This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
j0_downward failures on i686 with gcc 6.
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: GNU C Library <libc-alpha at sourceware dot org>
- Date: Tue, 1 Mar 2016 10:37:59 -0500
- Subject: j0_downward failures on i686 with gcc 6.
- Authentication-results: sourceware.org; auth=none
Joseph,
On i686 using GCC 6.0 I see some additional +1 ULP changes in j0.
Does this seem like normal compiler-induced changed in precision?
Like selecting slightly different instructions that produce a slightly
different result? Might it be OK to adjust this up by an ULP to avoid
the test failure or do we need to know exactly why this changed?
The machine in question is an Intel E5540, so nothing odd there.
testing float (without inline functions)
Failure: Test: j0_downward (0xap+0)
Result:
is: -2.45935813e-01 -0x1.f7ad32p-3
should be: -2.45935768e-01 -0x1.f7ad2cp-3
difference: 4.47034835e-08 0x1.800000p-25
ulp : 3.0000
max.ulp : 2.0000
Maximal error of `j0_downward'
is : 3 ulp
accepted: 2 ulp
Test suite completed:
58942 test cases plus 53146 tests for exception flags and
53146 tests for errno executed.
2 errors occurred.
testing float (finite-math-only)
Failure: Test: j0_downward (0xap+0)
Result:
is: -2.45935813e-01 -0x1.f7ad32p-3
should be: -2.45935768e-01 -0x1.f7ad2cp-3
difference: 4.47034835e-08 0x1.800000p-25
ulp : 3.0000
max.ulp : 2.0000
Maximal error of `j0_downward'
is : 3 ulp
accepted: 2 ulp
Test suite completed:
46382 test cases plus 41098 tests for exception flags and
0 tests for errno executed.
2 errors occurred.
testing float (inline functions)
Failure: Test: j0_downward (0xap+0)
Result:
is: -2.45935813e-01 -0x1.f7ad32p-3
should be: -2.45935768e-01 -0x1.f7ad2cp-3
difference: 4.47034835e-08 0x1.800000p-25
ulp : 3.0000
max.ulp : 2.0000
Maximal error of `j0_downward'
is : 3 ulp
accepted: 2 ulp
Test suite completed:
58518 test cases plus 0 tests for exception flags and
0 tests for errno executed.
2 errors occurred.
Cheers,
Carlos.