Reading "double" values from an Intel file

	Bottom Previous Topic Next Topic
Register To Post

thellier

Posted on: 2015/11/2 9:19 #1

Not too shy to talk

Hello

I am having some trouble to read a file coming from an PC application (so a binary with Intel ordering)
Some data are stored as "double" 64 bits but they dont seems to be the same as "double" in Motorola world

I have tested several byte-ordering without success ...
Any example code for converting that ??

Alain Thellier

Daytona675x

Re: Reading "double" values from an Intel file

Posted on: 2015/11/2 10:41 #2

Not too shy to talk

@thellier
It should be the reverse byte-order, just like it is for other types. If you tried that and if it doesn't work then you most likely simply got a typo in your shift / OR / AND orgy.
This should do:


#define ANTI_ENDIAN_64(a) \

(((a) << 56) | \

(((a)<<40) & 0x00FF000000000000ull) | \

(((a)<<24) & 0x0000FF0000000000ull) | \

(((a)<<8) & 0x000000FF00000000ull) | \

(((a)>>8) & 0x00000000FF000000ull) | \

(((a)>>24) & 0x0000000000FF0000ull) | \

(((a)>>40) & 0x000000000000FF00ull) | \

((a) >> 56))

Or use a union of double and 8 chars and simply swap char[x] and char[7-x] (x>=0, x<4).

Deniil

Re: Reading "double" values from an Intel file

Posted on: 2015/11/3 15:06 #3

Quite a regular

@thellier

Are you on 68k or PPC?

68k uses 80-bit floats internally. Not sure how the compilers implement double in this case. I think x86 also uses 80-bit floats, but PPC and probably x64 uses 64-bit floats.

Software developer for Amiga OS3 and OS4.
Develops for OnyxSoft and the Amiga using E and C and occasionally C++

Register To Post
	Top Previous Topic Next Topic

Currently Active Users Viewing This Thread: 1 ( 0 members and 1 Anonymous Users )