Re: [lmi] Currency class

[Vadim Zeitlin]

On Thu, 17 Sep 2020 21:24:31 +0000 Greg Chicares <gchicares@sbcglobal.net> 
wrote:

GC> This afternoon I've pushed a series of commits that carve the
GC> costly overhead into layers, in a way that I hope the commit
GC> messages make clear.

 Sorry, I still didn't have time to look at the first series of commits,
let alone the latter one.

GC> I would have thought that a function like
GC>   inline double identity(double d) {return d;}
GC> would have been optimized away, costing nothing.

 It really should be optimized away when optimizations are enabled.

GC> And I guess I'm a little surprised that multiplying by 100.0 * 0.01
GC> repeatedly costs so much.

 OK, I claimed that I would believe anything, but this one is really hard
to swallow, I'd really like to profile it to see what's going on...

VZ

pgpilMwsqDcyR.pgp
Description: PGP signature