Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 25 Apr 1995 18:23:41 -0700 (MST)
From:      Don Yuniskis <dgy@seagull.rtd.com>
To:        freebsd-hackers@freefall.cdrom.com (FreeBSD hackers)
Subject:   Re: benchmark hell.. (Really context switch & FPU)
Message-ID:  <199504260123.SAA20990@seagull.rtd.com>
In-Reply-To: <199504260012.RAA09198@ref.tfs.com> from "Poul-Henning Kamp" at Apr 25, 95 05:12:30 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> Well, you guys, when can we see the code ??
> 
> > I didn't explain.  Sorry!  I assumed we would want to do
> > lazy FP context save/restore.  Once you do that, further
> > refinements don't buy you very much.  If you reread what I
> > wrote _in this context_, may be it will make more sense.
> > 
> > > o       The FPU has a "last used by" state variable
> > > o       When the first FPU using process is run, this variable is
> > >         set to its PID.
> > Okay.

I've implemented this in a couple of real-time OS's and it really is
a big win -- especially if the FPU can execute in parallel with the 
CPU's execution unit (some FPU's can't) and you invest the time hooking
the FPU traps to do belated switches *only on demand*.  It does have
some impact on real-time determinism ... but I doubt that would
be a real issue in FBSD.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504260123.SAA20990>