<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
On 05/04/2015 08:49 PM, Levent Erkok wrote:<br>
<blockquote
cite="mid:CACOpX0fMYcfXq2SBmcCpWTN_rD9w-PdLvqpasA40GL7ezyR_3A@mail.gmail.com"
type="cite">
<div dir="ltr">Artyom: That's precisely the point. The true
IEEE754 variants where precision does matter should be part of a
different class. What Edward and Yitz want is an "optimized"
multiply-add where the semantics is the same but one that goes
faster.</div>
</blockquote>
No, it looks to me that Edward wants to have a more precise
operation in Num:<br>
<blockquote type="cite">I'd have to make a second copy of the
function to even try to see the precision win.</blockquote>
Unless I'm wrong, you can't have the following things
simultaneously:<br>
<ol>
<li>the compiler is free to substitute <i>a+b*c</i> with <i>mulAdd
a b c</i></li>
<li><i>mulAdd a b c</i> is implemented as <i>fma</i> for Doubles
(and is more precise)</li>
<li>Num operations for Double (addition and multiplication) always
conform to IEEE754</li>
</ol>
<p>
<blockquote type="cite">The true IEEE754 variants where precision
does matter should be part of a different class.</blockquote>
So, does it mean that you're fine with not having point #3 because
people who need it would be able to use a separate class for
IEEE754 floats?<br>
</p>
</body>
</html>