Abstract:
In the recent years, communication with embedded computers has become 
omnipresent. Mobile phones and personal digital assistants (PDAs) are 
a part of our business and personnel everyday life. Protecting contents 
with cryptographic primitives is inevitable and demands for highly 
efficient cryptographic applications in the sense of time and power 
consumption. Some cryptosystems allow for high efficiency assuming a 
thorough done implementation. Optimizing implementations on embedded 
devices often implies the use of the processor language (assembly). 
However, current compilers for the C programming language can often 
also produce highly optimized code regarding code size and/ or 
execution time.
In this contribution, we analyze the generated code of different 
compilers for embedded processors. The time and space consumption 
of the code - a modern cryptographic implementation, namely a 
hyperelliptic curve cryptosystem - determines We provide an accurate 
comparison of several important compilers (DIAB 5.1.2, GNU 3.3, 
MetaWare 4.5, ARMCC 2.0 and CodeWarrior 2.0) for embedded processors.

BibTeX:
@InProceedings{IP-BPPWW04,
author = {E. Barteska and J. Pelzl and C. Paar and V. Wittelsberger and T. Wollinger},
title = "{Case Study: Compiler Comparison for an Embedded Cryptographical Application}",
booktitle = {The 2004 International Conference on Embedded Systems and Applications - ESA, June 21-24, 2004, Las Vegas, USA},
year = {2004},
month = {June}
}