Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA12981; Tue, 13 Feb 96 19:21:45 EST
Received: from liquor.cabi.net by MIT.EDU with SMTP
	id AA27287; Tue, 13 Feb 96 19:21:16 EST
Received: by liquor.cabi.net (8.7.3/8.7.3) id OAA10363; Tue, 13 Feb 1996 14:02:41 -0500
Resent-Date: Tue, 13 Feb 1996 14:02:41 -0500
Date: Tue, 13 Feb 1996 10:58:39 -0800
From: geh@mda.ca (Graeme Hiebert)
Message-Id: <9602131858.AA07243@igneous.mda.ca>
To: java-linux@java.blackdown.org
Subject: Re: Performance issues
X-Sun-Charset: US-ASCII
Resent-Message-Id: <"XNVL62.0.7W2.6_D8n"@liquor>
Resent-From: java-linux@java.blackdown.org
X-Mailing-List: <java-linux@java.blackdown.org> archive/latest/1126
X-Loop: java-linux@java.blackdown.org
Precedence: list
Resent-Sender: java-linux-request@java.blackdown.org

Coincidentally, I was in the midst of a test run to compare the two
things under Solaris when this message came in.  The results there were
that, based on a firm statistical foundation of only on one run, Sun's
java was "slightly" faster than Netscape's.  (Of course, the difference
could be accounted for by any number of things: disk caches, other
users' activities, etc.)  The test involved compiling a single java
file, which indirectly induced compilation of 120 files.  With javac,
it took 1 minute and 59 seconds, with 90.6% CPU usage.  Netscape took
2 minutes and 1 second, at 93.6% CPU usage.  I calculate this as being
5% slower for Netscape than for Sun's javac.  (113.256 seconds for Netscape,
versus 107.814 for javac.)

I will be interested in seeing the results under Linux when I get
home.

   -g
----

> Date: Tue, 13 Feb 1996 10:37:24 -0800 (PST)
> From: Neil Archibald <neil@ohio.catapult.com>
> To: java-linux@java.blackdown.org
> Subject: Performance issues
> 
> 
> I'm sad to bring this up, but I've quit using the Linux java kit, (kind 
> of) on the basis that using the netscape hack way of compiling java 
> programs against the linux's try1 classes.zip IS ACTUALLY faster. Even 
> though I barely have enough RAM to run netscape, I get a average compile 
> time for 11 seconds under netscape compared to 28 seconds under javac. 
> So, now I have javac aliased to run netscape with the hack.
> 
> Oh, and using netscape to run java programs (ie 'java') is faster too.
> 
> But, keep up the good work, i'm sure there will be a time sometime soon 
> when I switch back again :)
> 
> Neil.
> 
> -----BEGIN PGP PUBLIC KEY BLOCK-----
> Version: 2.6.2
> 
> mQBtAzCOpJEAAAEDAN3Bv1/AJavy2KW4YsZaMgE1n/LYBJ1mFJrUrnGnKVOjQo7C
> eLEmoc/LKNKuUbyeIGpbRZYjNXOKnwspKBz5RKVP4GC2zSYoqUUkfNFlSN5/8ouQ
> bF7ak7JVYhyNs/OpRQAFEbQiTmVpbCBBcmNoaWJhbGQgPG5laWxAY2F0YXB1bHQu
> Y29tPg==
> =YzLB
> -----END PGP PUBLIC KEY BLOCK-----
> 
> 

