[PATCH] D31085: [InlineCost] Increase the cost of Switch

[Jun Bum Lim via Phabricator via llvm-commits]

junbuml added a comment.

While I'm working on the accurate version of case cluster calculation through the TTI hook in this change based on https://reviews.llvm.org/D31080, I also want to see if it make sense to apply different level of approximations depending on the number of case? So, for example, if the number of case is more than some threshold, we can use a very rough approximation (e.g, use the number case as number of cluster like https://reviews.llvm.org/D29870). Otherwise, we can use a somewhat accurate version of cluster calculation  based on https://reviews.llvm.org/D31080.


https://reviews.llvm.org/D31085