From: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
To: "Morten Brørup" <mb@smartsharesystems.com>,
"Bruce Richardson" <bruce.richardson@intel.com>,
dev@dpdk.org
Subject: Re: [PATCH v2 4/4] mempool perf test: test random bulk sizes
Date: Mon, 31 Mar 2025 17:48:42 +0300 [thread overview]
Message-ID: <3fd8b3ab-27e9-4972-850f-60466652e223@oktetlabs.ru> (raw)
In-Reply-To: <20250331100343.213430-5-mb@smartsharesystems.com>
On 3/31/25 13:03, Morten Brørup wrote:
> Bulk requests to get or put objects in a mempool often vary in size.
> A series of tests with pseudo random request sizes, to mitigate the
> benefits of the CPU's dynamic branch predictor, was added.
>
> Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
> Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
[snip]
> @@ -181,9 +240,9 @@ per_lcore_mempool_test(void *arg)
> }
>
> /* n_get_bulk and n_put_bulk must be divisors of n_keep */
> - if (((n_keep / n_get_bulk) * n_get_bulk) != n_keep)
> + if (!n_max_bulk && (((n_keep / n_get_bulk) * n_get_bulk) != n_keep))
IMHO n_max_bulk == 0 would be easier to read and as far as I remember
DPDK coding style recommends the same style.
> GOTO_ERR(ret, out);
> - if (((n_keep / n_put_bulk) * n_put_bulk) != n_keep)
> + if (!n_max_bulk && (((n_keep / n_put_bulk) * n_put_bulk) != n_keep))
same
> GOTO_ERR(ret, out);
> /* for constant n, n_get_bulk and n_put_bulk must be the same */
> if (use_constant_values && n_put_bulk != n_get_bulk)
> @@ -200,7 +259,9 @@ per_lcore_mempool_test(void *arg)
> start_cycles = rte_get_timer_cycles();
>
> while (time_diff/hz < TIME_S) {
> - if (!use_constant_values)
> + if (n_max_bulk)
n_max_bulk != 0
as DPDK coding style says
> + ret = test_loop_random(mp, cache, n_keep, n_max_bulk);
> + else if (!use_constant_values)
> ret = test_loop(mp, cache, n_keep, n_get_bulk, n_put_bulk);
> else if (n_get_bulk == 1)
> ret = test_loop(mp, cache, n_keep, 1, 1);
> @@ -261,9 +322,13 @@ launch_cores(struct rte_mempool *mp, unsigned int cores)
> use_external_cache ? external_cache_size : (unsigned int) mp->cache_size,
> cores,
> n_keep);
> - printf("n_get_bulk=%3u n_put_bulk=%3u constant_n=%u ",
> - n_get_bulk, n_put_bulk,
> - use_constant_values);
> + if (n_max_bulk)
same
> + printf("n_max_bulk=%3u ",
> + n_max_bulk);
> + else
> + printf("n_get_bulk=%3u n_put_bulk=%3u constant_n=%u ",
> + n_get_bulk, n_put_bulk,
> + use_constant_values);
>
> if (rte_mempool_avail_count(mp) != MEMPOOL_SIZE) {
> printf("mempool is not full\n");
[snip]
next prev parent reply other threads:[~2025-03-31 14:48 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-28 16:48 [PATCH] " Morten Brørup
2025-03-13 8:23 ` Morten Brørup
2025-03-25 7:15 ` Morten Brørup
2025-03-30 8:29 ` Andrew Rybchenko
2025-03-30 8:57 ` Morten Brørup
2025-03-31 10:03 ` [PATCH v2 0/4] " Morten Brørup
2025-03-31 10:03 ` [PATCH v2 1/4] mempool perf test: replace bare unsigned with unsigned int Morten Brørup
2025-03-31 10:03 ` [PATCH v2 2/4] mempool perf test: test default mempool with cache Morten Brørup
2025-03-31 10:03 ` [PATCH v2 3/4] mempool perf test: improve output readability Morten Brørup
2025-03-31 10:03 ` [PATCH v2 4/4] mempool perf test: test random bulk sizes Morten Brørup
2025-03-31 14:48 ` Andrew Rybchenko [this message]
2025-04-02 9:07 ` Morten Brørup
2025-03-31 14:48 ` [PATCH v2 0/4] " Andrew Rybchenko
2025-04-01 15:00 ` Morten Brørup
2025-04-01 15:00 ` [PATCH v2 1/4] mempool perf test: replace bare unsigned with unsigned int Morten Brørup
2025-04-01 15:00 ` [PATCH v2 2/4] mempool perf test: test default mempool with cache Morten Brørup
2025-04-01 15:00 ` [PATCH v2 3/4] mempool perf test: improve output readability Morten Brørup
2025-04-01 15:00 ` [PATCH v2 4/4] mempool perf test: test random bulk sizes Morten Brørup
2025-04-01 15:02 ` [PATCH v3 0/4] " Morten Brørup
2025-04-01 15:02 ` [PATCH v3 1/4] mempool perf test: replace bare unsigned with unsigned int Morten Brørup
2025-04-01 15:02 ` [PATCH v3 2/4] mempool perf test: test default mempool with cache Morten Brørup
2025-04-01 15:02 ` [PATCH v3 3/4] mempool perf test: improve output readability Morten Brørup
2025-04-01 15:02 ` [PATCH v3 4/4] mempool perf test: test random bulk sizes Morten Brørup
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3fd8b3ab-27e9-4972-850f-60466652e223@oktetlabs.ru \
--to=andrew.rybchenko@oktetlabs.ru \
--cc=bruce.richardson@intel.com \
--cc=dev@dpdk.org \
--cc=mb@smartsharesystems.com \
/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.
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 NNTP newsgroup(s).