Reply-to: Geoffrey Fox To: "Darema, Frederica" Subject: Is this what you want Date: Tue, 06 Aug 2002 11:26:41 -0500 From: Geoffrey Fox Geoffrey Fox gcf@indiana.edu FAX 8128567972 Phones Cell 315-254-6387 Home 8123239196 Lab 8128567977 CS 8128553788 ------- Forwarded Message Date: Fri, 14 Jun 2002 12:12:33 -0500 From: Geoffrey Fox To: "Darema, Frederica" Subject: Re: I can referee Getov/Lu paper! - --------------020300080403010209090701 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit This paper is as they say not wrong. However the paper contains "just measurements" of performance but gives no insight as to what causes Java slowness. From work by others (eg IBM) we know that slowness comes from several areas - --- handling of floating point arithmetic - --- Array index checks - ---- Lack of "real arrays" I see that adding some discussion of the reasons for results would make paper much better I note a) IBM and probably Sun have high performance Java compilers b) The ratio of Java/Fortran is different for particles & field parts of computation (see their tables) Why? c) The last sentence of section 2 is bizarre; why give the computer more communication burden? This paper is publishable with modest changes; it needs major upgrade to be a good paper Darema, Frederica wrote: >meant to say - "hard" copy! FD > >>-----Original Message----- >>From: Darema, Frederica >>Sent: Monday, May 20, 2002 2:20 PM >>To: 'Geoffrey Fox' >>Cc: Baker, Beverly P. >>Subject: RE: I can referee Getov/Lu paper! >>Importance: High >> >>Thank you Geoffrey. I get a soft copy of these, so Mrs. Baker will FedEx >>it to you today. >>Regards, Frederica >> >> -----Original Message----- >> From: Geoffrey Fox [SMTP:gcf@grids.ucs.indiana.edu] >> Sent: Monday, May 20, 2002 2:09 PM >> To: Darema, Frederica >> Subject: I can referee Getov/Lu paper! >> >> Please send electronically >> Thanks! >> >> Geoffrey Fox gcf@indiana.edu FAX 8128567972 >> Phones Cell 315-254-6387 Home 8123239196 Lab 8128567977 CS >>8128553788 >> > > - -- : : Geoffrey Fox gcf@indiana.edu FAX 8128567972 : Phones Cell 315-254-6387 Home 8123239196 Lab 8128567977 CS 8128553788 - --------------020300080403010209090701 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit This paper is as they say not wrong. However the paper
contains "just measurements" of performance but gives
no insight as to what causes Java slowness.
From work by others (eg IBM) we know that
slowness comes from several areas
- --- handling of floating point arithmetic
- --- Array index checks
- ---- Lack of "real arrays"
I see that adding some discussion of the reasons for
results would make paper much better

I note
a) IBM and probably Sun have high performance Java compilers
b) The ratio of Java/Fortran is different for particles & field parts
of computation (see their tables) Why?
c) The last sentence of section 2 is bizarre; why give the computer
more communication burden?

This paper is publishable with modest changes; it needs major upgrade
to be a good paper

Darema, Frederica wrote:

meant to say - "hard" copy!  FD

-----Original Message-----
From:	Darema, Frederica 
S
ent:	Monday, May 20, 2002 2:20 PM
To:	'Geoffrey Fox'
Cc:	Baker, 
Beverly P.
Subject:	RE: I can referee Getov/Lu paper!
Importance:
	High

Thank you Geoffrey.  I get a soft copy of these, so Mrs. Ba
ker will FedEx
it to you today.  
Regards, Frederica 

	-----Or
iginal Message-----
	From:	Geoffrey Fox [SMTP:gcf@grids.ucs.indiana.edu]

	Sent:	Monday, May 20, 2002 2:09 PM
	To:	Darema, Frederi
ca
	Subject:	I can referee Getov/Lu paper!

	Please send ele
ctronically 
	Thanks!

	Geoffrey Fox  gcf@indiana.edu FAX 8128567972

	Phones Cell 315-254-6387 Home 8123239196 Lab 8128567977 CS
812855378
8

-- 
:
: Geoffrey Fox  gcf@indiana.edu FAX 8128567972
: Phones Cell 315-254-6387 Home 8123239196 Lab 8128567977 CS 8128553788

- --------------020300080403010209090701-- ------- End of Forwarded Message