[PATCH] D27960: Use exact vector capacities to store DWARF line tables

[Mehdi AMINI via Phabricator via llvm-commits]

mehdi_amini added a comment.

I can believe that the dry-run could save a little bit of RSS, since it won't stress as much the memory allocator (but RSS is not necessarily the critical metric either here). 
All the remaining differences are in the noise? (Not sure if I got all the numbers right)


https://reviews.llvm.org/D27960