mark@mips.COM (Mark G. Johnson) (01/02/89)
In article <325@loligo.fsu.edu> mccalpin@masig1.ocean.fsu.edu (John D. McCalpin) writes: > >What format is used for the IEEE 128-bit numbers? All I have read >is the original proposed draft standard, and I don't recall this >length. Specifically, what size exponent does the 128-bit format use? You're right; the IEEE standard only specifies a "double extended" precision which has >= 15 bits of exponent, >= 64 bits of fraction, and an exponent range of at least -16382..+16383. It's possible for both 80-bit *and* 128-bit floating point formats to fit the IEEE definition of "double extended" precision as long as they meet these requirements. And the MIPS architecture defines floating-point formats in both lengths: 80- and 128-bit. They are both IEEE-standard "double extended": total width 80 bits 128 bits exponent bits 15 bits 15 bits leading one explicit explicit fraction bits 64 bits 112 bits exponent range -16382..16383 -16382..16383 >There is some hesitancy in the supercomputer community to switch to the >IEEE format because the exponent range of 64-bit numbers is so much >smaller than the range currently provided by Cray and CDC/ETA formats. >The IEEE 64-bit allows a range of about 1.0e-308 to 1.0e+308, while the >Cray and CDC/ETA machines allow a range of about 1.0e-4000 to 1.0e+4000. > >I do not believe that the 80-bit format increases the exponent range. >It might help if the 128-bit format did allow this..... > Agreed; for normalized numbers, the IEEE double precision format has an exponent range of -1022..1023 (roughly, a range of 10^-308 to 10^308 for normalized numbers). However, IEEE double-extended precision has a range of at least -16382..16383 (roughly, 10^-4900 to 10^4900). This is true for all IEEE double-extendeds, in particular, for 80-bit formats and larger. So, in at least the category of ``range of numbers represented'', existing microprocessor chips that implement the IEEE double-extended precision in 80 bits (like the 8087, 68881, et cetera) match up nicely against Cray, CDC, and ETA. :-). -- -- Mark Johnson MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086 ...!decwrl!mips!mark (408) 991-0208