Summary ======= Table generated by the tst-ieee754.tbl Perl script: LDBL FEM E NAN Z aarch64-linux-centos8 q 0 - - arm64-darwin-llvm14 - - - armv5tel-linux-zaurus - - - armv6l-linux-n810 - 0 - - hppa-hpux-hpcc q - - i386-freebsd6.2 e - - i686-linux-gcc-O0 e 2 - - - i686-linux-gcc-O2 E 2 + - - i686-linux-gcc-sse E 0 - - ia64-hpux-hpcc q - - ia64-linux-gcc-O0 E 0 - - ia64-linux-gcc-O2 e 0 - - ia64-linux-icc-O0 e 0 - - ia64-linux-icc-O1 e -1 1 - ia64-linux-icc-O2 e -1 1 N mips-linux-gcc-32 - 0 - - mips-linux-gcc-64 q 0 - - power8-aix-gcc - - - ppc-darwin8.11 d - - ppc-linux-deb-2008 d 0 - - ppc-linux-deb-2012 d 0 - - ppc64-linux-alma9 d 0 - - ppc64-linux-centos7 d 0 - - ppc64-linux-suse d 0 - - sparc-solaris-sunwspro5 q - - sparc-solaris-sunwspro6 q - - x86_64-linux-deb4-gcc E 0 - - x86_64-linux-deb4-opencc E 2 - - x86_64-linux-deb6-gcc e 0 - - x86_64-linux-debsid-lenny E 0 - - x86_64-linux-debsid-wheezy-gcc e 0 - - x86_64-linux-debsid-wheezy-tcc x 1 NZDU LDBL: "long double" arithmetic: -: IEEE-754 double precision (same as the "double" type); e: x87 extended precision (without the gcc bug on casts); E: ditto with bug https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36578; x: x87 extended precision with buggy macros of the double precision; d: double-double; q: IEEE-754R quadruple precision. FEM: value of FLT_EVAL_METHOD (ISO C99), when defined. E: when intermediate extended precision is used for the "double" type: -: values stored to a variable are converted into double precision; +: values are kept in extended precision (not conform to ISO C99). NAN: number of failing tests related to NaN (up to 8). Z: rounding modes for which -0 + -0 does not give -0. On MIPS, the GCC option -mabi=n32 gives the same results as -mabi=64 (tested on gcc23.fsffrance.org and gcc24.fsffrance.org under Debian 8 with GCC 4.9.2). $Id: Summary 161337 2023-09-07 13:10:08Z vinc17/cventin $