[PATCH] D12778: Change the probability to check if a hot successor has a more important predecessor in block-placement.

[David Li via llvm-commits]

davidxl added a comment.

While the fix seems reasonable, in reality without profile data (or with statically predicted branch probabilities), the effect of this change will be hard to predict.

On the other hand, I think we should do a better job in computing the branch costs and locality cost when profile data is available and does not rely on this magic 80/20 ratio.


http://reviews.llvm.org/D12778