[Bug math/3268] New: fma for soft-fp is not correct.

[Bug math/3268] New: fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1053 bytes --]

The current ./math/s_fma[fl].c implementation simply implements:

double
__fma (double x, double y, double z)
{
  return (x * y) + z;
}

Which does mot meet the requirement of 754r

"fused multiply-add: The operation fma(x,y,z) computes (x ×y )+z as if with
unbounded range and precision, rounding only once to the destination format; see
subclause 5.1."

This implies (for example) that for double that the 106 bit result of the
multiply be passed unrounded to the add (53 bit) and then final result found to
53 bits.

-- 
           Summary: fma for soft-fp is not correct.
           Product: glibc
           Version: 2.4
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
        AssignedTo: aj at suse dot de
        ReportedBy: sjmunroe at us dot ibm dot com
                CC: glibc-bugs at sources dot redhat dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-09-26 21:23 -------
Created an attachment (id=1326)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1326&action=view)
Initial patch for discussion.


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-09-26 21:24 -------
Created an attachment (id=1327)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1327&action=view)
Initial fma patch for PPC32 soft-fp for discussion


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joseph at codesourcery dot
                   |                            |com


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
  BugsThisDependsOn|                            |2749


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for soft-fp is not correct.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-09-26 21:38 -------
These patches create fmasf4.c and fmadf4.c in soft-fp. These function have to
built with soft-fp and exported from libc. Otherwise they can not access the
simulated FPU events, masks and rounding modes which are defined in libc but not
exports.

The soft-fp implementation of s_fma.c and s_fmaf.c reside in libm and are
exported from there. Currently there does not seem to be a good way to override
the generic (and incorrect) ./math/s_fms[f].c from the soft-fp directory. So for
now I placed the overrides in ports/sysdeps/powerpc/nofpu for testing.

I have not implemented a s_fmal.c/fmatf4.c for now. This would imply a soft-fp
(256-bit) extented implementation not supported by the current soft-fp. 

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-10-04 18:27 -------
Corrected title per Joseph Myer comment

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|fma for soft-fp is not      |fma for all targets without
                   |correct.                    |hardware fma instructions is
                   |                            |incorrect.


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-10-05 19:35 -------
Created an attachment (id=1348)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1348&action=view)
updated Ports patch for powerpc32 soft-fp FMA

Updated patch that resolves search patch  build problem.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1327 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2006-10-05 19:36 -------
Created an attachment (id=1349)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1349&action=view)
Updated FMA soft-fp implementation

Cleaned up source and updated comments

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1326 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 780 bytes --]


------- Additional Comments From sjmunroe at us dot ibm dot com  2006-10-05 19:47 -------
Created an attachment (id=1350)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1350&action=view)
New fma_test to verify the intermediate multiply product has full percision.

Add a new fma_test that verifies that:

"The operation fma(x,y,z) computes (x ×y )+z as if with unbounded range and
precision, rounding only once to the destination format"

This implies that for double that the 106 bit result of the multiply be passed
unrounded to the add (53 bit) and then final result found to 53 bits.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-01-22 23:54 -------
Created an attachment (id=1511)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1511&action=view)
Updated patch to ports to add fma for PPC32 soft-fp


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1348 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-01-23 00:01 -------
Created an attachment (id=1512)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1512&action=view)
Updated patch to add single and double fma to soft-fp

This adds macros to double.h and quad.h to copy internal soft-fp values between
the RAW, SEMIRAW, and CANNONICAL formats without using any float types.
Specifically this patch allows fmasf4.c and fmadf4.c to be implemented without
requiring TF type support. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1511 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-01-23 00:03 -------
Created an attachment (id=1513)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1513&action=view)
Updated fma_test to verify that the intermediate multiply product has full
percision


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1350 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-01-23 00:07 -------
Created an attachment (id=1514)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1514&action=view)
Updated patch to ports to add fma for PPC32 soft-fp


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1349 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-01-23 00:14 -------
Created an attachment (id=1515)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1515&action=view)
Updated fma_test to verify that the intermediate multiply product has full
percision

This one includes the change log entry. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1513 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[jakub at redhat dot com]

------- Additional Comments From jakub at redhat dot com  2007-02-05 20:50 -------
Why do you need a soft-fp fmaf other than generic
float fmaf (float x, float y, float z)
{
  return ((double) x * y) + z;
}
?

As DFmode has more than twice as wide mantissa as SFmode and bigger exponent
range as well, (double) x * (double) y will be IMHO always precise, so no
rounding will ever happen there.
Similarly for fma and IEEE quad long double.
Only fmal needs soft-fp or gmp implementation, on all architectures that don't
have it in hardware, and fma for IEEE extended long double or IBM 2x double long
double.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-02-06 16:30 -------
Yes what you suggest should work for the generic s_fmaf.c, but that is not the
current implmentation. Also fmasf4.c should be a little faster as it avoids some
of the more expensive intermediate conversions.

s_fma.c would be more problematic because it would require full IEEE quad and TF
type support. I believe that you youself pointed out that not all platforms
support IEEE quad/TF type, but those platforms still need fma() to work correctly. 

That is why I reworked the patch to avoid internal use of TF typs and support
direct RAW/SEMIRAW/CANNONICAL conversion for the intermediate steps of the
soft-fp fma. Again fmadf4.c should be faster then the generic.

Yes in the long run we will need to address soft-fp fmal but that requires the
invention of "oct or long long double" format. Most platforms that need soft-fp
don't implement long double 128-bit so we stage that function. 

But we have a fix for soft-fp float and double now. I don't see any reason to delay.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-02-06 16:31 -------
No one else is working on this so, I'll assign it to my self.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-02-06 16:32 -------
I think this bug is fixed. 

If you want I can open a separate bug for soft-fp fmal support.

Anyone disagree?

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[aj at suse dot de]

------- Additional Comments From aj at suse dot de  2007-02-18 13:51 -------
following comment #15.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|aj at suse dot de           |sjmunroe at us dot ibm dot
                   |                            |com


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

-- 
Bug 3268 depends on bug 2749, which changed state.

Bug 2749 Summary: powerpc32 does not build --without-fp
http://sourceware.org/bugzilla/show_bug.cgi?id=2749

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED

http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-04-03 18:58 -------
Created an attachment (id=1664)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1664&action=view)
Updated ports soft-fp fma patch for PPC32

Verified with CVS from April 3rd.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1514 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[joseph at codesourcery dot com]

------- Additional Comments From joseph at codesourcery dot com  2007-04-03 18:58 -------
Subject: Re:  fma for all targets without hardware fma instructions is incorrect.

Thank you for your message.

I am away from my email from 29 March to 2 April.  Messages sent
before 29 March will be read before I go away; messages sent during
that period will be read after I return.  If you need a response
before my return, please contact Nathan Sidwell
<nathan@codesourcery.com>.



-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-04-03 19:00 -------
Created an attachment (id=1665)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1665&action=view)
updated patch to add float/double fma to soft-fp

verified with glibc cvs from April 3rd.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1512 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[sjmunroe at us dot ibm dot com]

------- Additional Comments From sjmunroe at us dot ibm dot com  2007-04-03 20:39 -------
Created an attachment (id=1666)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=1666&action=view)
Updated fma_test for current CVS

Verfied on CVS from April 3rd

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #1515 is|0                           |1
           obsolete|                            |


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[vincent+libc at vinc17 dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vincent+libc at vinc17 dot
                   |                            |org


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[nbowler at draconx dot ca]

------- Additional Comments From nbowler at draconx dot ca  2010-02-02 21:56 -------
Created an attachment (id=4569)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=4569&action=view)
Example demonstrating fma failure on x86_64

The fma function is still incorrect on x86_64 (with glibc version 2.11) when
the multiplication overflows but the final result would not: the result is an
infinity instead of the rounded true result.

See the attached C program for a demonstration.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[nbowler at draconx dot ca]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nbowler at draconx dot ca


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug math/3268] fma for all targets without hardware fma instructions is incorrect.

[zimmerma+gcc at loria dot fr]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zimmerma+gcc at loria dot fr


http://sourceware.org/bugzilla/show_bug.cgi?id=3268

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.