performance regression on sphereflake

[Renato Golin]

On 19 February 2014 00:31, Bob Wilson <bob.wilson at apple.com> wrote:
> The APCS stack alignment is only 32 bits. At the time I made that change, we were not able to dynamically realign stack frames for ARM, so it didn't make sense to have a preferred aggregate alignment larger than 32 bits. Even now, given the cost of dynamic realignment, I don't think changing that would be a good idea.

It looks like the regression Gerolf is seeing was actually a bug that
did well, rather than a feature that regressed. I'm inclined to agree
with Bob. I'd rather have correct behaviour than fast behaviour any
day. And this being a 15 years old ABI, I think we shouldn't fiddle
too much.

cheers,
-renato