Re: Comments on the KSE option

Daniel Eischen wrote:
On Sat, 28 Oct 2006, Scott Long wrote:

Daniel Eischen wrote:
Actually, that's not quite true. I assume you know the thing you
left out: system scope threads compete against all the other
system scope threads in the system (from all applications, not
just within one application).


All this debate about the merits of process scope threads and fair
scheduling is great. But tell me, who was working on making this stuff
work well quickly and reliably (i.e. work well)? No one! I don't care
what AIX or Solaris or what else may or may not have done, who was making this work well for FreeBSD? Having a slow a thread subsystem is
a serious detriment, no matter how nice and flexible it looks on paper.

Process scope threads work well in libpthread. System scope
threads work well and fast in libthr. I think most people's
concept of "fast" as applied to process scope threads doesn't
quite mesh well with the fact that process scheduling is fair.


Btw, why is PTHREADS_INVARIANTS still enabled on -STABLE branches? We
don't have kernel INVARIANTS enabled there, so I don't understand why
libpthread needs to be different.

Scott

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

Relevant Pages

  • Re: Comments on the KSE option
    ... system scope threads in the system (from all applications, ... concept of "fast" as applied to process scope threads doesn't ... quite mesh well with the fact that process scheduling is fair. ...
    (freebsd-current)
  • Re: BIND9 performance issues with SMP
    ... >> contention is reasonably cheap. ... I observed this with system scope threads. ... One remarkable difference is that ULE *relatively* improved the ...
    (freebsd-current)
  • Re: Comments on the KSE option
    ... said that you can't have system scope threads. ... same priority as process scope threads. ... are we going to make system scope thread model fair ...
    (freebsd-current)
  • Re: Comments on the KSE option
    ... If your application requires 1000 threads of execution, you can write it three ways, with 1000 processes, 1000 system scope threads or 1000 process scope threads. ... This whole "fairness" argument is about making system scope threads have the same priority as process scope threads. ... are we going to make system scope thread model fair compared to process scope threaded model, or fair compared to the separate processes model? ...
    (freebsd-current)
  • Re: FreeBSD5.3-RC1 MySQL Performance
    ... >> the MySQL backend was MyISAM. ... > a serious problem with KSE process scope threads. ... David made a recent change that speeded up system scope threads. ... We're not sure why process scope threads don't do as well. ...
    (freebsd-current)