There was a change to the floating point compare function called by the C51 compiler called ?C?FPCMP made between v9.01 and v9.02a. According to Keil, this change addressed rounding issues with the previous version. According to Keil, this change was internal and not subject to disclosure on the 9.02a release notes.
Does anyone know what specific problems this addresses when performing a float compare of (a < b)?
this really is an issue for Keil, not a Keil user i.e. forum participant. Ask Keil directly
Erik
I have. They are unwilling to produce test cases demonstrating the rounding problem in v9.01 that is fixed in v9.02a. I am soliciting the community for information.