Re: [PATCH] scheduler patch, faster still
Richard Gooch (rgooch@atnf.csiro.au)
Tue, 29 Sep 1998 16:27:10 +1000
yodaiken@chelm.cs.nmt.edu writes:
> On Mon, Sep 28, 1998 at 08:26:41PM +0200, Rik van Riel wrote:
> > Come on, we've all seen the code. Scanning the longer run
> > queue is the _only_ thing that's done when you have a
> > long run queue. If you've read the code, you'll reach
> > the same conclusion I reached...
>
> Don't. Maybe this tiny difference is caused by cache misses.
On a 386???? 12.3 us switch time and add 7.4 us for every extra
process on the run queue. Intel 386DX33.
This shows you can't explain away all the consistent trends I've
measured (longer run queue has a significent effect) on cache misses.
Regards,
Richard....
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/