[Ffmpeg-devel] int vs. float: Hard Numbers

Mean fixounet
Fri May 20 23:10:39 CEST 2005


Athlon64, 32 bits mode

integer_adder() (10 adds) returned 50, 16 cycles used
float_adder() (10 adds) returned 50.000000, 8 cycles used
integer_mult() (10 mults) returned 9765625, 16 cycles used
float_mult() (10 mults) returned 9765625.000000, 8 cycles used

Via C3 800 (ezra core)

integer_adder() (10 adds) returned 50, 39 cycles used
float_adder() (10 adds) returned 50.000000, 39 cycles used
integer_mult() (10 mults) returned 9765625, 131 cycles used
float_mult() (10 mults) returned 9765625.000000, 39 cycles used










More information about the ffmpeg-devel mailing list