The sophisticated optimization technology of IAR Embedded Workbench for RL78 generates very compact code.
For this benchmark we took real customer applications and tested for size. The benchmark compares the total size of linked code and constants between code from IAR Embedded Workbench and Renesas CC.
In total for the tested applications, IAR Embedded Workbench generated 28 % smaller code than Renesas CCRL.
IAR Embedded Workbench for RL78 v2.21.1
Options: --core s3 --code_model near --data_model near --near_const_location rom0 -e -Ohz --double=32 --no_scheduling
Renesas CCRL v1.3.0
Options: -c -cpu=S3 -dbl_size=4 -dev="DR5F113GL.DVF" -g -Osize -Oalias=ansi -Osame_code
A number of customer applications covering different aspects of embedded software and representing were used in this benchmark. When compiled with IAR Embedded Workbench they ranged from 495 bytes to 43255 bytes. They were as follows:
|car_navig||Car navigation system|
|modeit||Embedded Internet Toolkit|
|Spirometer||Ventilation, movement of air measurement|
|decrypt||Functions to descramble CSS encrypted DVD content|
|bfix||Insert and extract bit fields|
|bittest||Boolean bit operations|
|float||Arithmetic float operations|