This is the mail archive of the kawa@sourceware.org mailing list for the Kawa project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: Performance

From: Luis Casillas <luis at casillas dot org>
To: Per Bothner <per at bothner dot com>
Cc: Tony <tonys-mailbox at hotmail dot com>, kawa at sourceware dot org
Date: Mon, 23 Jun 2008 13:35:23 -0700
Subject: Re: Performance
References: <BAY115-W22D4D18E1168CCA4F506AFE1AA0@phx.gbl> <485D4BE7.6080603@bothner.com> <485D8086.9050309@bothner.com>

I remember that a long time ago, in older versions of the original shootout, Java used to rank really badly until the shootout author made the following changes:

1. Added a Hello World test program to measure startup time.

2. Subtracted the time of the Hello World test from every other result in the comparison table.

I wonder if (a) the new shootout employs a similar method for comparing implementations, (b) given that, whether you're comparing the test runtimes properly.

The new shootout does have the Hello World test as the "startup" example.

On Jun 21, 2008, at 3:28 PM, Per Bothner wrote:

Per Bothner wrote:

binarytrees.java takes about 1.6s, while binarytrees.scm takes
about 1.2s.  These are unofficial numbers, of course.


Ooops that was backwards - I'm now (on JDK 1.5) getting
1.2s for binarytrees.java and 1.7s for binarytrees.scm.
Startup time is definitely a factor: Running the latter
10 times takes about 10.8s.
--
	--Per Bothner
per@bothner.com   http://per.bothner.com/

References:
- Performance
  - From: Tony
- Re: Performance
  - From: Per Bothner
- Re: Performance
  - From: Per Bothner

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]