Do you have ever consider to implement Fixed Point (s31.32) in FF as an
alternative for real floating point. I understand that regular floating point
on a scamp3 not is posible (software bug).
With kindley regards
Jan
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
after asking John to implement floatingpoint on the scamp , the answer was it is not posible by a bug in the Microchip assembler. So I tought fixedpoint math is an alternitive!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I have not used the FF floating point at all. No need for me. I tried it once and I got strange results.
I just now realized that I was using degrees instead of radians.
So probably the FF floating point is fine to use. I tried sin and cos using radians.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I want to try to program the floatingpoints fix s16.16 first. For that, I need the following sign operators: q*, q/, q-, q+, and qnegate. Do you perhaps have the code for that?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Mikael,
Do you have ever consider to implement Fixed Point (s31.32) in FF as an
alternative for real floating point. I understand that regular floating point
on a scamp3 not is posible (software bug).
With kindley regards
Jan
No I have not considered implementing fixed point.
What bug are you referring to ? Do you mean that the FF floating point routines have a bug?
after asking John to implement floatingpoint on the scamp , the answer was it is not posible by a bug in the Microchip assembler. So I tought fixedpoint math is an alternitive!
I have not used the FF floating point at all. No need for me. I tried it once and I got strange results.
I just now realized that I was using degrees instead of radians.
So probably the FF floating point is fine to use. I tried sin and cos using radians.
that in mind, John could make a version for the scamp3 without any problems?
Probably not. It will together with his BSP use most of the flash of his 64 Kbyte PIC24 chips.
A chip with 128 Kbytes or more would be needed and a special linker file to locate the math library outside the areas used by flashforth.
Leaving out the trigonometric functions would make it fit better.
I want to try to program the floatingpoints fix s16.16 first. For that, I need the following sign operators: q*, q/, q-, q+, and qnegate. Do you perhaps have the code for that?
No I do not.