Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 3 Dec 1996 14:23:21 -0500 (EST)
From:      John Dyson <dyson@dyson.iquest.net>
To:        lm@neteng.engr.sgi.com (Larry McVoy)
Cc:        dyson@freebsd.org, lm@neteng.engr.sgi.com, thorpej@nas.nasa.gov, davem@jenolan.rutgers.edu, jkh@time.cdrom.com, dennis@etinc.com, kpneal@pobox.com, hackers@freebsd.org, torvalds@cs.helsinki.fi, lm@relay.engr.SGI.COM, iain@sbs.de, sparclinux@vger.rutgers.edu
Subject:   Re: TCP/IP bandwidth bragging
Message-ID:  <199612031923.OAA01823@dyson.iquest.net>
In-Reply-To: <199612031858.KAA00415@neteng.engr.sgi.com> from "Larry McVoy" at Dec 3, 96 10:58:06 am

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> : > FreeBSD folks, please don't beat up David for optimizing for the I/O
> : > paths tested by lmbench.  While I agree with the load vs no load points
> : > raised, you are missing another one: smallness is goodness, and David
> : > is almost always optimizing by making things smaller.  There are plenty
> : > of people shoveling stuff into the kernel making it slower - David is
> : > making it smaller & faster, let him be, it's useful.
> : >
> : BTW, that is our emphasis also (in the VM code), and we are really
> : fast.  We are interested in friendly competition, but not interested
> : in bragging.  Some people take offense to bragging.
> 
> Hey, John, take the high road.
>
I have been.  Now the gloves are off :-).

>
>  You've heard the old saying: numbers
> talk, bullshit walks.
>
> ...Various other pieces of drivel from a drooler...
>    (this is only meant as a touche for the above insult)
>
Look Larry, I am trying to say that I don't want to get into the benchmark
waring game that you and your other Linux friends apparently want to play.
I have stated that your lmbench benchmarks DO NOT SHOW APPLICATION
PERFORMANCE...  I really don't care other than for QC about lmbench perf
numbers.  They are valuable to me, but the results measured by lmbench are
pretty much orthogonal to what the end user needs to measure for performance
comparisons.

I like and use lmbench as one of the tools (and only a small part of the
tools) that I use to make sure that FreeBSD performs well...  Lmbench is
just overblown by the uninitiated and people new to computing, so was
dhrystone for that matter.  I know that you believe in your invention,
and I do also, as far as it's bona fide scope takes it.

Sorry charlie (I mean Larry), them's the breaks...  For application
benchmarking, lmbench is the one doing the walking (per your statement
above)...

John

care about spending lots of time optimizing from 



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612031923.OAA01823>