[PATCH] D25722: Improved cost model for FDIV and FSQRT

[Simon Pilgrim via llvm-commits]

RKSimon added a comment.

In https://reviews.llvm.org/D25722#578664, @mkuper wrote:

> I'd really like to understand the discrepancy between Agner's numbers and the IACA numbers - especially since IACA does give the expected number for the latency.
>  Andrew, any chance you could ping the IACA people at Intel and ask them about this?


IACA doesn't have that great support these days - the website says development has been suspended, although they have announced they will do a BDW/SKL bugfix release at some point. Similar comments about dodgy costs don't seem to be answered.


https://reviews.llvm.org/D25722