From: Vibhav Pant <vibhavp@gmail.com>
To: "Clément Pit-Claudel" <cpitclaudel@gmail.com>
Cc: Stefan Monnier <monnier@iro.umontreal.ca>,
"emacs-devel@gnu.org" <emacs-devel@gnu.org>
Subject: Re: [PATCH]: Add new bytecode op `switch' for implementing branch tables.
Date: Wed, 8 Feb 2017 19:08:12 +0530 [thread overview]
Message-ID: <CA+T2Sh1Vg+WAyUp3sTq=NYZVd6Bo_m09cNHHwEQMn4LKwi8R5A@mail.gmail.com> (raw)
In-Reply-To: <1b07c68a-873e-83c8-246d-423bc83a3881@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 1209 bytes --]
On Tue, Feb 7, 2017 at 9:26 PM, Clément Pit-Claudel
<cpitclaudel@gmail.com> wrote:
> The timings fluctuate quite a bit, but the byte-switch branch seems to be about 5-7% slower. Hopefully linear-scan hash tables will make things much faster :)
The following patch makes hash_lookup use linear search when the number of keys
in the hash table is <= 5 (chosen arbitrarily). switch bytecode run with this
patch takes 15.96 seconds to run the benchmark, while the goto-if-nil code takes
17.15 seconds.
--
Vibhav Pant
vibhavp@gmail.com
diff --git a/src/fns.c b/src/fns.c
index ac7c1f265a..2940bfdfbb 100644
--- a/src/fns.c
+++ b/src/fns.c
@@ -3915,6 +3915,17 @@ hash_lookup (struct Lisp_Hash_Table *h,
Lisp_Object key, EMACS_UINT *hash)
ptrdiff_t start_of_bucket;
Lisp_Object idx;
+ if (HASH_TABLE_SIZE (h) <= 5 && h->test.cmpfn) {
+ /*Try doing a linear search first */
+ ptrdiff_t i;
+ for (i = 0; i < HASH_TABLE_SIZE (h); i++)
+ {
+ if (h->test.cmpfn (&h->test, key, HASH_KEY (h, i)))
+ return i;
+ }
+ return -1;
+ }
+
hash_code = h->test.hashfn (&h->test, key);
eassert ((hash_code & ~INTMASK) == 0);
if (hash)
[-- Attachment #2: gethash-linear-search.diff --]
[-- Type: text/plain, Size: 637 bytes --]
diff --git a/src/fns.c b/src/fns.c
index ac7c1f265a..2940bfdfbb 100644
--- a/src/fns.c
+++ b/src/fns.c
@@ -3915,6 +3915,17 @@ hash_lookup (struct Lisp_Hash_Table *h, Lisp_Object key, EMACS_UINT *hash)
ptrdiff_t start_of_bucket;
Lisp_Object idx;
+ if (HASH_TABLE_SIZE (h) <= 5 && h->test.cmpfn) {
+ /*Try doing a linear search first */
+ ptrdiff_t i;
+ for (i = 0; i < HASH_TABLE_SIZE (h); i++)
+ {
+ if (h->test.cmpfn (&h->test, key, HASH_KEY (h, i)))
+ return i;
+ }
+ return -1;
+ }
+
hash_code = h->test.hashfn (&h->test, key);
eassert ((hash_code & ~INTMASK) == 0);
if (hash)
next prev parent reply other threads:[~2017-02-08 13:38 UTC|newest]
Thread overview: 53+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-06 17:50 [PATCH]: Add new bytecode op `switch' for implementing branch tables Vibhav Pant
2017-02-06 18:30 ` Stefan Monnier
2017-02-07 8:45 ` Vibhav Pant
2017-02-07 14:41 ` Stefan Monnier
[not found] ` <CA+T2Sh3N09WGoHtNL3BbptsKA1ZdwWedTNDd0vmeBe0fTO5a1g@mail.gmail.com>
[not found] ` <CA+T2Sh2sa6=WOvyePzL_ACR0Nw=jTnZze_xXFcvL=22OURP=ZA@mail.gmail.com>
2017-02-07 15:21 ` Vibhav Pant
2017-02-07 13:50 ` Vibhav Pant
2017-02-07 15:56 ` Clément Pit-Claudel
2017-02-08 13:38 ` Vibhav Pant [this message]
2017-02-08 16:21 ` Vibhav Pant
2017-02-08 17:46 ` Vibhav Pant
2017-02-09 13:21 ` Stefan Monnier
2017-02-08 17:56 ` Eli Zaretskii
2017-02-23 4:35 ` Compiled code in Emacs-26 will fail in Emacs-25 if use pcase [was: Add new bytecode op `switch' for implementing branch tables.] Tino Calancha
2017-02-23 8:36 ` Richard Copley
2017-02-23 10:12 ` Tino Calancha
2022-07-04 12:24 ` Vibhav Pant
2017-02-23 13:39 ` Stefan Monnier
2017-02-23 14:33 ` Kaushal Modi
2017-02-23 15:10 ` Stefan Monnier
2022-07-02 20:31 ` Robert Weiner
2022-07-02 21:28 ` Stefan Monnier
[not found] ` <CA+T2Sh29UhuNrhRZG=PPQbYYHtONXwyb8dX4rBVLmwdORLELhg@mail.gmail.com>
2017-02-23 17:50 ` Vibhav Pant
2017-04-11 17:40 ` Compiled code in Emacs-26 will fail in Emacs-25 if use pcase Lars Brinkhoff
2017-02-23 16:15 ` Compiled code in Emacs-26 will fail in Emacs-25 if use pcase [was: Add new bytecode op `switch' for implementing branch tables.] raman
2017-02-23 17:24 ` Drew Adams
2017-02-23 17:50 ` T.V Raman
2017-02-23 18:02 ` Drew Adams
2017-02-23 18:24 ` T.V Raman
2017-02-23 19:00 ` Drew Adams
2017-02-24 2:06 ` defsubst VS defun or defmacro [was RE: Compiled code in Emacs-26 will fail in Emacs-25 if use pcase] Tino Calancha
2017-02-24 3:14 ` Eric Abrahamsen
2017-02-24 3:56 ` raman
2017-02-24 4:20 ` Eric Abrahamsen
2017-02-24 3:51 ` raman
2017-02-09 17:32 ` [PATCH]: Add new bytecode op `switch' for implementing branch tables Clément Pit-Claudel
2017-02-09 19:15 ` Clément Pit-Claudel
2017-02-10 4:12 ` Vibhav Pant
2017-02-10 4:17 ` Clément Pit-Claudel
2017-02-10 5:03 ` Vibhav Pant
2017-02-10 6:07 ` Stefan Monnier
2017-02-10 13:51 ` Vibhav Pant
2017-02-10 15:12 ` Stefan Monnier
2017-02-10 17:59 ` Vibhav Pant
2017-02-10 18:25 ` Vibhav Pant
2017-02-10 20:47 ` Paul Eggert
2017-02-10 20:58 ` Stefan Monnier
2017-02-11 15:07 ` Vibhav Pant
2017-02-12 3:10 ` Vibhav Pant
2017-02-13 7:18 ` Vibhav Pant
2017-02-09 16:37 ` Vibhav Pant
2017-02-06 19:32 ` Eli Zaretskii
2017-02-07 14:26 ` Vibhav Pant
2017-02-07 16:14 ` Eli Zaretskii
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: https://www.gnu.org/software/emacs/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CA+T2Sh1Vg+WAyUp3sTq=NYZVd6Bo_m09cNHHwEQMn4LKwi8R5A@mail.gmail.com' \
--to=vibhavp@gmail.com \
--cc=cpitclaudel@gmail.com \
--cc=emacs-devel@gnu.org \
--cc=monnier@iro.umontreal.ca \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).