[PATCH] D21920: Fix for Bug 26903 adds support to inline __builtin_mempcpy

[Sunita_Marathe via llvm-commits]

Sunita_Marathe added a comment.

In http://reviews.llvm.org/D21920#472778, @majnemer wrote:

> On the other hand, we might want to turn memcpy + add into mempcpy as it might be a little faster/make live ranges a little smaller.


Hi David,

Thanks for participating in the review.

Are you suggesting that a memcpy in the IR followed at some point by a DST+N should be converted to a to a mempcpy call in the IR so as to reduce the live range of DST and N?   And then that mempcpy would be lowered as a memcpy followed by as DST+N as  per this change set?

Can you please clarify?


http://reviews.llvm.org/D21920