Atmel CAVR-4 Manual De Usuario

CAVR-4

Descriptions of options

Reference Guide

For additional information, see --cross_call_passes, page 174.

This option is related to the Optimization options in the C/C++ Compiler category in 
IAR Embedded Workbench.

--no_cse

--no_cse 

Use 

--no_cse

 to disable common subexpression elimination. 

At optimization levels 6 and 9, the compiler avoids calculating the same expression 
more than once. This optimization normally reduces both code size and execution time. 
However, the resulting code may be difficult to debug.

Note: This option has no effect at optimization levels below 6.

To set the equivalent option in IAR Embedded Workbench, select 
Project>Options>C/C++ Compiler>Code.

--no_inline

--no_inline 

Use 

--no_inline

 to disable function inlining. 

Function inlining means that a simple function, whose definition is known at compile 
time, is integrated into the body of its caller to eliminate the overhead of the call. 

This optimization, which is performed at optimization level 9, normally reduces 
execution time and increases code size. The resulting code may also be difficult to 
debug.

The compiler heuristically decides which functions to inline. Different heuristics are 
used when optimizing for speed.

Note: This option has no effect at optimization levels below 9.

To set the equivalent option in IAR Embedded Workbench, select 
Project>Options>C/C++ Compiler>Code.

--no_rampd

--no_rampd

Specifying this option makes the compiler use the 

RAMPZ

 register instead of 

RAMPD

. This 

option corresponds to the instructions 

LDS

 and 

STS

.

Note that this option is only useful on processor variants with more than 64 Kbytes data 
(

-v4

 and 

-v6

).