Re: take this famous emacs benchmark test

Re: take this famous emacs benchmark test

[Lute Kamstra]

Dan Jacobson <jidanni@dman.ddts.net> writes:

> Fellas, take my nifto emacs benchmark test, which I pegged together
> just moments ago without proper computer training...
>
>     j=;for i in '' -nw -q --no-site-file;do j=$i\ $j;c="time emacs $j -f
>     kill-emacs";echo -e \\n\$ $c;$c;done
>
> ...to see just how much slower your emacs plods along due to your
> and your "Linux distribution" crud-encrusting it. Just cut and paste it
> into your shell to-day! Results here at Nerdsburg Laboratories:
>
> $ time emacs -f kill-emacs
> 0.97user 0.13system 0:01.30elapsed 84%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (1632major+952minor)pagefaults 0swaps
>
> $ time emacs -nw -f kill-emacs
> 0.80user 0.12system 0:00.93elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (1305major+823minor)pagefaults 0swaps
>
> $ time emacs -q -nw -f kill-emacs
> 0.44user 0.07system 0:00.53elapsed 96%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (1296major+675minor)pagefaults 0swaps
>
> $ time emacs --no-site-file -q -nw -f kill-emacs
> 0.07user 0.03system 0:00.09elapsed 105%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (1056major+311minor)pagefaults 0swaps

I'm curious.  What happens when you do these four tests in reverse order?

[...]

Lute.

-- 
(spook) => "Skipjack 9705 Samford Road propaganda"
(insert-file-contents "~/.signature") => (error "`~/.signature' too rude")

Re: take this famous emacs benchmark test

[Dan Jacobson]

L> I'm curious.  What happens when you do these four tests in reverse order?

About the same. Any special reason not?
$ sh /tmp/tt
0.06user
0.47user
0.75user
0.96user
$ cat /tmp/tt
set -- --no-site-file -q -nw ''; t=$$$$
for i; do /usr/bin/time -a -o /tmp/$t emacs $@ -f kill-emacs; shift; done
awk '/^[0-9]\./{print $1}' /tmp/$t; rm /tmp/$t

Re: take this famous emacs benchmark test

[Thomas A. Horsley]

Speaking of remarkable emacs results, I don't know if it is the DDR memory
or the RAID-0 disk or the Altlon XP chip (or some combination), but my copy
of emacs for windows starts up faster than notepad on my fastest PC.  By the
time I get my finger off the mouse button, it is already up and finished
loading all my init code.

I spent several hours trying to figure out what was wrong with my emacs
installation on that machine the first time I tried it :-).
--
>>==>> The *Best* political site <URL:http://www.vote-smart.org/> >>==+
      email: Tom.Horsley@worldnet.att.net icbm: Delray Beach, FL      |
<URL:http://home.att.net/~Tom.Horsley> Free Software and Politics <<==+

take this famous emacs benchmark test

[Dan Jacobson]

Fellas, take my nifto emacs benchmark test, which I pegged together
just moments ago without proper computer training...

    j=;for i in '' -nw -q --no-site-file;do j=$i\ $j;c="time emacs $j -f
    kill-emacs";echo -e \\n\$ $c;$c;done

...to see just how much slower your emacs plods along due to your
and your "Linux distribution" crud-encrusting it. Just cut and paste it
into your shell to-day! Results here at Nerdsburg Laboratories:

$ time emacs -f kill-emacs
0.97user 0.13system 0:01.30elapsed 84%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1632major+952minor)pagefaults 0swaps

$ time emacs -nw -f kill-emacs
0.80user 0.12system 0:00.93elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1305major+823minor)pagefaults 0swaps

$ time emacs -q -nw -f kill-emacs
0.44user 0.07system 0:00.53elapsed 96%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1296major+675minor)pagefaults 0swaps

$ time emacs --no-site-file -q -nw -f kill-emacs
0.07user 0.03system 0:00.09elapsed 105%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1056major+311minor)pagefaults 0swaps

Wow, 105% CPU. I.e. with all the junk stripped, it runs faster than
the machine, eh?  Bet that's how Stallman runs it, yup, raw. Like them
defaults, man.  Naw... [Here we go again] Say, has anybody seen his
.emacs file?  Do a wc(1) on it while he's not looking.  Better yet,
post it(s URL) so we can examine it.  I wonder what shape it is in.

Back to the results above. I won't go over the details [because I have
um no idea what...] but it looks like the 0.97 0.80 ... does the
talking. Here's an additional result we should add, 

$ /usr/bin/time emacs --no-site-file -q -f kill-emacs
0.23user 0.02system 0:00.43elapsed 57%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (1412major+599minor)pagefaults 0swaps
-- 
http://jidanni.org/ Taiwan(04)25854780

Re: take this famous emacs benchmark test

[Thien-Thi Nguyen]

Dan Jacobson <jidanni@dman.ddts.net> writes:

   Say, has anybody seen his
   .emacs file?  Do a wc(1) on it while he's not looking.  Better yet,
   post it(s URL) so we can examine it.  I wonder what shape it is in.

last time i looked (~1994) it was not as sparse as i had expected.
that was on a static gnu machine, no idea what his portable sports.

no worries, someone will eventually post instructions on how to do M-x
dump-custom-emacs, at which point you can add that as a write-file hook
local to your ~/.emacs (of course, debugging that remotely would be more
difficult but that's another question).

thi