This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug math/14283] Inaccurate range reduction for float (sinf, cosf)


http://sourceware.org/bugzilla/show_bug.cgi?id=14283

Liubov Dmitrieva <liubov.dmitrieva at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |liubov.dmitrieva at gmail
                   |                            |dot com
         Resolution|                            |FIXED

--- Comment #1 from Liubov Dmitrieva <liubov.dmitrieva at gmail dot com> 2012-06-25 10:38:25 UTC ---
The patch works well.

Before patch <http://sourceware.org/ml/libc-alpha/2012-06/msg00650.html> 
applied Intel proprietary test framework yields the following results:

On I386:
Summary:
----------------------------------
Function     PASS/  Max Ulp   Roun
             FAIL    error    ding
----------------------------------
SINF         FAIL 6.987e+06   Near
SINF         FAIL 2.334e+07  +Inf 
SINF         FAIL 1.191e+07  -Inf 
SINF         FAIL 2.229e+06   Zero
----------------------------------
COSF         FAIL 1.141e+05   Near
COSF         FAIL 4.102e+05  +Inf 
COSF         FAIL 2.003e+05  -Inf 
COSF         FAIL 27.631139   Zero
----------------------------------

On x86_64:
Summary:
----------------------------------
Function     PASS/  Max Ulp   Roun
             FAIL    error    ding
----------------------------------
SINF         FAIL 7.390e+06   Near
SINF         FAIL 2.054e+07  +Inf 
SINF         FAIL 3.759e+06  -Inf 
SINF         FAIL 3.759e+06   Zero
----------------------------------
COSF         FAIL 2.489e+05   Near
COSF         FAIL 2.754e+05  +Inf 
COSF         FAIL 2.754e+05  -Inf 
COSF         FAIL 2.754e+05   Zero
----------------------------------


I applied the patch, and results now good (I allowed error 5 ulp, hence no
errors):

On I386:
Summary:
----------------------------------
Function     PASS/  Max Ulp   Roun
             FAIL    error    ding
----------------------------------
SINF         PASS  0.949884   Near
SINF         PASS  1.501194  +Inf 
SINF         PASS  1.542971  -Inf 
SINF         PASS  1.542971   Zero
----------------------------------
COSF         PASS  2.905172   Near
COSF         PASS  2.905172  +Inf 
COSF         PASS  3.905172  -Inf 
COSF         PASS  3.905172   Zero
----------------------------------


On x86_64:
Summary:
----------------------------------
Function     PASS/  Max Ulp   Roun
             FAIL    error    ding
----------------------------------
SINF         PASS  1.213484   Near
SINF         PASS  3.542971  +Inf 
SINF         PASS  1.854113  -Inf 
SINF         PASS  1.875448   Zero
----------------------------------
COSF         PASS  2.905172   Near
COSF         PASS  3.544283  +Inf 
COSF         PASS  3.905172  -Inf 
COSF         PASS  3.905172   Zero
----------------------------------

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]