[PATCH] D24593: Standford/Bubble sort code restructure

[Matthias Braun via llvm-commits]

MatzeB requested changes to this revision.
MatzeB added a comment.
This revision now requires changes to proceed.

We should refrain from changing the benchmarks unless there is a good reason. Undefined behavior would be a good reason, the compiler not being smart enough to catch some optimisation opportunities is not IMO. If we change it will no longer be the "Stanford" benchmark, there is also no reason to believe that code out in the wild doesn't look like the existing benchmark which the compiler should handle.

At a first glance I don't see any "flakiness" and the changes just look like manual loop peeling to me.


https://reviews.llvm.org/D24593