emacs-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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


From: Stephen Berman
Subject: Re: The Emacs master is much slower than the emacs-27 branch.
Date: Fri, 04 Dec 2020 11:14:36 +0100
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/28.0.50 (gnu/linux)

On Fri, 4 Dec 2020 09:53:40 +0000 Alan Mackenzie <acm@muc.de> wrote:

> Hello, Steve.
>
> On Fri, Dec 04, 2020 at 10:32:16 +0100, Stephen Berman wrote:
>> On Fri, 4 Dec 2020 09:04:49 +0000 Alan Mackenzie <acm@muc.de> wrote:
>> > On Thu, Dec 03, 2020 at 23:17:18 +0100, Stephen Berman wrote:
>> >> On Thu, 3 Dec 2020 18:15:30 +0000 Alan Mackenzie <acm@muc.de> wrote:
>
>> >> > Hello, Emacs.
>
>> >> > I've got pretty much up to date copies of both the master and the
>> >> > emacs-27 branches.
>
>> >> > When I run my customary benchmark (see below) on master's src/xdisp.c in
>> >> > master, it's taking around 34s.  On emacs-27, it's taking 22s.
>
>> >> Hm, this is odd: I just ran your benchmark on src/xdisp.c (starting
>> >> Emacs with -Q, both configured with CFLAGS='-Og -g3') and got vastly
>> >> different timings than you did:
>
>> >> master:   21.785817861557007
>> >> emacs-27: 61.3885281085968
>
>> > Maybe the emacs-27 is built without optimisation for debugging.
>
>> As noted above, I build both master and emacs-27 with CFLAGS='-Og -g3'.
>
> Apologies.  I hadn't read your post properly.
>
>> Given that, is such a difference still expected?  Eli wrote:
>> "In unoptimized builds I get almost no difference: Emacs 28 is about 4%
>> slower than Emacs 27", which also contrasts sharply with my timings.
>
> I can't understand why you're getting such a slowdown in your emacs-27.
> You've presumably got a modern machine, with plenty of cores and plenty
> of RAM, so it shouldn't be swapping, or some other process taking up the
> processor.

The CPU is an i7-8700 with 6 cores (12 threads) and 16 GB RAM.  I ran
the benchmarks twice in sequence, first using xdisp.c from master and
then using xdisp.c from emacs-27, with nearly identical results (with
the latter the timings were insignificantly faster, but the differences
between the runs from master and those from emacs-27 were the same).

> There's something very strange happening.

Indeed.

Steve Berman



reply via email to

[Prev in Thread] Current Thread [Next in Thread]