SystemC Myth Busters
David Black, CTO
Charles Wilson, Member of Technical Staff
ESLX, Inc., USA

SystemC has been around since 1999, and many folks have used it with varying degrees of success. Over almost a decade, it is natural that a number of commonly accepted myths have been created. But are these myths true? This presentation looks into a number of these myths by performing directed experimentation. Here is a sample of beliefs that may be tested:

• Prefer SC_METHOD over SC_THREAD (context switching overhead is too high)
• Dynamic sensitivity is slow (prefer static)
• Dynamic processes are slow (avoid)
• QT-threads is faster than P-Threads (by how much?)
• STL is too slow for SystemC
• Fixed-point construction is too slow
• Conditionals cost a lot
• Debug slows simulation (gcc -g)

Source code for all tests will be made available for inspection and comment.