually caused by either buggy hardware &/or drivers or a combination of other software that together with the OS make it unstable in some scenarios.
From the article:
FreeBSD 5.x enjoyed an excellent head start in the fully 64-bit AMD64 operating system arena, but now trails the pack, with only Windows XP 64-bit behind it in speed and completeness.
The link provided to show how “slow” AMD64 under FreeBSD 5.3 is supposed to be does in no way even mention FreeBSD and in fact only tests SuSE and Fedora Core along with Windows XP SP1. The link proveded about completeness goes along in that line and also does not mention FreeBSD. If one is to say something about the speed and the completeness then at least the links provided should be able to back that up. Just saying that FreeBSD is slow(er) without providing data or even just saying that you “feel” it’s slower is FUD. If I say Gnome 2.8 feels faster on my computer than Fedora Core then someone can either “trust” my feeling or say they disagree (based on their feelings or benchmarks).
The section “Improvements since 5.2.1” is also ignoring the more important improvments and even leaves out a few facts. X.org was not “upgraded” but rather replaced XFree86 in 5.3 (XFree was default X-server in 5.2.1). The new multithreaded network stack is mentioned elsewhere but not here and that is a major improvment in 5.3. Also the inclusion of pf in the system is a bigger “improvment” than the “upgrades” of Gnome and KDE which more or less follows their respective schedules. Preemption, improvements to 4BSD etc. Way to go in other words…
The ULE (which is not an acronym; its full name is SCHED_ULE as opposed to the older SCHED_4BSD) scheduler continues to have stability and performance problems and was totally disabled instead of being made the default process scheduler in 5.3 as planned.
No way José. ULE is the name of the scheduler. SCHED_ULE is the name of the configuration option you put in the kernel configuration file (default /usr/src/sys/i386/conf/GENERIC). So the name ULE is a play with “words” SCHED_4BSD vs SCHED_ULE. http://www.chesapeake.net/~jroberson/ULE.pdf
Also the problems with ULE and PREEMTION have recently been fixed and backported into 5-STABLE (but needs to be manually “enabled”).
Again from the article:
Where FreeBSD used to be a highly usable, reliable, and scalable operating system, the last three releases have been increasingly substandard, culminating in a hardly usable operating system on our test machines.
Assuming 5.0, 5.1, 5.2 (5.2.1 was a bugfix) was the releases then, please gimmie a break. Those releases was far more usable than the corresponding releases of Linux 2.5 which was also development/technological previews. Why not mention something about how changes in Linux 2.6 broke API:s and made updating without recompiling/replacing software impossible. Thats what 5-STABLE is about, being able to make updates to the system (say from 5.3 -> 5.4) without having to recompile ports.
I’ve had no problems with FreeBSD 5.3. In fact 5.3 are the best release yet for my laptop. Just remember, YMMV so use the system that suits you and your hardware/software best, be it Windows, Linux, *BSD etc.
I also noticed the link about speed did not even mention FreeBSD. His references are useless. Why does he link to sites that have no information about FreeBSD? I can only think that he is spreading FUD. This article is horrible! I don’t understand why someone would dishonor his own name this way.
I have been using 5.3 on my laptop as a desktop and a couple of servers for a while now and it handles every piece of hardware I throw at it flawlessly. Not seen a panic yet and the speed the system runs at is fantastic. Congrats to the FreeBSD team for this superb OS and a “wtf?” goes to Jem for clearly not doing any research.
Just saying that FreeBSD is slow(er) without providing data or even just saying that you “feel” it’s slower is FUD.
should’ve read:
Just saying that FreeBSD is slow(er), without providing data or even just saying that you “feel” it’s slower, is FUD/flamebait/trolling.
That makes my statement slightly clearer
Many people have been using FreeBSD 5.X in their servers for months without problems.
I think that saying that it’s not stable is misguided. 5.3 is a “.0” release even if it’s 5.3 just because it’s the first time it has hit the “stable” flag. A small minority of users with problems it’s expected, but it don’t means “don’t use it”.
I’d rather wait for the next release however. Just like I do with every other software
I hate to name call, and it does seem harsh and I apoligize. However I get the impression the author isn’t really a FreeBSD user.
“This is in stark contrast to GNU/Linux which, as a complete operating system, has no central, cohesive developer base and is packaged in myriad different ways by myriad different distribution projects and companies”
No Linux isn’t a complete OS, it is a kernel with a set of utilities.
“Where FreeBSD used to be a highly usable, reliable, and scalable operating system, the last three releases have been increasingly substandard, culminating in a hardly usable operating system on our test machines”
Hence why there weren’t stable releases and why FreeBSD said not to use them in production enviornment…duh..
Besides, anyone here ever try Fedora? I wouldn’t recommend it, least stable of all.
“FreeBSD 5.x enjoyed an excellent head start in the fully 64-bit AMD64 operating system arena, but now trails the pack, with only Windows XP 64-bit behind it in speed and completeness”
FreeBSD is always a little slower because the developers are making sure the code is consistent and stable. Linux moves in to many directions and has stability and organization issues. FreeBSD is the same where ever you go…clean code and eveything is in a orranized manner. You don’t have configuration files sitting here and there, you know where they are.
“Considering the long list of significant problems in FreeBSD 5.3-RELEASE”
What problems….did you list any? I have been running 5.3 for a little while (I adit i was a little nervous at first, but took the plunge) and haven’t had any problems at all. It is running my MySQL databases for the web back ends without a problem at all. My server went down once, and that was hardware failure not OS failure.
I am not some FreeBSD zealot who thinks every other OS sucks, I use Windows XP Pro at work/home. We have Windows/HP-UX/Sun/Linux boxes here at work and all have their good points. However maybe they should have had a FreeBSD user write that article, that way things could have been explained better. That’s my gripe….I’m done.
I agree with you Chris. But when you mention “wtf”, you are refering to the program “wtf” listed under games, right?
😉
Just having some fun and introducing a port.
just a silly question i am always readin from various places that people never or hardly ever have to reboot the system, what happens when you get a kernel update like in linux? or is it that kernel updates are provided rarely ?
Also can someone provide a noob like me, from the linux side of things a link to introduce to FreeBSD and difference i may face ?
last question, does pkgsrc compile everything from source ?
The kernel and userland go through lots of update. To get them you cvsup your userland. There are generally several changes or more a day to the src files for RELENG_5. Basically cvsup see if it would be any thing useful. If so update your test box. If they make a difference prepare to roll them out to your servers.
So the same as nearly any thing else which you would track using cvsup.
Can’t say I agree with the majority here on how great fbsd is. Just the other day I tried 5.3, after having tried previous 4.x and 5.x releases, hoping this time it would be better. Nope, at least not on my hardware.
From what I understood, 5.3’s emu10k1 driver is supposed to finally have audigy support, cool. However, when I do a kldload of the driver, the whole system locks up hard, cold reboot is the only way I can get out. Reason is, so far as I can tell, is that my audigy uses IRQ 5. So does my PCI modem, which by itself, works fine. Have them both access the same IRQ, bang, system is dead. I’d experienced this before, in other fbsd releases, but usually using some third party driver like the chibis one. This was with a stock install.
That’s pretty bad, I don’t see this problem with the other BSDs, Linux, shoot even Windows. Sorry to say, I think after this I’ll have finally given up on fbsd.
Check your bios settings and them reconfig your kernel to make or vice versa.
Also some drivers like to be loaded at boot time too.
If you would like to see something done about it submit a PR and begin checking the mailing lists.
I doubt it is IRQ sharing cuasing the problem, since that works here fine.
I love FreeBSD. I have since the first time I installed it on a system.
But when I recently tried installing 5.3 on my new AMD64 Desktop, I had much frustration. The network interface would crash with a “watchdog timeout”. I tried swapping out several different network cards, but all of them resulted with the same error which caused the network to go down.
So after trying several different kernel configurations/rebuild, many of which would crash on boot, and trying both the i386 and amd64 distributions, I just gave up.
“Also the problems with ULE and PREEMTION have recently been fixed and backported into 5-STABLE”
I know this was discussed on the -current mailing list but has it actually been done yet? Looking at the dates in sched_ule.c, kern_sig.c, and kern_switch.c, they all seem to be from october while those in head are from this month. Based on that it doesn’t look like the changes have been merged yet? Unless I am missing something.
You may be experiencing instability on your system. AMD64 is a fairly new port. It is not really surprising that there are bugs. But the article is still horrible. Nothing is backed up. The author is simply trolling, so I find it is going a bit far as to say that the article is right.
<p>But the article is still horrible. Nothing is backed up. The author is simply trolling, so I find it is going a bit far as to say that the article is right.</p>
<p>What article did you read? The linked review of FreeBSD 5.3 backs up every claim with a valid link or the reviewer’s own testing and experience. I too had to downgrade back to 5.2.1 because of networking problems with 5.3, so I find it both honest and accurate. On the other hand, I don’t have an agenda to push.</p>
<p>You FreeBSD zealots need to face the fact that 5.3 is full of problems rather than attacking the people telling you about them.</p>
Check your bios.
BIOS can result in this. Mainly the APIC portion.
Becuase for the most part no one here or in the article is actually mentioning any problems in any useful fashion. Justing ranting on about a problem them and portraying them as if every body is using them or as if the system is entirely broken becuase of it.
FreeBSD is always a little slower because the developers are making sure the code is consistent and stable. Linux moves in to many directions and has stability and organization issues. FreeBSD is the same where ever you go…clean code and eveything is in a orranized manner. You don’t have configuration files sitting here and there, you know where they are.
You’re bloody joking, right? Stop spouting this nonsense about Linux “moving in many directions and having stability and organiszation issues”. What “directions” exactly, are you talking about? And what “issues”? Eh?
And yes, FreeBSD is slower than Linux. That is nothing to do with clean code, and everything to do with something terribly rotten in the code base. Have a look at this:
http://marc.theaimsgroup.com/?l=freebsd-current&m=110367308611797&w…
Linux 2.6 is shown to be from 2 to literally nearly 30 times faster than FreeBSD 5.3 in a threaded bind9 throughput test.
Ahh, fun, a useless linux zealot…
Well first off, linux is a massive organization problem, but since this thread is not about those, there is no point in going into it. It is still not non-sense just becuase you don’t like it.
Is FreeBSD RELENG_5_3 slower than linux in some cases yes, but so what. Since are spouting this, you most likely don’t understand it all, so here it is… RELENG_5_3 is the newest stable branch. What does this mean? It means it is stable enought to roll out but there is still optimization and tweaking to be done.
It would also be interesting to see them redone, since there have been some pleasant improvements since beta7.
Any ways, so bloody what? For most things there is no performance differnce.
BTW if you are going to try to make points using mailing list posts, it is generally best not to use a lame search engine like that. One that obscures the email address is redicolous and annoying.
Well first off, linux is a massive organization problem, but since this thread is not about those, there is no point in going into it. It is still not non-sense just becuase you don’t like it.
Oh how convenient for you. Unfortunately I don’t live in a fairy land where someone’s conjecture becomes reality.
Is FreeBSD RELENG_5_3 slower than linux in some cases yes, but so what. Since are spouting this, you most likely don’t understand it all, so here it is… RELENG_5_3 is the newest stable branch. What does this mean? It means it is stable enought to roll out but there is still optimization and tweaking to be done.
Sorry, 20 times worse threading performance in a real world application isn’t just a matter of “tweaking”. There is obviously a fundamental problem somewhere, and it is crazy that it hadn’t been picked up earlier – it gives you a good indication about the sort of QA the FreeBSD developers do.
It would also be interesting to see them redone, since there have been some pleasant improvements since beta7.
He tested with a release candidate too.
Any ways, so bloody what? For most things there is no performance differnce.
So what? You tell me: you’re the one who got your knickers in a twist about it.
BTW if you are going to try to make points using mailing list posts, it is generally best not to use a lame search engine like that. One that obscures the email address is redicolous and annoying
I have a better idea. I’ll continue to use whatever archive I want, and you deal with it.
Oh how convenient for you. Unfortunately I don’t live in a fairy land where someone’s conjecture becomes reality.
The recent interview where Linus dizzes Solaris should show an obvious problem with Linux development, planning or more precise the lack thereof. I really can’t plan my way out of a cardboard box. All my long-term stuff is very fuzzy “intuitive” stuff, not something I could really put into words. I try to avoid having very specific goals in the long term, and instead have more of a general feel for what kinds of things I like and don’t like. Some people may see that as undirected, and hell yes, it is. (From: http://news.com.com/Torvalds+a+Solaris+skeptic/2008-1082_3-5498799…. )
To some, this is a major problem and makes Linux a harder target to write software to. Especially if you need a feature X that is not present in the kernel.
Sorry, 20 times worse threading performance in a real world application isn’t just a matter of “tweaking”. There is obviously a fundamental problem somewhere, and it is crazy that it hadn’t been picked up earlier – it gives you a good indication about the sort of QA the FreeBSD developers do.
No if you would’ve read your link then you would also see that a fairly simple kernel hack solved that. From your link: I’ve made a quick-hack patch in the FreeBSD kernel and confirmed that this is the case.
Interrestingly, FreeBSD 5.3 RC1 was faster than Linux with no threads and standrad BIND.
Also, if you have followed the development of FreeBSD 5.3 then you also know that quite a few bugs (many performance hindering) was squashed going from 5.3RC1 to 5.3 RELEASE and that the BIND test have not been run on anything later than 5.3 RC1 (subsequent posts on [email protected]). It has not been tested on either 5-STABLE or 6-CURRENT. It was also tested using ULE at a time when it was known to be broken (because of PREEMTION) so it was not surprising to find that 4BSD was better.
In subsequent posts there is also a comment about Linux possibly cutting corners but I would not put any weight in that and I am the wrong person to judge such a thing.
Next time you troll, please try to do better. Linux is a great system so there is no need to be hostile or even making facts up as Linux have enough merits on its own.
“BTW if you are going to try to make points using mailing list posts, it is generally best not to use a lame search engine like that. One that obscures the email address is redicolous and annoying
I have a better idea. I’ll continue to use whatever archive I want, and you deal with it. ”
No, if you ever wish to make point that are any thing more than a lame troll, you need to.
“Sorry, 20 times worse threading performance in a real world application isn’t just a matter of “tweaking”. There is obviously a fundamental problem somewhere, and it is crazy that it hadn’t been picked up earlier – it gives you a good indication about the sort of QA the FreeBSD developers do.”
Ever consider it is something that has not been seen before? I and many other ppl have not seen it.
“He tested with a release candidate too.”
So bloody what? Same damn thing. A lot has changed since then.
“So what? You tell me: you’re the one who got your knickers in a twist about it.”
Nah, for me, it is your annoyingness. ^_^
The recent interview where Linus dizzes Solaris should show an obvious problem with Linux development, planning or more precise the lack thereof.
Err no, if *you* had any idea about Linux development, then no.
Linus does not plan it. Get your head around that, he is not the dictator who coordinates all development.
He puts things together at the end, and generally provides some guidance… but there are definitely people planning and designing and building things before they go to Linus.
No if you would’ve read your link then you would also see that a fairly simple kernel hack solved that. From your link: I’ve made a quick-hack patch in the FreeBSD kernel and confirmed that this is the case.
He didn’t even post the results without that hack. And with the “hack” FreeBSD was still much slower.
Interrestingly, FreeBSD 5.3 RC1 was faster than Linux with no threads and standrad BIND.
Err, no, that’s not interesting. More like “in the noise”. Statistically insignificant. Even if it was statistically valid, it is less than a percent.
In subsequent posts there is also a comment about Linux possibly cutting corners but I would not put any weight in that and I am the wrong person to judge such a thing.
Yeah, that was just a clueless FreeBSD developer doing some fine trolling and trying to cover his own ass because his grand release is so bad.
Next time you troll, please try to do better. Linux is a great system so there is no need to be hostile or even making facts up as Linux have enough merits on its own.
Err, what fact did I make up? Troll.
And yes, Linux does have enough merits on its own. One of those merits is that it is fast, has great SMP scalability, and a threading system that works.
No, if you ever wish to make point that are any thing more than a lame troll, you need to.
Starting on personal attacks now. Nice.
So you base the merits of someone’s post based on what mailing list archive they link to? That’s great for you, you small, sad, petty little man.
In the meantime, I’ll continue to use whatever mailing list archive I want. And no amount of your pathetic whining will make any difference.
Ever consider it is something that has not been seen before? I and many other ppl have not seen it.
No, I would be not at all surprised if none of the FreeBSD developers had tested anything like such a standard workload as a DNS server (BIND too, _the_ DNS server). Hence my comment about QA.
“He tested with a release candidate too.”
So bloody what? Same damn thing. A lot has changed since then.
Yeah, because the release candidates were so broken. But I think we’ll find 5.3 is no better.
“So what? You tell me: you’re the one who got your knickers in a twist about it.”
Nah, for me, it is your annoyingness. ^_^
Oooh pooooor baby. Are you annoyed because I pointed out FreeBSD got beaten by Linux? Oh you poor thing.
>>The recent interview where Linus dizzes Solaris
>>should show an obvious problem with Linux development,
>>planning or more precise the lack thereof.
Err no, if *you* had any idea about Linux development, then no.
Linus does not plan it. Get your head around that, he is not the dictator who coordinates all development.
I never said he was. BUT Linus is ultimately the one who decides what goes and what stays in the official kernel releases.
He puts things together at the end, and generally provides some guidance… but there are definitely people planning and designing and building things before they go to Linus.
And you don’t see this as a problem?! Let’s make next version of a good car and not have a plan of what the goals are. True, OpenBSD and recently FreeBSD have switched towards timebased releases but that does not stop them from having set goals and people that manage those goals. What if Linus doesn’t like the plans/designs and the stuff they built? Fork… yeah suuure.
Remember what happened during the 2.4 release cycle with the VM? Remember what recently happend when 2.6.8 was released? Not a problem you say. I beg to differ. I want to target a stable platform, not a moving target with no clear goal. To me and a lot of developers it does matter, not to everyone and I stated that in my previous post.
He didn’t even post the results without that hack. And with the “hack” FreeBSD was still much slower.
Did you even read your own link?! The results published are for standard RC1 and Beta 7 except in one test using his improved version of BIND (which he calls BIND++).
From your link:
B. tests with FreeBSD 5.3 RC1 on Xeon 3000MHz x 4
threads BIND BIND++ BIND++-kernel_patch
0 16253
1 7953 14600 14438
2 3591 19840 23854
3 1012 24268 30268
4 533 25447 30434
See, the quick kernel hack improves the threading a bit but without it is not a slouch either.
>>Interrestingly, FreeBSD 5.3 RC1 was faster than Linux >>with no threads and standrad BIND.
Err, no, that’s not interesting. More like “in the noise”. Statistically insignificant. Even if it was statistically valid, it is less than a percent.
A lot of apps are NOT threaded and a lot is badly threaded ex. BIND 9.3.0 as shown by your link, so apps running without threading is running fast on FreeBSD (in this case faster than Linux 2.6).
I run a lot of bio-software that does not utilise threads and pushes the CPU to the max for days or weeks in a run so this performance matters. I probably should take my tests of different systems and write an article about doing biological-computing.
Yeah, that was just a clueless FreeBSD developer doing some fine trolling and trying to cover his own ass because his grand release is so bad.
Yeah right, Scott Long clueless… You just proved your ignorance. While I do not like GPL or Microsoft, I would not call Bill Gates or RMS clueless, because that would be ignorant, just as you calling Scott Long is ignorant.
You have not only managed to show that you cannot read your own links properly but also showed an inability to put two and two together.
Fact, FreeBSD threading works very nicely (stated in your link)
Fact, FreeBSD 5.3-RELEASE was NOT tested in your link so conclusions cannot be extended to 5.3-RELEASE
Fact, BIND 9.3.0 have issues with threading on BOTH SuSE and FreeBSD
>>Next time you troll, please try to do better.
>>Linux is a great system so there is no need to be
>>hostile or even making facts up as Linux have
>>enough merits on its own
Err, what fact did I make up? Troll.
I just showed you.
I never said he was. BUT Linus is ultimately the one who decides what goes and what stays in the official kernel releases.
No, this is just not true. Andrew Morton has a lot of weight in various different areas in the kernel (memory manager, vfs, filesystems in particular), David Miller basically decides what goes and what stays in the network stack, Jeff Garzik network drivers, James Bottomley SCSI layer, Jens Axboe block layer, Ingo Molnar scheduler, etc.
These are all people who can (and do) override Linus in their respective fields.
What if Linus doesn’t like the plans/designs and the stuff they built? Fork… yeah suuure.
If Linus alone doesn’t like it, and everyone else does, then he’ll either have to come up with something better himself, or he will concede the point. Linux development has changed a lot.
Remember what happened during the 2.4 release cycle with the VM? Remember what recently happend when 2.6.8 was released? Not a problem you say. I beg to differ. I want to target a stable platform, not a moving target with no clear goal. To me and a lot of developers it does matter, not to everyone and I stated that in my previous post.
FreeBSD 4.9 (or was it 4.10? No 4.9 I think) introduced bad SMP instability and filesystem corruption with softupdates.
As far as the 2.4 VM goes, the problem was with large “highmem” (i386 PAE) systems, of 16GB and more. I doubt FreeBSD 5.3, 6, or any of the 4 series would even *boot* one of those systems.
Linux VM developers simply didn’t have access to these type of machines, so it is no failing that the VM didn’t always work well on them.
See, the quick kernel hack improves the threading a bit but without it is not a slouch either.
How about you read the link. BIND++ lacks critical locking and so cannot be considered a production system. Nor is that interesting in a test of real world threaded apps.
A lot of apps are NOT threaded and a lot is badly threaded ex. BIND 9.3.0 as shown by your link, so apps running without threading is running fast on FreeBSD (in this case faster than Linux 2.6).
I told you already, that isn’t statistically significant.
But excuse me, BIND9 looks like it is threaded quite well indeed. It scales nicely on a real operating system. No, it looks like FreeBSD “is badly threaded” to me.
Yeah right, Scott Long clueless… You just proved your ignorance. While I do not like GPL or Microsoft, I would not call Bill Gates or RMS clueless, because that would be ignorant, just as you calling Scott Long is ignorant.
He is clueless when it comes to Linux as evidenced by his weak attempt to salvage some pride after his operating system was shown to be soundly beaten.
Fact, FreeBSD threading works very nicely (stated in your link)
Fact, FreeBSD 5.3-RELEASE was NOT tested in your link so conclusions cannot be extended to 5.3-RELEASE
Fact, BIND 9.3.0 have issues with threading on BOTH SuSE and FreeBSD
Fact, BIND9 on Linux Linux trounces FreeBSD when it comes to threaded operation and multiprocessor scalability.