Subject: Re: C434 JGFSI Reviews
Resent-Date: Fri, 12 Nov 1999 08:40:21 -0500
Resent-From: Geoffrey Fox <gcf@npac.syr.edu>
Resent-To: p_gcf@npac.syr.edu
Date: Thu, 4 Nov 1999 15:34:09 +0000 (GMT)
From: Mark Bull <markb@epcc.ed.ac.uk>
To: Geoffrey Fox <gcf@npac.syr.edu>

Dear Geoffrey,

Please find attached the revised version of paper C434. We have
responded to the referees' comments as follows. Please let me know if
you would like to see any further changes.

Best wishes,
Mark.

Referee 1:  all suggested changes implemented

Referee 2:

    1) Agreed - remark added to last para of Section 2.

    2) Point taken - will look at including this in future versions.

    3) Remark added to para 1 of Section 6.1

    4) Point noted. This would however require some in-depth analysis
       which is beyond the intended scope of this paper.

    Comment about JIT warmup. We completely agree with the referee
    for kernel and application type benchmarks. However, for low-level
    benchmarks (which as programs are very artificial)
    it would seem to make some sense to try and obtain performance
    which is representative of the best the JVM can manage (i.e. under
    conditions which favour optimisation).

Referee 3:  no specific comments