From owner-freebsd-java Thu Mar 28 18:34:14 2002 Delivered-To: freebsd-java@freebsd.org Received: from cafe.muraoka.info.waseda.ac.jp (cafe.muraoka.info.waseda.ac.jp [133.9.68.49]) by hub.freebsd.org (Postfix) with ESMTP id AD28337B404 for ; Thu, 28 Mar 2002 18:34:10 -0800 (PST) Received: from muraoka.info.waseda.ac.jp (shudoh@localhost [127.0.0.1]) by cafe.muraoka.info.waseda.ac.jp (8.9.1a/3.7W) with ESMTP id LAA02202; Fri, 29 Mar 2002 11:33:58 +0900 Message-Id: <200203290233.LAA02202@cafe.muraoka.info.waseda.ac.jp> To: green@redhat.com Cc: freebsd-java@FreeBSD.ORG Subject: Re: [press@apache.org: PRESS RELEASE: ASF Reaches Agreement with Sun to Allow Open Source Java Implementations] From: shudo@computer.org In-Reply-To: <1017330145.2206.84.camel@dhcppc2> References: <20020327221634.M1335-100000@yez.hyperreal.org> <20020328064218.GA2973@gnuppy.monkey.org> <1017330145.2206.84.camel@dhcppc2> Date: Fri, 29 Mar 2002 11:33:58 +0900 Sender: owner-freebsd-java@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org Anthony Green wrote: > I think the gcj project has been very successful. > The performance of the resulting binaries is good. It's competitive > with state-of-the-art JIT systems. Meaning... sometimes faster, > sometimes slower. When it's faster, it can be much faster. For > instance, certain crypto operations used in SSL can run 10x faster than > state-of-the-art commercial JITs. I agree completely on your view that the GCJ is being mature in both performance and stability. Performance tests also endorse the fact: Performance comparison of JITs (Jan 2002) http://www.shudo.net/jit/perf/ But, your wording like `GCJ is 10 times faster than modern JITs' is misleading even it is not false at all. You may refer the following report, which says the GCJ do a work in a second but the HotSpot VM takes over 10 seconds. GCJ success stories (was: Re: status of gcj's boehm collector?) http://gcc.gnu.org/ml/java/2001-12/msg00108.html > And start up times are always much faster. Start up time (latency) and throughput (or peak performance) can be measured separately and evaluating separately is the way we as engineers prefer, I suppose. Bill Huey wrote: > IMO, from looking at all the HotSpot/JVM internals over this year and a > half time frame, that's impossible for an open source group to commit > the necessary resources to create a useable J2SE clone. You need too > many domain experts with too long term a focus to complete each > subsystem. I'm also sure there are a small number of people who can contribute to state-of-the-art language systems. But see GCC and GCJ. They are going forward certainly. Besides, a certain number of experts are working on open source Java projects, like GCJ, Hans Boehm, John Whaley, Kaffe, SableVM, OpenJIT, Wonka and others. Each one of us only do what will lead the world to what each hope. Kazuyuki Shudo shudo@computer.org http://www.shudo.net/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-java" in the body of the message