[NTLUG:Discuss] [Fwd: Linux vs Solaris stats]
Alton Pouncey
arpounce at earthlink.net
Mon Aug 9 23:47:07 CDT 1999
Not to sound like a Sun apologist or anything but why wasn't a benchmark done on an
UltraSparc machine as well? Solaris isn't exactly optimized (no matter how many
times you recompile the kernel) for the Intel architecture. I would have liked to
see how UltraLinux would have fared against Solaris on the sun4u architecture.
To me, it is obvious that Linux will beat most any OS (maybe all??) on a PC.
Alton Pouncey
"J. Reeves Hall" wrote:
> Georgia Tech did some tests on Linux and Solaris... Take a look at the
> numbers. They're either amazing, or pathetic, depending on which
> OS you're supporting :)
>
> -Reeves
>
> ------------------------------------------------------------------------
>
> Subject: Re: Linux vs Solaris stats
> Date: Sat, 7 Aug 1999 11:01:46 -0400 (EDT)
> From: vernard at cc.gatech.edu (Vernard Martin)
> To: reeves at earthling.net (J. Reeves Hall)
>
> > I took a look at
> > http://www.cc.gatech.edu/projects/ihpcl/lmbench-results/lmbench.html...
> > It would appear that Linux ate Solaris for lunch. Is Linux really that
> > much more efficient than Solaris, or am I looking at the numbers
> > crosseyed?
>
> Sorry it took so long to respond. I had a death in the family.o
> Yes, those comparisons are correct. Except in exactly one case, linux beats the
> absolute stuffing out of solaris hands down. Now this was Solaris 2.5. Solaris
> 2.7 is out and I don't believe that we have run the benchmarks on it yet.
>
> > Any idea how NT stacks up?
>
> Afraid not. We don't have the POSIX package for NT so lmbench won't run on it.
> My gut instinct is that it'll hold its own for most of the tests, win a few
> handily and loose a few very badly.
>
> Vernard
> --------------------------------------------------------------------------------
> Darkness has a hunger that's insatiable Vernard Martin (vernard at cc.gatech.edu)
> Lightness has a call that's hard to hear http://www.cc.gatech.edu/~vernard/
More information about the Discuss
mailing list