🚀 go-pugleaf

Jonathon Sivier (jsivier@ux1.cso.uiuc.edu) wrote:
: snow@castor.hahnemann.edu (Donald C. Snow) writes:
: the VAX uses Intel byte order, the first two chars should be 'II'.  The floating
: point is more difficult.  Not only is the byte order reversed, but thats easy to
: fix, also the floating point formats are different.  I tried to determine the
: format, but gave up and transfered the floating point data as ASCII, which I
: could read in and let the program convert to floating point.  I'm sure
: someone knows the formats and how to convert from one to the other, but this
: was the easiest way as far as I cxould see.

The floatting point format of the MIPS machines is the following:

31	30-23		22-0
63	62-52		51-0
---------------------------------
S	e		f
Sign	Exponent	Fraction


the filesds are:
- 1bit sign: s
- biased exponent: e=E+bias
- fraction: f=.b1b2...bp-1

The values (v) are calculated in the following way:

1. if E=Emax+1 and f#0, then v is a NAN
2. if E+Emax+1 and f=0, then v=(-1)^s \infinity
3. if Emin<=E<=Emax, then v=(-1)^s*2^E*(1.f)
4. if E=Emin-1 and f#0, then v=(-1)^s*2^(Emin)*(0.f)
5. if E=Emin-1 and f=0, then v=(-1)^s*0

f		24 	53
Emax		+127	+1023
Emin		-126	-1022
Exp bias	+127	+1023

The R4000 and R6000 have a fixed point format:

31	30-0
S	integer 
, which is in 2's complementary format.

I do not have data about the VAX numbers, perhaps You would like to
post to c.s.vax groups

Convert and enjoy.

-- Penio.