Using C-to-Silicon

Hi Marco,
 
The Cadence C-to-Silicon Compiler accepts most built-in and user-defined data types for C, C++, and SystemC.  The required coding styles can be summarized by one major rule that CtoS needs to be able to statically determine all the data structures at compiler time, thus you need to avoid the use of constructs such as dynamic memory and recursive function calls.   Pointers are fine to use as long as CtoS can determine what they will be pointing to.  We have a "Preparing Models for CtoS" Application Note that goes into further details.  Please contact your local Sales rep or AE.
 
One of the advantages of CtoS is that it has Cadence's RTL Compiler logic synthesis tool built-in, so CtoS is implementation-aware when determining how to generate optimal RTL for decisions like when to squeeze more logic into a clock cycle or whether or not to share large resources.  This helps allow CtoS to often match the QoR of hand-written RTL, plus makes for predictable timing closure with the down-stream logic synthesis.   However, this built-in RTL Compiler is not required for trying multiple micro-architectures.  We have a recommended methodology where you use CtoS to quickly try multiple micro-architectures, then once you find some good candidates, you run it through the flow with fine-grained timing enabled.
 
Hope this helps.
-mark

Dear mark Warren. thanks, your answer is useful for me.

Could you give me "Preparing Models for CtoS"?

Thnak in advance