[LLVMdev] GSoC 2011: Superoptimization for LLVM IR

[John Regehr]

> It worries me that you talk about "extending the number of
> transformations."  The point of a superoptimizer is that it has no
> specific model of transformations.  Rather, it is simply looking for cheap
> code fragments that are equivalent to expensive ones.

I could stand to be more clear...

Could you describe explictly the role of transformations in your 
superoptimizer framework?  And also give examples of what kinds of 
transformations you are talking about?

Could you describe how it is established that two code fragments are 
equivalent?

John