* 21.3 smokes 21.2
@ 2003-03-01 6:44 Matt Swift
2003-03-01 20:43 ` Stefan Monnier
0 siblings, 1 reply; 3+ messages in thread
From: Matt Swift @ 2003-03-01 6:44 UTC (permalink / raw)
21.3 is three to four times faster than 21.2 when loading files at
startup.
I just made objective measurements and confirmed a subjective
impression that 21.3 smokes 21.2 at startup. In summary, it is three
times faster at loading the same disk files, and subtract 5 more
seconds for not having to load Mule-ucs (this bumps to ratio to four
times faster for me).
These results not controlled enough to be a reliable benchmark. The
compilation of 21.2 may have been done with gcc-2.95 vs gcc-3.2, and
of the approx 100 files loaded, a handful are not the same in each
case. But whatever the cause, it sure is a delight.
(The output below is generated by code I wrote to measure and
summarize the time it takes to `require' files at startup (I will put
it on gnu.emacs.sources when it is more polished.)
21.2 21.3
un-define 4.37s
..mule-uni 0.00s
..mucs-error 0.00s
..mucs-type 0.00s
....trans-util 0.00s
mucs 0.24s
..mucs-ccl 0.22s
....ccl 0.21s
tbl-mg 0.00s
un-tools 0.00s
..un-data 0.00s
bbdb-autoloads 0.01s bbdb-autoloads 0.05s
..wid-edit 0.26s ..wid-edit 0.02s
..cus-load 0.29s ..cus-load 0.02s
cus-edit 0.90s cus-edit 0.19s
..cus-start 0.22s ..cus-start 0.06s
edb-fixes-1\.21 0.03s edb-fixes-1\.21 0.00s
eldav 0.06s eldav 0.02s
..cl 0.04s ..cl 0.01s
....apel-ver 0.00s ....apel-ver 0.00s
alist 0.35s alist 0.12s
..product 0.34s ..product 0.12s
....pym 0.07s ....pym 0.11s
......static 0.00s ......static 0.00s
mailcrypt-init 0.00s mailcrypt-init 0.00s
mmm-auto 0.06s mmm-auto 0.02s
..mmm-vars 0.06s ..mmm-vars 0.02s
....mmm-compat 0.04s ....mmm-compat 0.01s
......regexp-opt 0.04s ......regexp-opt 0.00s
....working 0.00s ....working 0.00s
....assoc 0.05s ....assoc 0.00s
..semantic 0.11s ..semantic 0.14s
....semantic-util 0.03s ....semantic-util 0.11s
......semantic-util-modes 0.01s ......semantic-util-modes 0.01s
..eieio 0.28s ..eieio 0.06s
..semantic-load 0.00s ..semantic-load 0.00s
..tree-buffer 0.27s ..tree-buffer 0.10s
....avoid 0.05s ....avoid 0.00s
....ecb-util 0.00s ....ecb-util 0.01s
....ecb-compilation 0.00s ....ecb-compilation 0.00s
....ecb-create-layout 0.01s ....ecb-create-layout 0.01s
......ecb-mode-line 0.00s ......ecb-mode-line 0.00s
..ecb-layout 0.11s ..ecb-layout 0.08s
....compile 0.06s ....compile 0.02s
..ecb-help 0.00s ..ecb-help 0.00s
..ecb-navigate 0.00s ..ecb-navigate 0.01s
..ecb-eshell 0.22s ..ecb-eshell 0.00s
..ecb-cycle 0.00s ..ecb-cycle 0.00s
..ecb-face 0.02s ..ecb-face 0.10s
..ecb-upgrade 0.01s ..ecb-upgrade 0.01s
ecb 1.14s ecb 0.60s
..ecb-tod 0.00s ..ecb-tod 0.00s
..poe 0.40s **poe 0.28s
....nofeature 0.05s **..nofeature 0.00s
....bytecomp 0.04s
..pcustom 0.01s ..pcustom 0.00s
......pces 0.05s ......pces 0.10s
........pces-e20 0.04s ........pces-e20 0.10s
..........pces-20 0.03s ..........pces-20 0.09s
............broken 0.01s ............broken 0.00s
bitmap 0.62s bitmap 0.60s
..bitmap-ci 0.20s ..bitmap-ci 0.31s
....poem 0.13s ....poem 0.12s
......poem-e20 0.07s ......poem-e20 0.01s
........poem-e20_3 0.06s ........poem-e20_3 0.00s
dired 0.07s **......dired 0.19s
install 0.02s install 0.04s
..path-util 0.01s ..path-util 0.02s
initz-vars 0.25s initz-vars 0.01s
initz-globals 0.02s initz-globals 0.01s
initz-util 0.01s initz-util 0.00s
..find-func 0.32s ..find-func 0.02s
....loadhist 0.05s ....loadhist 0.01s
swift-extensions 0.33s swift-extensions 0.03s
..merge-list 0.00s ..merge-list 0.00s
..comint 0.34s ..comint 0.13s
....ring 0.04s ....ring 0.01s
..rfc822 0.04s ..rfc822 0.01s
....uniquify 0.29s ....uniquify 0.03s
....saveplace 0.04s ....saveplace 0.11s
....recentf 0.06s ....recentf 0.02s
....mwheel 0.28s ....mwheel 0.01s
......syntax 0.01s
....font-lock 0.41s ....font-lock 0.18s
......jit-lock 0.32s ......jit-lock 0.02s
....image-file 0.05s ....image-file 0.04s
......image 0.01s
....jka-compr 0.06s ....jka-compr 0.17s
....require-vulnerable 0.01s ....require-vulnerable 0.01s
....srx 0.01s ....srx 0.00s
......webster 0.01s ......webster 0.00s
....alz 0.27s ....alz 0.01s
......oed-netscape 0.01s ......oed-netscape 0.00s
.........ls-lisp 0.01s
..........w32-symlinks 0.01s
....swift-dired 1.30s ....swift-dired 0.30s
......dired-x 1.28s ......dired-x 0.29s
........dired-aux 0.75s ........dired-aux 0.05s
....outline-x 0.57s ....outline-x 0.18s
......outline 0.55s ......outline 0.17s
....pp* 0.01s ....pp* 0.00s
....python-x 0.25s ....python-x 0.00s
....lisp-mnt 0.59s ....lisp-mnt 0.04s
....generic-x 2.64s ....generic-x 0.69s
......generic 0.52s ......generic 0.03s
....ange-ftp 0.86s ....ange-ftp 0.07s
......format-spec 0.02s ......format-spec 0.00s
......shell 0.54s ......shell 0.15s
......tramp-ftp 0.03s ......tramp-ftp 0.01s
....tramp 0.95s ....tramp 0.38s
......tramp-smb 0.02s ......tramp-smb 0.12s
..dotemacs 9.77s ..dotemacs 2.94s
....tex-site 0.05s ....tex-site 0.16s
personal-elisp 11.12s personal-elisp 3.30s
..browse-url-opera 0.88s ..browse-url-opera 0.05s
....browse-url 0.86s ....browse-url 0.04s
0 other features loaded in < 0.0s each 0.00s 0 other features loaded in < 0.0s each 0.00s
111 features TOTAL 19.7 s 105 features TOTAL 5.1 s
It takes about 5 seconds to load Mule-ucs
in Emacs 21.2, so the figures for file-loading are
about 3:1 and for starting up overall, 4:1.
lines with ** appear out of order for the sake of comparison
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: 21.3 smokes 21.2
2003-03-01 6:44 21.3 smokes 21.2 Matt Swift
@ 2003-03-01 20:43 ` Stefan Monnier
2003-03-03 16:19 ` Sean Neakums
0 siblings, 1 reply; 3+ messages in thread
From: Stefan Monnier @ 2003-03-01 20:43 UTC (permalink / raw)
Cc: emacs-devel
> 21.3 is three to four times faster than 21.2 when loading files at
> startup.
Glad to hear that, but I'm wondering if anybody has an
idea why that is. I can't think of any change between 21.2 and
21.3 that would justify such a change.
Stefan
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: 21.3 smokes 21.2
2003-03-01 20:43 ` Stefan Monnier
@ 2003-03-03 16:19 ` Sean Neakums
0 siblings, 0 replies; 3+ messages in thread
From: Sean Neakums @ 2003-03-03 16:19 UTC (permalink / raw)
commence Stefan Monnier quotation:
>> 21.3 is three to four times faster than 21.2 when loading files at
>> startup.
>
> Glad to hear that, but I'm wondering if anybody has an
> idea why that is. I can't think of any change between 21.2 and
> 21.3 that would justify such a change.
Just a thought: Matt doesn't say much about the conditions under which
the test was done. For example, if the machine was freshly booted,
and the 21.2 test run was done before the 21.3 run, the 21.3 run would
be reading straight from the filesystem cache rather than having to go
to disk as the 21.2 run would have had to do.
Further: I've been using a build of 21.3 for a few weeks myself, and I
can't say I've noticed a dramatic reduction in startup times.
--
/ |
[|] Sean Neakums | Size *does* matter.
[|] <sneakums@zork.net> | That's why I use Emacs.
\ |
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-03-03 16:19 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-03-01 6:44 21.3 smokes 21.2 Matt Swift
2003-03-01 20:43 ` Stefan Monnier
2003-03-03 16:19 ` Sean Neakums
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/emacs.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).