Has anyone any idea where I can get a double precision maths library which will work with the Keil C51 compiler.
We have found whilst doing a project that we need to do some calculations which require greater than 11 digits of accuracy. The normal float only supplies 7/8! We have done the calculations in Visual C++ but the only way to get them to work properly was to use double types. When the code is transported to C51 the inaccuracies are too great.
"We have found whilst doing a project that we need to do some calculations which require greater than 11 digits of accuracy." Now, how about the range requirements? If you don't need the range that having an exponent gives you, then a using 64-bit fixed-point format might do the trick.
Unfortunately, we need a range which includes numbers from 0.0005--- to 16,000,000^3 both positive and negative. We have looked into using 64bit integers but this requires an aweful lot of extra scaling which we feel would be easier to achieved in fp.
how lovely achieved in fp is that "fixed point" or "floating point" ambigous abbreviations hurt when you sit down. erik
I'd always assumed fp meant 'Floating Point' and nothing else!
I'd always assumed fp meant 'Floating Point' and nothing else! And post by the assupmption that someone reading this in Farawyistan makes the same assumption. Erik
We seem to have lost the thread! Double precision maths for the 8051 - any ideas?
"Double precision maths for the 8051 - any ideas?" I thought you had it when you said, "Many thanks Graham. That will do nicely with some slight modifications." in your previous thread?! http://www.keil.com/forum/docs/thread7167.asp
We have looked into using 64bit integers but this requires an aweful lot of extra scaling which we feel would be easier to achieved in fp. I really feel for you. you may actually have to work to get what you need. How horrible Erik