home *** CD-ROM | disk | FTP | other *** search
- CHANGES IN FIXFLOAT LIBRARY 1.1 -> 1.2
-
- 1 - Bugfix in ffcos routine.
-
- 2 - Better performance of signed multiply by IMUL & IDIV
- instructions instead of sign analysis.
-
- 3 - Inline assembly versions of ffmul, ffsmul, ffsdiv, ffdiv have been
- added to remove CALL / RET instructions when multiplying/dividing
- from C/C++.
-
- The following Assembler macros have been added to make provide the
- same effect in asm files :
-
- MFFMUL, MFFSMUL, MFFDIV, MFFSDIV
-
- 4 - The divide functions ( not macros or inline asm functions ) have been
- made foolproof so that they cannot overflow 32 bits and generate
- exceptions. Instead largest possible positive or negative number is
- returned.
-
- 5 - A function to produce the argument of a vector is provided.
-
- 6 - A function to solve 2:nd degree polynomials is added. Works with
- 64 bit intermediate results.
-
- 7 - A function to approximate 'sqrt( a*a + b*b )' which executes in approx
- 18 clocks has been added.
-
- 8 - Documentation have been extended.
-
- 9 - An extensive C example demonstrating the functioning of approx 80%
- of the provided functions is included.
-
-
-
-