[Bug math/18613] New: tgamma inaccurate in non-default rounding modes

[jsm28 at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=18613

            Bug ID: 18613
           Summary: tgamma inaccurate in non-default rounding modes
           Product: glibc
           Version: 2.21
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
          Assignee: unassigned at sourceware dot org
          Reporter: jsm28 at gcc dot gnu.org
  Target Milestone: ---

In non-default rounding modes, tgamma can be slightly less accurate than
permitted by glibc's accuracy goals, e.g. for double on x86_64:

Failure: Test: tgamma_towardzero (-0x6.fe4638p+0)
Result:
 is:         -2.98383122695617016006e-02  -0x1.e8def3d7660a50000000p-6
 should be:  -2.98383122695616634367e-02  -0x1.e8def3d76609a0000000p-6
 difference:  3.81639164714897560770e-17   0x1.60000000000000000000p-55
 ulp       :  11.0000
 max.ulp   :  0.0000

-- 
You are receiving this mail because:
You are on the CC list for the bug.