Mailing-List: contact harmony-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: harmony-dev@incubator.apache.org
Received-SPF: pass (hermes.apache.org: domain of david.griffiths@gmail.com
 designates 64.233.162.197 as permitted sender)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
        s=beta; d=gmail.com;
        h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type;
        b=e5q2YkcZknwB2dnbfNYjZonRhH1kcmvGjXvk7xLntUIOwtaEHp5KjrrUcRtzgS9Ajh+VmIoZx55OBtdr7KNnIH6W/djx/Z68dlTZXp06R5DXSKZ9JRRGLwdapaTJfUn3QmFC+PIIeE33P3RkpUVIzW+GYJ9JJCaGDX/5CZnto5Y=
Message-ID: <892f710b05052304596c4b3b7b@mail.gmail.com>
Date: Mon, 23 May 2005 12:59:28 +0100
From: David Griffiths <david.griffiths@gmail.com>
Reply-To: David Griffiths <david.griffiths@gmail.com>
To: harmony-dev@incubator.apache.org
Subject: Java bytecode metrics?
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_Part_16906_25835576.1116849568325"

------=_Part_16906_25835576.1116849568325
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

I've been wondering about the cost of adding gc points to every backwards=
=20
branch in LLVM (as would need to be done to make it multithreaded). A paper=
=20
here http://research.sun.com/techrep/1998/abstract-70.html suggests the cos=
t=20
is around 5% of total running time (compared to code patching). What I was=
=20
wondering was, does anyone know of any code metrics for large collections o=
f=20
Java bytecode? Eg what is the average number of bytecodes in a method, what=
=20
is the average ratio of backward branches per bytecode and so on.

Is code patching a technique that has been widely employed in JIT compilers=
?=20
(I liked the idea of using an access to a write-protected page to reduce th=
e=20
cost of polling!)

Cheers,

Dave

------=_Part_16906_25835576.1116849568325--