Re: buildworld times

From: Vincent Poy (vince_at_oahu.WURLDLINK.NET)
Date: 02/29/04

  • Next message: Vincent Poy: "Re: HEADSUP: Sleep queues added to kernel, so be careful." 
    Date: Sat, 28 Feb 2004 16:41:38 -1000 (HST)
To: Robert Watson <rwatson@freebsd.org>

    On Sat, 28 Feb 2004, Robert Watson wrote:

    >
    > On Sat, 28 Feb 2004, Vincent Poy wrote:
    >
    > > On Sat, 28 Feb 2004, Robert Watson wrote:
    > >
    > > > On Sat, 28 Feb 2004, Vincent Poy wrote:
    > > >
    > > > > > 2.5 hours seems a little high and would seem to indicate that you're
    > > > > > building world under a kernel that has all the debugging options enabled.
    > > > > > Could you make your kernel config available?
    > > > >
    > > > > Yeah, I recall that -CURRENT on a Mobile PII 366Mhz took the
    > > > > same amount of time so I would imagine the P4 would be faster since I
    > > > > remember PIII running at 800Mhz or so would take about 2 hours and the
    > > > > Athlon K7 at 1.2Ghz was about a little over an hour. I'm using the
    > > > > standard GENERIC kernel and do have the debugging options enabled.
    > > >
    > > > Turning off the debugging options (especially WITNESS) can have a dramatic
    > > > effect on performance. Give it a spin. :-)
    > >
    > > Besides the WITNESS and WITNESS_SKIPSPIN, do these have a big
    > > effect as well?
    > >
    > > options DDB
    >
    > No runtime cost, and useful if anything goes wrong :-).
    >
    > > options INVARIANTS
    >
    > Small but non-zero cost.
    >
    > > options INVARIANT_SUPPORT
    >
    > Space overhead, but other than memory overhead, no run-time cost.
    >
    > > makeoptions DEBUG=-g
    >
    > Makes your kernel build real slow, but won't change the kernel speed.
    >
    > As you point out, debugging problems is a lot easier if you run with
    > INVARIANTS+WITNESS. However, you may well see build times halve or more
    > by running without WITNESS.

            I always thought INVARIANTS caused slow performance as well.
    Ofcourse, it does seem all the stuff mentioned above is disabled in a
    -RELEASE GENERIC kernel so that may be it as well.

    Cheers,
    Vince - vince@WURLDLINK.NET - Vice President ________ __ ____
    Unix Networking Operations - FreeBSD-Real Unix for Free / / / / | / |[__ ]
    WurldLink Corporation / / / / | / | __] ]
    San Francisco - Honolulu - Hong Kong / / / / / |/ / | __] ]
    HongKong Stars/Gravis UltraSound Mailing Lists Admin /_/_/_/_/|___/|_|[____]
    Almighty1@IRC - oahu.DAL.NET Hawaii's DALnet IRC Network Server Admin

    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

  • Next message: Vincent Poy: "Re: HEADSUP: Sleep queues added to kernel, so be careful."