make check errors

["Mike Black" <mblack at csihq dot com>]

Seems that "make check" is failing on a good number of math tests with small differences.

This is on a PIII/600 with gcc version pgcc-2.95.2 19991024 (release)  Diff occurs at 16th significant digit.

 

more math/test-idouble.out

testing double (inline functions)
Failure: Test: Real part of: ctanh (-2 - 3 i) == -0.9653858790221331242 + 0.0098843750383224937 i
Result:
 is:         -9.65385879022133019944e-01  -0x1.ee470ed4d72e20000000p-1
 should be:  -9.65385879022133130967e-01  -0x1.ee470ed4d72e30000000p-1
 difference:  1.11022302462515654042e-16   0x1.00000000000000000000p-53
 ulp       :  1.0000
 max.ulp   :  0.0000
Maximal error of real part of: ctanh
 is      :  1.0000 ulp
 accepted:  0.0000 ulp
Maximal error of imaginary part of: ctanh
 is      :  1.0000 ulp
 accepted:  1.0000 ulp
Failure: Test: yn (3, 0.7) == -15.819479052819633505
Result:
 is:         -1.58194790528196396195e+01  -0x1.fa392c2273c1d0000000p+3
 should be:  -1.58194790528196342905e+01  -0x1.fa392c2273c1a0000000p+3
 difference:  5.32907051820075139403e-15   0x1.80000000000000000000p-48
 ulp       :  3.0000
 max.ulp   :  2.0000

 

Test suite completed:
  2413 test cases plus 2205 tests for exception flags executed.
  3 errors occured.

________________________________________
Michael D. Black   Principal Engineer
mblack@csihq.com  321-676-2923,x203
http://www.csihq.com  Computer Science Innovations
http://www.csihq.com/~mike  My home page
FAX 321-676-2355