Freescale Semiconductor PXR4030, Reference Manual

Page: 1423 / 1434

Device Performance Optimization
PXR40 Microcontroller Reference Manual, Rev. 1
Freescale Semiconductor 33-11
performance of the application comes at the expense of larger code size. Compilers use a host of features,
such as loop unrolling, function inlining and application profile feedback to make the desired trade-offs
between enhanced performance and minimized code size.
The data in Figure 33-5 shows the effects of compiler optimization on a simple application. In this case,
the Dhrystone benchmark was run under three conditions:
• Optimized for small code size
• Optimized for high performance
• A Trade-off between code size and performance
Although this is an extreme example, it highlights how significant the role of the compiler and linker is in
determining the overall performance of an application.
Figure 33-5. Influence of Compiler Settings on Application Performance and Code Size
NOTE
Data measured using Dhrystone version 2.1 run on a Power Architecture
based Powertrain device using a standard commercial compiler.
The compiler optimizations do not necessarily have to be applied to the entire application. Analysis of an
application can identify time critical functions that may subsequently be targeted for performance
optimization, without incurring the impact of optimizing the entire application.
There are several other aspects of the compiler and linker that should be considered. In particular, the use
of Small Data Areas (SDAs, sometimes referred to as Special Data Areas) can make a significant
performance improvement. Refer to compiler documentation for usage guidelines on Small Data Areas.
Performance vs. Code Size
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Normalised Code Size
N
o
rm
ali
s
ed
E
xec
u
ti
o
n
T
im
e
Size Optimized
Trade-off
Speed Optimized