Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 17 Oct 2003 05:46:29 -0400 (EDT)
From:      Jeff Roberson <jroberson@chesapeake.net>
To:        Sean Chittenden <sean@chittenden.org>
Cc:        current@freebsd.org
Subject:   Re: More ULE bugs fixed.
Message-ID:  <20031017054543.E30029-100000@mail.chesapeake.net>
In-Reply-To: <20031017092434.GA45975@perrin.nxad.com>

next in thread | previous in thread | raw e-mail | index | archive | help

On Fri, 17 Oct 2003, Sean Chittenden wrote:

> > I think you cvsup'd at a bad time.  I fixed a bug that would have
> > caused the system to lock up in this case late last night.  On my
> > system it freezes for a few seconds and then returns.  I can stop
> > that by turning down the interactivity threshold.
>
> Hrm, I must concur that while ULE seems a tad snappier on the
> responsiveness end, it seems to be lacking in terms of real world
> performance compared to 4BSD.

Thanks for the stats.  Is this on SMP or UP?

>
> Fresh CVSup (~midnight 2003-10-17) and build with a benchmark from
> before and after.  I was "benchmarking" a chump calc program using
> bison vs. lemon earlier today under 4BSD
> (http://groups.yahoo.com/group/sqlite/message/5506) and figured I'd
> throw my hat in on the subject with some relative numbers.  System
> time is down for ULE, but user and real are up.
>
>
> Under ULE:
>
> Running a dry run with bison calc...done.
> Running 1st run with bison calc... 52.11 real 45.63 user 0.56 sys
> Running 2nd run with bison calc... 52.16 real 45.52 user 0.69 sys
> Running 3rd run with bison calc... 51.80 real 45.32 user 0.87 sys
>
> Running a dry run with lemon calc...done.
> Running 1st run with lemon calc... 129.69 real 117.91 user 1.10 sys
> Running 2nd run with lemon calc... 130.26 real 117.88 user 1.13 sys
> Running 3rd run with lemon calc... 130.76 real 117.90 user 1.10 sys
>
> Time spent in user mode   (CPU seconds) : 654.049s
> Time spent in kernel mode (CPU seconds) : 7.047s
> Total time                              : 12:19.06s
> CPU utilization (percentage)            : 89.4%
> Times the process was swapped           : 0
> Times of major page faults              : 34
> Times of minor page faults              : 2361
>
>
> And under 4BSD:
>
>  Running a dry run with bison calc...done.
>  Running 1st run with bison calc... 44.22 real 37.94 user 0.85 sys
>  Running 2nd run with bison calc... 46.21 real 37.98 user 0.85 sys
>  Running 3rd run with bison calc... 45.32 real 38.13 user 0.67 sys
>
>  Running a dry run with lemon calc...done.
>  Running 1st run with lemon calc... 116.53 real 100.10 user 1.13 sys
>  Running 2nd run with lemon calc... 112.61 real 100.35 user 0.86 sys
>  Running 3rd run with lemon calc... 114.16 real 100.19 user 1.04 sys
>
>  Time spent in user mode (CPU seconds) : 553.392s
>  Time spent in kernel mode (CPU seconds) : 6.978s
>  Total time : 10:40.80s
>  CPU utilization (percentage) : 87.4%
>  Times the process was swapped : 223
>  Times of major page faults : 50
>  Times of minor page faults : 2750
>
>
> Just a heads up, it does indeed look as thought hings have gone
> backwards in terms of performance.  -sc
>
> --
> Sean Chittenden
>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031017054543.E30029-100000>