Am 03.09.21 um 12:52 schrieb Павел Бельтюков: > > What would be the impact of different calling conventions in case of > fixed point operations and fix16 (Q16.16) data format? >> See https://github.com/PetteriAimonen/libfixmath > <https://github.com/PetteriAimonen/libfixmath> for example...
None. That library uses a 32-bit integer type as fixed-point type, with functions using that as parameters and return type. Both the old and the new convention pass all 32-bit parameters on the stack. The new and old convention use the same registers to return 32-bit values. Both the new and old convention use caller-cleanup of stack parameters for 32-bit integer return values. Philipp _______________________________________________ Sdcc-user mailing list Sdcc-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sdcc-user
