"The FreeBSD operating system is finally through it's buggy 5.x series and into the more reliable 6.x series. Most of the problems of the old days - kernel panics on multi-CPU machines, AMD64 troubles galore, and shaky network drivers - are gone. FreeBSD still isn't perfect, but at least with 6.0-RELEASE it's more stable and functional than it has been in the recent past."
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:"Bring on the benchmarks. If this is true, it should be in FreeBSD's interest to get that out there."
No need benchmark it, just install FreeBSD-6.0 on your dual-boot Linux and see for your self the difference.
"How is it possible to put forth such a claim when THE PROOF IS ALREADY IN THE PUDDING?? And why aren't they publishing any benchmarks that demonstrates that they are even in the ballpark of Linux' SMP performance?"
With the enhanced ULE scheduler, FreeBSD's SMP is much better than Linux, squidclam is hogging to death on Gentoo Linux while FreeBSD-6.0 with ULE sched has no effort at all
"Remember, because of the SMP work that went into 5.x, 4.x performed significantly better on the common case, UP machines. That kind of regression would NEVER be accepted in the Linux kernel."
I really accept that FreeBSD-4.x is fast, and IMHO, FreeBSD-5.x is the developmental version which put FreeBSD-6.0 on its scalability and performance.
BTW, if linux is _STABLE_ why they're putting "EXPERIMENTAL" stuff on it? Does it mean that "EXPERIMENTAL" is "STABLE"?
RE[2]: Some more material on FreeBSD top performance
RE[3]: Some more material on FreeBSD top performance
Member since:
2005-08-28
"There are reports of FreeBSD starting to benchmark much faster than Linux for certain filesystem tasks on SMP systems."
Bring on the benchmarks. If this is true, it should be in FreeBSD's interest to get that out there.
"BSDs in general are also much more corporate friendly than Linux
This is so funny. How about taking a look at what actually takes place in the real world? Both in terms of contributions to the kernel and actual deployment. Almost all the people who work on the Linux kernel these days have that as part of their job description, and the Linux eco system has actually SPAWNED a number of corporations (Red Hat alone has almost 1k employees! not bad for a corporation relying entirely on GPL code), in addition to attracting behemoths like IBM and Intel.
"Linux took an approach that was fairly easy and non-disruptive, but ultimately has limitations that we feel will limit its scalability for general computing,"
How is it possible to put forth such a claim when THE PROOF IS ALREADY IN THE PUDDING?? And why aren't they publishing any benchmarks that demonstrates that they are even in the ballpark of Linux' SMP performance?
Btw, immediately when I saw that quote, I KNEW it came from Scott Long. That guy just makes stuff up.
Also, Linux' approach was non-disruptive because of the Linux kernel development philosophy. Basically, most work should and is done in small incremental steps and each step needs to be a step in the right direction. This makes it much easier for people to review the work, and it also makes it easier for the people doing to work to keep up with current kernels. The Linux kernel people have demonstrated again and again that most work can be completed in this fashion.
In FreeBSD, they prefer to do major work in the perifery on private Perforce repositories and then pull things into the main cvs repository whole hog. No wonder they experience disruptions. It also makes it alot harder for people to do a proper review of all the work.
Remember, because of the SMP work that went into 5.x, 4.x performed significantly better on the common case, UP machines. That kind of regression would NEVER be accepted in the Linux kernel.