Microchip Technology XC8 Standard Compiler (Workstation) SW006021-1 SW006021-1 User Manual

®

DS52053B-page 268

 2012 Microchip Technology Inc.

The assembler performs optimizations on assembly code in addition to those 
optimizations performed by the code generator directly on the C code, see 
Section 5.13 “Optimizations”.

The xc8 driver by default instructs the assembler to optimizer assembly code that is 
generated from C code, but to not perform optimizations on hand-written assembly 
source modules. The latter code can be optimized if required, see 
Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. Assembly which is added 
in-line (see Section 5.12.2 “#asm, #endasm and asm()”) with C code is never 
optimized.

The optimizations that can be performed by the assembler include the following.

In-lining of small routines so that a call to that routine is not required. Only very 

small routines (typically a few instructions) will be changed so that code size 
is not impacted. This will speed code execution, without a significant 
increase in code size.

Procedural abstraction is performed on assembly code sequences that appear 

more than once. This is essentially a reverse in-lining process. The code se-
quence is abstracted into a callable routines will use a label, PLx, where x is 
a number, and a call to this routine will replace every instance of the original 
code sequence. This will reduce code size considerable, which a small 
impact on code speed.

Jump-to-jump type optimizations are made, primarily to tidy the output related to 

conditional code sequences that follow a generic template and which can 
leave jump instructions that jump to other jump instructions.

Unreachable code is removed. Code may become orphaned by other optimizations 

and cannot be reached during normal execution, e.g., instructions after a re-
turn instruction. The presence of any label is considered a possible entry 
point and code following a label is always considered reachable.

Peephole optimizations are performed on every instruction. These optimizations 

consider the state of execution at, and immediately around, each instruction 
– hence the name. They either alter or delete one or more instructions at 
each step. For example, if W is known to contain the value 0, and an instruc-
tion moves W to an address (MOVWF), this might be replaceable with a CLRF 
instruction.

Psect merging may be performed to allow other optimizations to take place. Code 

within the same psect is guaranteed to be located in the same program 
memory page, and so calls and jumps within the psect do not need to have 
the page selection bits set before executing. Code using the LJMP and 
FCALL

 instructions will benefit from this optimization, see 

Assembly optimizations can often interfere with debugging in tools such as MPLAB IDE 
and it may be necessary to disable them when debugging code, if possible, see 
Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. The assembler optimiza-
tions can drastically reduce code size, although typically have little effect on RAM 
usage.