76448b8654
commit broke the 2**24 cases where |x| > DBL_MAX/2. There are exponent range problems not just for denormals (underflow) but for large values (overflow). Doubles have more than enough exponent range to avoid the problems, but I forgot to convert enough terms to double, so there was an x+x term which was sometimes evaluated in float precision. Unfortunately, this is a pessimization with some combinations of systems and compilers (it makes no difference on Athlon XP's, but on Athlon64's it gives a 5% pessimization with gcc-3.4 but not with gcc-3.3). Exlain the problem better in comments. |
||
|---|---|---|
| .. | ||
| bind | ||
| csu | ||
| libalias | ||
| libarchive | ||
| libatm | ||
| libautofs | ||
| libbegemot | ||
| libbluetooth | ||
| libbsnmp | ||
| libbz2 | ||
| libc | ||
| libc_r | ||
| libcalendar | ||
| libcam | ||
| libcom_err | ||
| libcompat | ||
| libcrypt | ||
| libdevinfo | ||
| libdevstat | ||
| libdisk | ||
| libedit | ||
| libexpat | ||
| libfetch | ||
| libform | ||
| libftpio | ||
| libgeom | ||
| libgpib | ||
| libgssapi | ||
| libio | ||
| libipsec | ||
| libipx | ||
| libkiconv | ||
| libkse | ||
| libkvm | ||
| libmagic | ||
| libmd | ||
| libmemstat | ||
| libmenu | ||
| libmilter | ||
| libmp | ||
| libncp | ||
| libncurses | ||
| libnetgraph | ||
| libngatm | ||
| libopie | ||
| libpam | ||
| libpanel | ||
| libpcap | ||
| libpmc | ||
| libpthread | ||
| libradius | ||
| librpcsvc | ||
| libsbuf | ||
| libsdp | ||
| libsm | ||
| libsmb | ||
| libsmdb | ||
| libsmutil | ||
| libstand | ||
| libtacplus | ||
| libtelnet | ||
| libthr | ||
| libthread_db | ||
| libufs | ||
| libugidfw | ||
| libusbhid | ||
| libutil | ||
| libvgl | ||
| libwrap | ||
| liby | ||
| libypclnt | ||
| libz | ||
| msun | ||
| ncurses | ||
| Makefile | ||
| Makefile.inc | ||