Dear All;
Is there any work that discusses LLVM framework’s short comings regarding auto vectorization and polyhedral optimizations?
Regards,
Hi,
did you look at the Polly project:
http://polly.llvm.org/
It performs polyhedral optimizations on LLVM-IR and allows the LLVM
vectorizer to vectorize more loops by:
- Creating alias free region with runtime-alias checks
- Annotating parallel loops
- Moving parallel dimensions to the innermost loop level
I added the Polly discussion list to the CC in case you want to know
more or start a discussion.
Best regards,
Johannes