<html><body>
<p><font size="2" color="#800080">----- Forwarded by Robert Enenkel/Toronto/IBM</font><font size="2" color="#800080"> on 06/03/2010 01:08 PM</font><font size="2" color="#800080"> -----</font><br>
<br>
<table width="100%" border="0" cellspacing="0" cellpadding="0">
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">From:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Ian McIntosh/Toronto/IBM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">To:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Rajan Bhakta/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Cc:</font></td><td width="100%" valign="middle"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Raymond Mak/Toronto/IBM@IBMCA, Robert Enenkel/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Date:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">06/02/2010 07:45 PM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Subject:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Re: Fw: proposed masked test to categorize floating point values</font></td></tr>
</table>
<hr width="100%" size="2" align="left" noshade style="color:#8091A5; "><br>
<br>
In
<ul> <font color="#0000FF"> Issue: The _Imaginary types need to be supported.</font><br>
<font color="#0000FF"> One way would be to say that this same family of functions works on imaginary types too; eg, testfpf64x would work on _Imaginary _Float64x.</font><br>
<font color="#0000FF"> Another approach would be to define additional functions.</font><br>
<b><font color="#FF0000">We can't do this without tgmath support.</font></b></ul>
<br>
Which suggestion can we not do without tgmath support (just using the same functions? or defining additional functions? or doing anything at all?)<br>
What are the implications?<br>
Do we currently have tgmath support?<br>
Do we plan it?<br>
Do others have it?<br>
What wording should I put?<br>
<br>
<br>
<br>
<b><font size="4">All:</font></b> I should send this tomorrow morning. Since we haven't had time to come to a single IBM position, it will probably go out roughly as is. I'm wondering if I should change the "testfp" suggestion to "classifyfp" or "fpclassify" (the existing macro name)?<br>
<br>
- Ian McIntosh IBM Canada Lab Compiler Back End Support and Development<br>
<br>
<br>
<br>
<br>
<table width="100%" border="0" cellspacing="0" cellpadding="0">
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">From:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Rajan Bhakta/Toronto/IBM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">To:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Ian McIntosh/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Cc:</font></td><td width="100%" valign="middle"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Raymond Mak/Toronto/IBM@IBMCA, Robert Enenkel/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Date:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">06/02/2010 12:51 PM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Subject:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Re: Fw: proposed masked test to categorize floating point values</font></td></tr>
</table>
<hr width="100%" size="2" align="left" noshade style="color:#8091A5; "><br>
<br>
My preferences in <font color="#FF0000">red </font>below.<br>
My comments in <b><font color="#FF0000">bold red </font></b>below.<br>
<br>
Regards,<br>
<br>
Rajan Bhakta<br>
z/OS C/C++ Technical Architect<br>
ISO C Standards Representative for Canada, IBM<br>
C Compiler Development<br>
Contact: rbhakta@ca.ibm.com, Rajan Bhakta/Toronto/IBM<br>
Telephone: (905) 413-3995<br>
<br>
<br>
<br>
<br>
<table width="100%" border="0" cellspacing="0" cellpadding="0">
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">From:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Ian McIntosh/Toronto/IBM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">To:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Robert Enenkel/Toronto/IBM@IBMCA, Raymond Mak/Toronto/IBM@IBMCA, Rajan Bhakta/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Date:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">06/02/2010 10:54 AM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Subject:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Fw: proposed masked test to categorize floating point values</font></td></tr>
</table>
<hr width="100%" size="2" align="left" noshade style="color:#8091A5; "><br>
<br>
A couple of changes<font color="#0000FF">, in blue</font>.<br>
<br>
- Ian McIntosh IBM Canada Lab Compiler Back End Support and Development<br>
<br>
<font size="2" color="#800080">----- Forwarded by Ian McIntosh/Toronto/IBM on 06/02/2010 10:04 AM -----</font><br>
<br>
<table width="100%" border="0" cellspacing="0" cellpadding="0">
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">From:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Ian McIntosh/Toronto/IBM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">To:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">Robert Enenkel/Toronto/IBM@IBMCA, Raymond Mak/Toronto/IBM@IBMCA, Rajan Bhakta/Toronto/IBM@IBMCA</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Date:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">06/01/2010 08:04 PM</font></td></tr>
<tr valign="top"><td width="1%"><img width="96" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2" color="#5F5F5F">Subject:</font></td><td width="100%"><img width="1" height="1" src="cid:1__=0ABBFDA4DFCDB5938f9e8a93df938@ca.ibm.com" border="0" alt=""><br>
<font size="2">proposed masked test to categorize floating point values</font></td></tr>
</table>
<hr width="100%" size="2" align="left" noshade style="color:#8091A5; "><br>
<br>
This is based on what Robert and I discussed, with some additional options. Please comment.<br>
<br>
<font color="#0000FF">This would offer a capability similar to the fpclassify macro, but with a mask parameter. It could, like fpclassify, be a macro instead of a function.</font><br>
<br>
Function Name:<br>
<font color="#0000FF"> Issue: What name(s) should be used?</font><br>
<font color="#0000FF">Proposal 1:</font><br>
<b>testfp</b><suffix> where <suffix> is null for double, "f" for float, "ld" for long double, "f128", "f64", "f32" or "f16" for binary, "f64x", "f32x" etc. for extended binary, "d32", "d64" or "d128" for dfp.<br>
This tests one scalar value.<br>
<font color="#0000FF"> Issue: The _Imaginary types need to be supported.</font><br>
<font color="#0000FF"> One way would be to say that this same family of functions works on imaginary types too; eg, testfpf64x would work on _Imaginary _Float64x.</font><br>
<font color="#0000FF"> Another approach would be to define additional functions.</font><br>
<b><font color="#FF0000">We can't do this without tgmath support.</font></b><br>
<font color="#0000FF"> Issue: The _Complex types need to be supported.</font><br>
<font color="#0000FF"> One way would be to define additional functions; eg, testfpcomplexf64x, with semantics like testfpvectorany below.</font><br>
<font color="#0000FF"> Another would be to define functions like all three below.</font><br>
<br>
<font color="#0000FF"> The C Standard does not include vector types, but many implementations do. The following are included here not to become part of the standard yet, but to suggest a future approach.</font><br>
<b>testfpvector</b><#><suffix> where <#> is the vector length (number of values) and <suffix> is as above.<br>
This tests a vector of # values of the type implied by the suffix.<br>
Example: testfpvector4f32 tests 4 32 bit values and gives 4 results.<br>
<b>testfpvectorany</b><#><suffix> similar to testfpvector, returning a single result.<br>
<b>testfpvectorall</b><#><suffix> similar to testfpvector, returning a single result.<br>
<font color="#0000FF"> Issue: What about a vector of complex?</font><br>
<b><font color="#FF0000">Can be treated as suggested for _Complex types above.</font></b><br>
<font color="#0000FF"> Proposal 2:</font><br>
<font color="#0000FF"> Same, but using "classify" instead of "test".</font><br>
<font color="#0000FF"> Proposal 3:</font><br>
<font color="#0000FF"> Same, but using "categorize" instead of test".</font><br>
<font color="#0000FF"> Proposal 4:</font><br>
<font color="#0000FF"> Shorten names by abbreviating the verb and/or omitting "fp".</font><br>
<font color="#0000FF"> Proposal 5:</font><br>
<font color="#0000FF"> Make it an overloaded macro, like fpclassify is.</font><br>
<font color="#0000FF"> Proposal 6:</font><br>
<font color="#0000FF"> Overload the fpclassify macro - if there is only one parameter it acts like now, but if there is a second parameter is is a mask.</font><br>
<br>
<font color="#0000FF">Issue: S</font>hould <font color="#0000FF">we </font>define macros set by the compiler to indicate which functions are present and fast/slow<font color="#0000FF">, as is done for the fma functions?</font><br>
<br>
Function Parameters:<br>
value type double, float, long double, _Float#, _Float#x, _Decimal#, etc., or a vector of one of those.<br>
mask unsigned long, containing zero or more mask bits ored together:<br>
We need to choose macro names for these:<br>
<br>
Proposal 1:<br>
Result is true iff any of the conditions is true. The sign is ignored.<br>
0x01 zero<br>
0x02 normal not equal to zero<br>
0x04 subnormal<br>
0x08 infinity<br>
0x10 quiet NaN<br>
0x20 signaling NaN<br>
0x40 noncanonical (dfp only)<br>
eg, 0x38 means infinities and NaNs.<br>
Proposal 2:<br>
Result is true iff either of the first two conditions is true and any of the other conditions is true.<br>
0x01 positive<br>
0x02 negative<br>
0x04 zero<br>
0x08 normal not equal to zero<br>
0x10 subnormal<br>
0x20 infinity<br>
0x40 quiet NaN<br>
0x80 signaling NaN<br>
0x100 noncanonical (dfp only)<br>
eg, 0xe3 means negative or positive infinities and NaNs, 0x1fe means all negatives, 0x15 means positive zero or subnormal.<br>
Proposal 3:<br>
Same as proposal 2, except that 0x01 means "only positive" and 0x02 means "only negative", so in the normal case one need not add/or in 3 to get both signs (simplifies typical usage).<br>
<font color="#0000FF">Issue: Which proposal? Or something else? </font>I recommend proposal 3.<br>
<font color="#0000FF"> Issue: Should the type be int or unsigned int?</font><br>
<font color="#FF0000">I think it should stay unsigned long (for flexibility/expansion), and it certainly should not be a signed type.</font><br>
<font color="#0000FF"> Issue: We should define macros for each of these; for example, in proposal 3 the way to specify infinities and NaNs might be "FP_CLASS_INF | FP_CLASS_SNAN | FP_CLASS_QNAN".</font><br>
<font color="#FF0000">I think this would be useful for a programmer.</font><br>
<br>
Function Result:<br>
For testfp: _Bool, true if the condition is true, false if it is not.<br>
For testfpvector: unsigned long, with # rightmost bits true or false, corresponding to the vector values<br>
<font color="#0000FF"> Issue: Should the type be int or iunsigned int, like fpclassify returns?</font><br>
<font color="#FF0000">I think it should stay unsigned long (for flexibility/expansion), and it certainly should not be a signed type.</font><br>
For testfpvectorany: _Bool, true if f the condition is true for <b>any</b> of the vector values, false if it is not.<br>
For testfpvectorall: _Bool, true if f the condition is true for <b>all</b> of the vector values, false if it is not.<br>
<br>
<br>
- Ian McIntosh IBM Canada Lab Compiler Back End Support and Development<br>
<br>
<br>
<br>
<br>
<br>
</body></html>