Horton

Re: appendBinaryStringInfo stuff

John Naylor <john.naylor@enterprisedb.com>

View thread View on postgr.es
From: John Naylor <john.naylor@enterprisedb.com>
To: David Rowley <dgrowleyml@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Andres Freund <andres@anarazel.de>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2022-12-22T12:20:37Z
Lists: pgsql-hackers 
On Thu, Dec 22, 2022 at 4:19 PM David Rowley <dgrowleyml@gmail.com> wrote:
>
> Test 1 (from earlier)
>
> master + escape_json using appendStringInfoCharMacro
> $ pgbench -n -T 60 -f bench.sql -M prepared postgres | grep latency
> latency average = 1.807 ms
> latency average = 1.800 ms
> latency average = 1.812 ms (~4.8% faster than master)

> 23.05%  postgres          [.] pg_utf_mblen

I get about 20% improvement by adding an ascii fast path in
pg_mbstrlen_with_len, which I think would work with all encodings we
support:

@@ -1064,7 +1064,12 @@ pg_mbstrlen_with_len(const char *mbstr, int limit)

        while (limit > 0 && *mbstr)
        {
-               int                     l = pg_mblen(mbstr);
+               int                     l;
+
+               if (!IS_HIGHBIT_SET(*mbstr))
+                       l = 1;
+               else
+                       l = pg_mblen(mbstr);

--
John Naylor
EDB: http://www.enterprisedb.com

Commits

  1. Change argument type of pq_sendbytes from char * to void *

  2. Remove useless casts to (void *) in hash_search() calls

  3. Change argument of appendBinaryStringInfo from char * to void *

  4. Use appendStringInfoString instead of appendBinaryStringInfo where possible