Date: Fri, 12 Apr 2002 12:39:24 -0400 (EDT) From: Garrett Wollman <wollman@lcs.mit.edu> To: "Tim J. Robbins" <tim@robbins.dropbear.id.au> Cc: freebsd-standards@FreeBSD.ORG Subject: Re: standards/36783 Message-ID: <200204121639.g3CGdOZ90234@khavrinen.lcs.mit.edu> In-Reply-To: <200204121240.g3CCe3a52899@freefall.freebsd.org> References: <200204121240.g3CCe3a52899@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
<<On Fri, 12 Apr 2002 05:40:03 -0700 (PDT), "Tim J. Robbins" <tim@robbins.dropbear.id.au> said: > This format string for long double is the same as that used for double. > <machine/float.h> says #define LDBL_DIG DBL_DIG , but it seems > odd to be printing out long doubles with no more precision than doubles. The exact type of a `long double' differs from machine to machine. On IA-32 machines, it's usually 80-bit ``extended double precision''. On SPARCv9, it's always ``quad precision'' (which is not an IEEE-sanctioned type, but is the obvious analogue of IEEE double with twice as many bits in the representation). Some other processors have other representations. Because of the way the IA-32 FPU is configured by default, using extended precision generally doesn't do any good, and I think in the current compilers a double and a long double are both implemented as FPU doubles. (This certainly was true at one time.) -GAWollman To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-standards" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200204121639.g3CGdOZ90234>