Date: Tue, 21 Feb 2012 07:23:42 -0800 From: Steve Kargl <sgk@troutmask.apl.washington.edu> To: "Thomas D. Dean" <tomdean@speakeasy.org> Cc: freebsd-amd64@freebsd.org Subject: Re: Gcc46 and 128 Bit Floating Point Message-ID: <20120221152342.GA12463@troutmask.apl.washington.edu> In-Reply-To: <4F4351F0.2000608@speakeasy.org> References: <4F3EA37F.9010207@speakeasy.org> <CAGE5yCpvF0-b1iKAVGbya=fUNaYbGyrpj1PHSQxw4BvycNMLDg@mail.gmail.com> <4F3EC0B4.6050107@speakeasy.org> <20120221072053.GA10302@troutmask.apl.washington.edu> <4F4351F0.2000608@speakeasy.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Feb 21, 2012 at 12:12:32AM -0800, Thomas D. Dean wrote: > On 02/20/12 23:20, Steve Kargl wrote: > > >float --------> 24 bit significand, 32 bit size. > >double -------> 53 bit significand, 64 bit size. > >long double --> 53 bit significand, 80 bit size, i386 > >long double --> 64 bit significand, 80 bit size, x86_64 > >long double --> 113 bit significand, 128 bit size, sparc64 > >__float128 ---> 113 bit significand, 128 bit size, i386, x86_64 > > > > Nice, if you are not working on my platform. In my initial message, > > From dmesg: > CPU: Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz (4160.12-MHz K8-class CPU) > Origin = "GenuineIntel" Id = 0x206d6 Family = 6 Model = 2d > Stepping = 6 I saw your initial message, and yes, I noted that you are using an x86_64 class cpu. > And, I also said the __float128 is likely to leave me with some code > that is orphaned. As long as you use gcc 4.something or higher, you should have access to __float128. So, code won't be orphaned. Now, if you meant nonportable code, I agree. If you need more precision that C99's long double and portability, then use mpfr. -- Steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120221152342.GA12463>