[LLVMdev] "Scheduling Hazards" on common architectures

orthochronous orthochronous at gmail.com
Wed Jul 7 17:53:02 PDT 2010


I'm not a compiler expert, so this is a basic questions. I'm trying to
get some new instruction scheduling code working, and I'm trying to do
things as simply as possible in the initial prototype. Looking through
the code it looks like only PPC architecture can return NoopHazard,
and I'm not interested in PPC. If in addition I'm preventing avoidable
stalls by different means, that means I can completely ignore the
HazardRecognizer and CurCycle (for now)?

Many thanks for any help,
David Steven Tweed

More information about the llvm-dev mailing list