Re: macOS metal rendering engine in mac port

[Alan Third]

On Thu, May 27, 2021 at 08:47:57PM +0300, Eli Zaretskii wrote:
> > Date: Thu, 27 May 2021 18:40:54 +0100
> > From: Alan Third <alan@idiocy.org>
> > Cc: aaronjensen@gmail.com, emacs-devel@gnu.org,
> >     mituharu@math.s.chiba-u.ac.jp
> > 
> > > > 3.7s 3.7s 4s
> > > 
> > > So I'm sure I understand: it takes you just 3 seconds to scroll
> > > through all of xdisp.c, top to bottom, with the command below?
> > > 
> > > (defun scroll-up-benchmark ()
> > >   (interactive)
> > >   (let ((oldgc gcs-done)
> > >         (oldtime (float-time)))
> > >     (condition-case nil (while t (scroll-up) (redisplay))
> > >       (error (message "GCs: %d Elapsed time: %f seconds"
> > >                       (- gcs-done oldgc) (- (float-time) oldtime))))))
> > 
> > Yes, although the initial run through, where the file is font-locked,
> > is more like 22 seconds. I didn't count that, am I measuring this wrongly?
> 
> Yes, I meant to measure just the first scroll, and restart Emacs each
> time before another measurement.
> 
> So maybe we were comparing apples and oranges all the time.

Possibly. I started off in this thread trying to measure the final
to the glass drawing time, so excluding the first run with the font
locking made sense, but I guess not so much for the question of what
happens with line numbers.

I suspect from Aaron's number's he's been doing the same as me. Unless
the M1 Macs really are that fast...
-- 
Alan Third