mark@mips.COM (Mark G. Johnson) (02/14/90)
Colin Plumb writes: > > ... i860 ... could lay claim to being ... 64-bit microprocessor. > It has a 64-bit external bus, and a fair number of 64-bit operations. > It still has 32-bit addresses and 32-bit integer registers, but the > floating point unit and the (to my mind, very wierd) graphics units > are 64 bits internally. I suspect that many folks believe you gotta have N bit integers and >=N bit addresses to legitimately be called an N bit processor. Fodder for thought-experiments: Intel 8086/7 had 16 bit external bus, 16-bit integers, and an 80-bit floating point unit. Yet it is considered to be a 16-bit machine. Addresses are 16 bits plus extra sophistication :-) due to segments. B-5000 SPARC chip (ECL from B.I.T.) has 64 bit external bus, a 64 bit floating point unit, 32 bit integers, and 32 bit addresses. Just like the i860. Yet SPARCs of today, B-5000 included, are typically called 32 bit machines. Intel 432 had an 80 bit floating point unit, 32 bit integers, and a 16 bit external bus. Addresses were >=40 bits. Four different numbers, quite fun :-). Typically it is called a 32 bit machine. Top of the line VAXes have >64 bit external busses, >64 bit floating point units, and 32 bit integers. Usually the VAX is called a 32 bit machine. Opinion: I speculate that the 8088 caused all of this confusion. It is a 16-bit machine which marketeers wrongly labelled "an 8 bit processor" and the misinformation stuck. Question: If a student did a MOSIS chip that implemented the PDP-8 with a doubleword (24-bit) interface at its pins, would that reclassify the PDP-8 as a 24 bit computer? Remember that the PDP-8 floating point included formats > 12 bits. -- -- Mark Johnson MIPS Computer Systems, 930 E. Arques, Sunnyvale, CA 94086 (408) 991-0208 mark@mips.com {or ...!decwrl!mips!mark}