[daip] forwarded message from John Martellaro

[Eric Greisen]

Comments?

Eric
------- start of forwarded message (RFC 934 encapsulation) -------
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii" ; format="flowed"
Received: from revere.aoc.nrao.edu (revere [146.88.1.15])
	by zia.aoc.nrao.edu (8.12.9/8.12.5) with ESMTP id h6AE5QpP020364
	for <egreisen at aoc.nrao.edu>; Thu, 10 Jul 2003 08:05:26 -0600 (MDT)
Received: from mail-out1.apple.com (mail-out1.apple.com [17.254.0.52])
	by revere.aoc.nrao.edu (8.11.6/8.11.6) with ESMTP id h6AE5Nj23549
	for <egreisen at aoc.nrao.edu>; Thu, 10 Jul 2003 08:05:23 -0600
Received: from mailgate2.apple.com (A17-129-100-225.apple.com [17.129.100.225])
	by mail-out1.apple.com (8.12.9/8.12.9) with ESMTP id h6AE5IiB003917
	for <egreisen at aoc.nrao.edu>; Thu, 10 Jul 2003 07:05:18 -0700 (PDT)
Received: from scv2.apple.com (scv2.apple.com) by mailgate2.apple.com
 (Content Technologies SMTPRS 4.2.1) with ESMTP id <T6357700676118164e16b0 at mailgate2.apple.com> for <egreisen at aoc.nrao.edu>;
 Thu, 10 Jul 2003 07:05:21 -0700
Received: from [192.168.1.8] (vpn-scv-x2-201.apple.com [17.219.193.201])
	by scv2.apple.com (8.12.9/8.12.9) with ESMTP id h6AE5EaI005701
	for <egreisen at aoc.nrao.edu>; Thu, 10 Jul 2003 07:05:15 -0700 (PDT)
X-Sender: jmartellaro at mail.apple.com
Message-Id: <p06001400bb3320cce9e2@[192.168.1.8]>
In-Reply-To: <16140.17075.524147.345165 at primate.aoc.nrao.edu>
References: <p06001415bb30dff5ad78@[192.168.1.8]>
 <16140.17075.524147.345165 at primate.aoc.nrao.edu>
X-MailScanner: Found to be clean
X-MailScanner-SpamCheck: not spam, SpamAssassin (score=-1.7, required 7,
	EMAIL_ATTRIBUTION, HOT_NASTY, IN_REP_TO, REFERENCES,
	SPAM_PHRASE_00_01)
From: John Martellaro <jm at apple.com>
To: Eric Greisen <egreisen at zia.aoc.NRAO.EDU>
Subject: Re: Hello from Apple
Date: Thu, 10 Jul 2003 08:05:19 -0600

It was great talking to you.  Thanks for all the insights and 
connecting me to Mr Hunt.

Yesterday, we talked briefly about matching an algorithm to the 
machine architecture.  I agree with you that time consuming tweaks 
for a specific architecture are often not possible. However, when one 
elects NOT to tweak, then one has to keep in mind that average 
performance is not an inherent characteristic of the product.

For example, I have permission to pass this on to you from the 
Brookhaven National Laboratory.

Cheers,
John

At 8:55 AM -0400 7/10/03, XXXX XXXXXXXXX wrote:
>>  >   Folks,
>>  >
>>  >   I've got an Apple Xserve 1U unit on loan for two weeks, for evaluation.
>>  >   I just ran a quick benchmark, and I'm still chuckling over it.
>>  >
>>  >   The optimal method for comparing DNA and protein sequences is the
>>  >   Smith-Watermann algorithm, which is NxM in both time and memory,
>>  >   where N and M are the lengths of the two sequences.  It is almost
>>  >   never used for searching large databases, because its usually too slow.
>>  >
>>  >   I just obtained a copy of a Smith-Watermann, from the Pearson group at
>>  >   U. Virginia, which has been optimized for the (G4) Altivec Velocity
>>  >   Engine.
>>  >   Here are the timings for searching for matches to two 500
>>  >   amino-acid-long
>>  >   proteins from a set of 28,000+ sequences from the Arabidopsis thaliana
>>  >   genome (12 million amino acids total), compared to running times of
>>  >   the same program (non-Altivec) and data set on an SGI Origin 200
>>  >   (Irix 6.5) and a Pentium IV (OpenBSD)
>>  >
>>  >     600 MHz  R14000 SGI                3 min 34 sec
>>  >
>>  >     1.7 GHz Pentium IV                 3 min 36 sec
>>  >
>>  >     1.33 GHz PPC G4                          25 sec    (!!!)
>>  >
>>  >  This is a REAL test of a program that we REALLY use, on REAL data.
>>  >  Clearly this is the FASTEST computer in our department, and is probably
>>  >  one of the fastest onsite.  Its certainly the fastest CPU I'VE ever used.
>>  >
>>  >  I will keep repeating this, relentlessly, until I hear people saying it
>>  >  to themselves and to others:  The PPC G4 is no ordinary computer - IT IS
>>  >  A SUPERCOMPUTER.  Other computers are simply NOT in the same LEAGUE and
>>  >  do NOT deserve the same RESPECT.

- -- 

John Martellaro
Federal Account Manager
Business and Government Sales
Apple Computer, Inc.
ph:   303.816.5071
cell: 303.513.3929
http://www.apple.com/federal

------- end -------