Feedback-ID: ia2694551:Fastmail
Date: Mon, 6 Feb 2023 11:11:32 +0100
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Yugo NAGATA <nagata@sraoss.co.jp>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Fabien COELHO <coelho@cri.ensmp.fr>,
	pgsql-hackers@postgresql.org
Subject: Re: pgbench: using prepared BEGIN statement in a pipeline could
 cause an error
Message-ID: <20230206101132.3oge5d5ovoirvxhb@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20220930100743.e08533490f3b1b6f797cf431@sraoss.co.jp>
Archived-At: <https://www.postgresql.org/message-id/20230206101132.3oge5d5ovoirvxhb%40alvherre.pgsql>
Precedence: bulk

I'm writing my own patch for this problem.  While playing around with
it, I noticed this:

struct Command {
    PQExpBufferData            lines;                /*     0    24 */
    char *                     first_line;           /*    24     8 */
    int                        type;                 /*    32     4 */
    MetaCommand                meta;                 /*    36     4 */
    int                        argc;                 /*    40     4 */

    /* XXX 4 bytes hole, try to pack */

    char *                     argv[256];            /*    48  2048 */
    /* --- cacheline 32 boundary (2048 bytes) was 48 bytes ago --- */
    char *                     varprefix;            /*  2096     8 */
    PgBenchExpr *              expr;                 /*  2104     8 */
    /* --- cacheline 33 boundary (2112 bytes) --- */
    SimpleStats                stats;                /*  2112    40 */
    int64                      retries;              /*  2152     8 */
    int64                      failures;             /*  2160     8 */

    /* size: 2168, cachelines: 34, members: 11 */
    /* sum members: 2164, holes: 1, sum holes: 4 */
    /* last cacheline: 56 bytes */
};

Not great.  I suppose this makes pgbench slower than it needs to be.
Can we do better?

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"Nunca se desea ardientemente lo que solo se desea por razón" (F. Alexandre)