From owner-freebsd-chat  Wed Dec  3 20:09:53 1997
Return-Path: <owner-freebsd-chat@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id UAA15863
          for chat-outgoing; Wed, 3 Dec 1997 20:09:53 -0800 (PST)
          (envelope-from owner-freebsd-chat@FreeBSD.ORG)
Received: from zeus.gel.usherb.ca (zeus.gel.usherb.ca [132.210.70.7])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id UAA15858
          for <chat@freebsd.org>; Wed, 3 Dec 1997 20:09:47 -0800 (PST)
          (envelope-from alex.boisvert@gel.usherb.ca)
Received: (from boia01@localhost)
	by zeus.gel.usherb.ca (8.8.8/8.8.8) id XAA21077;
	Wed, 3 Dec 1997 23:09:40 -0500 (EST)
Date: Wed, 3 Dec 1997 23:09:39 -0500 (EST)
From: "Alex.Boisvert" <boia01@gel.usherb.ca>
To: chat@FreeBSD.ORG
Subject: Java benchmarks on FreeBSD
Message-ID: <Pine.SUN.3.91.971203230734.20918A-100000@zeus>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-freebsd-chat@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org


Good news for Java programmers on FreeBSD.

As I mentioned earlier today in this list, I have Sun's JDK 1.1.2
running on FreeBSD 2.2-stable.  JavaWorld and Volano just published
a series of benchmarks for Java virtual machines running on
differents hardware/software platforms.

See the article at:
http://www.javaworld.com/javaworld/jw-12-1997/jw-12-volanomark.html)

As stated in the article, no benchmarks were published for FreeBSD's
current JDK 1.1 port (from Keith White) because it couldn't execute the
benchmark program correctly.

Well, I downloaded and ran the VolanoMark and the Client/Server benchmarks
and to my suprise (!) they both worked!

So here are the preliminary results on my home computer
(Intel Pentium 133 MHz) along with the results on NT/Solaris/Linux
with different JVMs running on an Pentium Pro 200 MHz...


VolanoMark v1.0                                         Result
=================================================================
Windows NT, Microsoft Java SDK v2.0                     756
(includes JIT and native threads)

Solaris, Sun's JDK 1.1.3                                334
(includes JIT and native threads)

Solaris, JavaSoft's JDK 1.1.4                           240
(this is the reference release, no JIT)

OS/2 Wrap, IBM's JDK 1.1.4                              239
(include JIT and native threads)

Red Hat Linux 4.2, JavaSoft JDK 1.1.3                   223
(no JIT, green threads)

Windows NT, JavaSoft's JDK 1.1.4                        149
(this is the reference release, no JIT)

FreeBSD 2.2-stable, JavaSoft JDK 1.1.2                  145
(no JIT, green threads, and running on
 a P133 MHz, not a PPro 200 MHz)


Client/Server benchmark                                Result
===============================================================
Windows NT, Microsoft Java SDK v2.0                    1070
(includes JIT and native threads)

OS/2 Wrap, IBM's JDK 1.1.4                              858
(include JIT and native threads)

Solaris, Sun's JDK 1.1.3                                695
(includes JIT and native threads)

Red Hat Linux 4.2, JavaSoft JDK 1.1.3                   521
(no JIT, green threads)

FreeBSD 2.2-stable, JavaSoft JDK 1.1.2                  331
(no JIT, green threads, and running on
 a P133 MHz, not a PPro 200 MHz)

Although we can't compare (nor accurately scale) the results
between my P133 and the PPro200 that was used for the actual tests,
the good news are:  1) it works! and 2) the results on FreeBSD seem
on-par with Linux (and maybe a bit better than Windows NT running
JavaSoft's plain-vanilla JDK 1.1.4).

Moral of the story:  expect a release "real soon now".

Alex.