Return to README CVS log

Up to [local] / OpenXM_contrib / gmp / mpfr

Annotation of OpenXM_contrib/gmp/mpfr/README, Revision 1.1.1.1

1.1       maekawa     1:
                      2:                         THE MPFR LIBRARY
                      3:
                      4:      ***** EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL *****
                      5:      ***** EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL *****
                      6:      ***** EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL EXPERIMENTAL *****
                      7:
                      8: This is a new function class, 'mpfr', for accurate floating-point arithmetic
                      9: with rounding control. Unlike the 'mpf' function class, 'mpfr' is a smooth
                     10: extension to the IEEE P754 arithmetic. In fact, setting the precision to 53,
                     11: 'mpfr' functions should work exactly like IEEE doubles.
                     12:
                     13: This code is *experimental* and it is not fully integrated into GMP. It is
                     14: also not as well tested as the other parts of GMP. The original 'mpfr' library
                     15: integrated into GMP was version 0.4.
                     16:
                     17: This code might at some point replace the 'mpf' library.
                     18:
                     19:
                     20:
                     21:                         MORE INFORMATION
                     22:
                     23: This code was contributed by the French FIABLE project. Please see the 'mpfr'
                     24: project page for more info, http://www.loria.fr/projets/mpfr/
                     25:
                     26:
                     27:
                     28:                           INSTALLATION
                     29:
                     30: To install this package to the libgmp library you add the "--enable-mpfr" to
                     31: configure. This has the unfortune side effect that all programs using GMP
                     32: will also depend on 'libm'.
                     33:
                     34: Two test cases were disabled, 'tcmp' 'tset_d'.
                     35:
                     36:
                     37:
                     38:                         REPORTING BUGS
                     39:
                     40: Report bugs to bug-gmp@gnu.org and mpfr@loria.fr.  What information is
                     41: needed in a good bug report is described in the manual.  The same address
                     42: can be used for suggesting modifications and enhancements.
                     43:
                     44:
                     45:
                     46:                 NOTES FROM THE ORIGINAL INSTALL DOCUMENT
                     47:
                     48: We tested the installation with GMP 2.0.2 on the following architectures:
                     49:
                     50:     Pentium II and III/Linux 2.2.10
                     51:     Alpha 21264/OSF1 V4.0
                     52:     SGI R10000/IRIX 6.5 (both using 32- and 64-bit limbs)
                     53:     Sun UltraSparc/Solaris 2.6
                     54:     HP 9000/HP-UX A.09.05
                     55:     Pentium/Windows with Cygnus compiler
                     56:
                     57: Known problems.
                     58: ===============
                     59:
                     60: 1) On some architectures (for example Alpha/OSF), the default 'make' command
                     61:    crashes in step 3. GNU make is preferred.
                     62:
                     63: 2) Under HP-UX and Pentium/Linux, don't use the -pedantic option, otherwise
                     64:    gcc complains with the longlong.h file from GMP.
                     65:
                     66: 3) On the Alpha, the test file for the square root (tsqrt) may produce
                     67:    errors as follows:
                     68:
                     69:      libm.a differs from mpfr_sqrt for a=9.89438396044940260000e-134, rnd_mode=2
                     70:      libm.a gives 3.14553397063986620000e-67, mpfr_sqrt gives 3.14553397063986680000e-67 (1 ulp)
                     71:      libm.a differs from mpfr_sqrt for a=7.86528588050363750000e+31, rnd_mode=1
                     72:      libm.a gives 8.86864469944739500000e+15, mpfr_sqrt gives 8.86864469944739400000e+15 (-1 ulp)
                     73:
                     74:    Also, Linux on Alpha is known to have problems with IEEE conformance. See
                     75:    http://www.linux.org.uk/VERSION/relnotes.2211.html.
                     76:
                     77:    In most cases, as above, the error comes from the libm.a. On Alpha/Linux,
                     78:    which is known to have a bad libm.a, you may even find errors in the
                     79:    addition or multiplication.
                     80:
                     81: