<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
I would do a comparison of -O2 and -O3 against the baseline to
demonstrate that they do make a difference. Then the question is
will all of your optimizations produce the same results at -O3 -- or
do you have to back down to -O2.<br>
<br>
That being said, here is what I recommend.<br>
<br>
1. Create your baseline and then do the -O, -O2, -O3 comparison on
one chart.<br>
<br>
2. Then try out the -march -mtune option to see if that helps. Use
the highest level of optimization from above that gives consistent
results.<br>
<br>
Once you have done these things (which we will count as two
optimization steps) you should be ready to demonstrate how any code
rewriting you do impacts optimization. The problem here is that
some of your code rewriting - coupled to the optimization flags -
may impact your results. For that reason you should try to pick the
optimization level above that works consistently across all of your
code rewriting optimizations. Your ultimate goal is to produce
faster code that gives identical results. If you come up with
something that dramatically increases the speed of the code, but
slightly modifies the results, then count that as a good
optimization but demonstrate the differences it produces in your
paper and be able to explain why you think that is happening.<br>
<br>
Does that help?<br>
<br>
<br>
<br>
<div class="moz-cite-prefix">On 02/27/2016 10:23 AM, Nicholas Allan
Rasmuson wrote:<br>
</div>
<blockquote
cite="mid:SN1PR01MB1629174E098FA29A2F464377EFB80@SN1PR01MB1629.prod.exchangelabs.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
<div id="divtagdefaultwrapper"
style="font-size:12pt;color:#000000;background-color:#FFFFFF;font-family:Calibri,Arial,Helvetica,sans-serif;">
<p>Dr. Pounds:</p>
<p><span> For the optimizations do you want me to count
things like differing compiler flags as their own
optimizations and run them separately? Such as testing -O2
and -O3 as well as the cpu base flag as three separate
optimizations and graph the effects, or lack thereof, as
such?</span><br>
</p>
<p><span>Sincerely, </span></p>
<p><span>Nicholas Rasmuson</span></p>
</div>
</blockquote>
<br>
<pre class="moz-signature" cols="72">--
Andrew J. Pounds, Ph.D. (<a class="moz-txt-link-abbreviated" href="mailto:pounds_aj@mercer.edu">pounds_aj@mercer.edu</a>)
Professor of Chemistry and Computer Science
Mercer University, Macon, GA 31207 (478) 301-5627
<a class="moz-txt-link-freetext" href="http://faculty.mercer.edu/pounds_aj">http://faculty.mercer.edu/pounds_aj</a>
</pre>
</body>
</html>