Subject: Re: C433 JGFSI Reviews
Resent-Date: Sun, 28 Nov 1999 16:26:28 -0500
Resent-From: Geoffrey Fox <gcf@npac.syr.edu>
Resent-To: p_gcf@npac.syr.edu
Date: Sun, 28 Nov 1999 14:33:51 -0500
From: Maurice Herlihy <mph@cs.brown.edu>
To: Geoffrey Fox <gcf@npac.syr.edu>

In my next message, I will send you the revised version of our Java Grande
paper.  I appreciate your patience.

MAJOR
As far as major comments, two referees felt that we should do a more careful
analysis of benchmarks other than the simple counting benchmark.  To that
end,
Section 6 describes a much more thorough set of experiments on three
additional
benchmarks.  We found that the relative performance of the Home and Arrow
directories is largely a function of the cache hit ratio: the worse the
ratio,
the greater the performance difference.  (The principal reason for our delay
was arranging interference-free access to enough workstations.)

One of the referees suggested we run our experiments on Wintel workstations.
Unfortunately, we don't have access to a sufficient number, so we had to
stick
with Solaris and Digital Unix.

The referees pointed out a number of typos, instances of bad or ambiguous
phrasing, poorly-drawn figures, etc.  We have adapted these suggestions
exactly
as requested.

        Maurice