<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 12 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
span.EmailStyle17
{mso-style-type:personal-reply;
font-family:"Calibri","sans-serif";
color:#1F497D;}
.MsoChpDefault
{mso-style-type:export-only;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 72.0pt 72.0pt 72.0pt;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]--></head><body lang=EN-GB link=blue vlink=purple><div class=WordSection1><div><p class=MsoNormal>On Sat, Apr 27, 2013 at 7:26 PM, Nick Lewycky <<a href="mailto:nicholas@mxc.ca" target="_blank">nicholas@mxc.ca</a>> wrote:<o:p></o:p></p><div><div><blockquote style='border:none;border-left:solid #CCCCCC 1.0pt;padding:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm'><p class=MsoNormal>I just want LLVM to behave the same on whatever platform it's run on. People already accept that depending on iteration order is a bug, but it's been harder to get people to accept that llvm needs bit-exact floating point constant folding, especially given the implementation difficulty.<o:p></o:p></p></blockquote><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal><span style='color:#1F497D'>| </span>I'll give a huge +1 to this.<o:p></o:p></p></div><div><p class=MsoNormal><o:p> </o:p></p></div><div><p class=MsoNormal><span style='color:#1F497D'>| </span>I think that long-term LLVM should, out of the box, support strictly correct (according to IEEE) floating point folding outside of fast-math, and should provide aggressive as hell folding inside of fast-math.<o:p></o:p></p></div><div><p class=MsoNormal><span style='color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D'>I'm a bit conflicted. On the one hand we have _some_ user code where things like addition of literal +0.0f in LLVM IR are being used for their side effects so that having a default mode that guaranteed not to fold them away would be helpful. On the other hand, as someone who sometimes writes numeric code I don't like the lumping of lots and lots of things to which the user code might have differing sensitivity under just one fast-math switch. For example, it might be appropriate to use the equivalent of gcc's ffinite-math-only without assuming the equivalent of gcc's funsafe-math-optimizations (which could dramatically lose precision even though for finite inputs it did ensure finite outputs.) In particular, it'd be nice to be able to have options to get constant-folding-to-host-accuracy without going in to the minefield of enabling full-blown fast-math. Unfortunately I'm not in a position to actually do anything about this presently.<o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D'><o:p> </o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D'>Cheers,<o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D'>Dave<o:p></o:p></span></p></div></div></div></div></div></body></html>