Date: Thu, 22 Jan 2004 19:42:52 +1100 From: Peter Jeremy <PeterJeremy@optushome.com.au> To: Thomas David Rivers <rivers@dignus.com> Cc: current@freebsd.org Subject: Re: historical stuff in math(3) Message-ID: <20040122084252.GG908@cirb503493.alcatel.com.au> In-Reply-To: <200401211141.i0LBfnp57845@lakes.dignus.com> References: <14141.1074670873@critter.freebsd.dk> <200401211141.i0LBfnp57845@lakes.dignus.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Jan 21, 2004 at 06:41:49AM -0500, Thomas David Rivers wrote: >> Isn't all the vax-D format related stuff math(3) getting pretty old ? > > It's very similar to the IBM mainframe format. So, a port > of FreeBSD to the IBM mainframe could still use it. (The VAX > format was just a copy of the IBM one with an extra precision bit > thrown in every now-and-then.) Not really. The IBM S/360 uses base-16 whereas virtually everyone else (including VAX) uses binary. The S/360 double precision format has a 14-digit (56-bit) fraction (no implicit digit), a fraction sign and a 7-bit signed exponent. The VAX-D documentation in math(3) is totally irrelevant to the S/360. Any serious math library would need significant re-work to handle the increased range and reduced/variable precision. Someone else mentioned the Alpha - VAX-format FP is specified in the architecture to simplify migration from the VAX. The early chips included it in hardware - do the recent chips still include it? > But - even the mainframe has an available/alternate IEEE format now, > and the mainframe version of gcc uses that... I think this must be new in the S/390. It's definitely not part of the S/360 or S/370 families and I don't believe it existed on the 30xx or 43xx families. Peter
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040122084252.GG908>