On Thu, Jan 24 2019 18:08:25 -0600, Scott Cheloha wrote:
> not to be standoffish, I'm just travelling so I can't really look into it.

yeah, it's not like I have infinite time either. I'll try to do
something with this over the weekend but no promises.

> if you came forward with malloc numbers before/after with sufficient
> N to get to a standard CI, that would help your case and be useful
> in the future.

It's been a while since I did any statistical analysis. To be honest,
this seems super overkill for measuring a code path that is not actually
taken most of the time; the test would have to compile both of the greps
under test with -DSMALL to avoid the mmap path. But fine, since it seems
so important to you I'll try to get *something*...

-- 
Lauri Tirkkonen | lotheac @ IRCnet

Reply via email to