Re: The Emacs master is much slower than the emacs-27 branch.

[Eli Zaretskii]

> Date: Sat, 05 Dec 2020 11:20:01 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Sat, 05 Dec 2020 11:14:25 +0200
> > From: Eli Zaretskii <eliz@gnu.org>
> > Cc: acm@muc.de, stephen.berman@gmx.net, emacs-devel@gnu.org
> > 
> > I wasn't going to ask that, but I would like to point out one more
> > potential difference: I didn't compare two builds of master before and
> > after that commit.  Instead, I timed 32-bit builds of the current
> > emacs-27 and master branches which were both built with the following
> > optional switches to configure:
> > 
> >   --with-wide-int --with-modules --enable-checking=yes,glyphs 'CFLAGS=-O0 
> > -gdwarf-4 -g3'
> > 
> > Again, I don't expect any of this to matter so dramatically, so this
> > is just for completeness' sake.
> 
> And one more nit, again for completeness' sake: in both cases I timed
> the scrolling through the master branch version of xdisp.c, not each
> branch with its version of xdisp.c.

It occurred to me that the differences in the relative performance
could be due to a different GCC version and/or the effect of inlining
on 32-bit code vs 64-bit code produced by GCC.

In any case, Alan's original report was for optimized builds, and he
observed (although was later unable to reproduce) that Emacs 28 was
slower than Emacs 27.  So the important part of this is to compare the
performance of the optimized (-O2) builds and see if we have any
degradation in Emacs 28 in that case.