James E Wilson wrote:
The only way to avoid double rounding is to have separate float, double, and long double arithmetic instructions, which unfortunately the x87 FP unit doesn't have.
And, to add some history - which Jim will know, doubtlessly - the m68k gcc port added the separate float and double arithmetic instructions as soon as they became available (in the Motorola 68040 processor).
-- Toon Moene - e-mail: toon@xxxxxxxxxxxxxxxxxxxx - phone: +31 346 214290 Saturnushof 14, 3738 XG Maartensdijk, The Netherlands Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html A maintainer of GNU Fortran 95: http://gcc.gnu.org/fortran/
