Performance of 4.x vs 5.x (Re: Lifetime of FreeBSD branches)
kris at obsecurity.org
Mon May 23 14:35:23 PDT 2005
On Mon, May 23, 2005 at 02:31:55PM -0700, Kris Kennaway wrote:
> On Mon, May 23, 2005 at 11:21:13PM +0200, Matthias Buelow wrote:
> > Kris Kennaway wrote:
> > > One thing that probably confuses and misleads a lot of people is when
> > > they build world or a kernel and notice that it's taking much longer
> > > than it did under 4.x, so they assume this means that 5.x is slower
> > > than 4.x. It doesn't. What it means is that 5.x and 4.x have
> > > different C compilers, and gcc 3.x is much slower at compiling code
> > > than gcc 2.x. You have to be very careful to draw conclusions based
> > > on subjective assessments like this.
> > Another thing might be that interactive response time seems to be worse.
> > While I (or rather ports) unpack the firefox/thunderbird source, the
> > machine is pretty much bogged down (mouse cursor jumps around, audio
> > stutters...). Haven't seen that on FreeBSD since the 386 days.
> I don't run FreeBSD on my desktop machines so I haven't seen this
> myself. One obvious guess is that it's due to VFS being under Giant,
> which causes lots of contention with other subsystems that also
> require Giant, and therefore introduces latency. If so, you'd see a
> substantial performance improvement on 6.0 with debug.mpsafevfs=1.
> This option isn't yet ready for production use (especially on SMP
> machines) since it still contains bugs, but it would be interesting if
> someone who sees this problem could test it on 6.0.
Also try defining PREEMPTION in your kernel on 5.x and above (if you
are running i386 or amd64). There have been very occasional reports
of panics with this option enabled (although I use it everywhere and
have not seen problems on my heavily loaded machines), but interactive
response should be much better.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20050523/33d55283/attachment.bin
More information about the freebsd-stable