From: Noah Goldstein <goldstein.w.n@gmail.com>
To: Sunil Pandey <skpgkp2@gmail.com>
Cc: GNU C Library <libc-alpha@sourceware.org>
Subject: Re: [PATCH v1] Benchtest: Add additional benchmarks for strlen and strnlen
Date: Wed, 28 Sep 2022 13:41:55 -0700 [thread overview]
Message-ID: <CAFUsyf+n6q3XPGiQ=yH2wxR1=NJfLWtSFPUxvzwhNxr08R3Qpg@mail.gmail.com> (raw)
In-Reply-To: <CAMAf5_fJ65S_8KOH+AtP2XVxLp5hj3seXbcWomA_S3=zV5-fUw@mail.gmail.com>
On Wed, Sep 28, 2022 at 12:27 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
>
> On Wed, Sep 28, 2022 at 12:20 PM Noah Goldstein via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > Current benchmarks are missing many cases in the mid-length range
> > which is often the hottest size range.
>
> How do you know if the mid-length range is the hottest size range?
> Can you please share data to back your claim.
Its moreso mid/small, just we are missing mid range.
>
> > ---
> > benchtests/bench-strlen.c | 33 ++++++++++++++++++++---
> > benchtests/bench-strnlen.c | 55 +++++++++++++++++++++++++++++++++++---
> > 2 files changed, 81 insertions(+), 7 deletions(-)
> >
> > diff --git a/benchtests/bench-strlen.c b/benchtests/bench-strlen.c
> > index 602e52ee83..696e4541d7 100644
> > --- a/benchtests/bench-strlen.c
> > +++ b/benchtests/bench-strlen.c
> > @@ -128,10 +128,10 @@ test_main (void)
> > /* Checking with only 4 * N alignments for wcslen, other alignments are wrong for wchar_t type arrays*/
> >
> > for (i = 1; i < 8; ++i)
> > - {
> > - do_test (&json_ctx, sizeof (CHAR) * i, i);
> > - do_test (&json_ctx, 0, i);
> > - }
> > + {
> > + do_test (&json_ctx, sizeof (CHAR) * i, i);
> > + do_test (&json_ctx, 0, i);
> > + }
> >
> > for (i = 2; i <= 12; ++i)
> > {
> > @@ -141,6 +141,31 @@ test_main (void)
> > do_test (&json_ctx, sizeof (CHAR) * i, (size_t)((1 << i) / 1.5));
> > }
> >
> > + for (i = 0; i < 512; i += 32)
> > + {
> > + do_test (&json_ctx, 0, i / sizeof (CHAR));
> > + }
> > +
> > + for (i = 512; i < 1024; i += 64)
> > + {
> > + do_test (&json_ctx, 0, i / sizeof (CHAR));
> > + }
> > +
> > + for (i = 1024; i < 2048; i += 128)
> > + {
> > + do_test (&json_ctx, 0, i / sizeof (CHAR));
> > + }
> > +
> > + for (i = 2048; i < 4096; i += 256)
> > + {
> > + do_test (&json_ctx, 0, i / sizeof (CHAR));
> > + }
> > +
> > + for (i = 4096; i < 8192; i += 512)
> > + {
> > + do_test (&json_ctx, 0, i / sizeof (CHAR));
> > + }
> > +
> > json_array_end (&json_ctx);
> > json_attr_object_end (&json_ctx);
> > json_attr_object_end (&json_ctx);
> > diff --git a/benchtests/bench-strnlen.c b/benchtests/bench-strnlen.c
> > index 2cd561dfd5..13b46b3f57 100644
> > --- a/benchtests/bench-strnlen.c
> > +++ b/benchtests/bench-strnlen.c
> > @@ -63,6 +63,11 @@ do_one_test (json_ctx_t *json_ctx, impl_t *impl, const CHAR *s, size_t maxlen,
> > ret = 1;
> > return;
> > }
> > + /* Warmup. */
> > + for (i = 0; i < iters / 16; ++i)
> > + {
> > + CALL (impl, s, maxlen);
> > + }
> >
> > TIMING_NOW (start);
> > for (i = 0; i < iters; ++i)
> > @@ -110,7 +115,7 @@ do_test (json_ctx_t *json_ctx, size_t align, size_t len, size_t maxlen,
> > int
> > test_main (void)
> > {
> > - size_t i;
> > + size_t i, j;
> > json_ctx_t json_ctx;
> >
> > test_init ();
> > @@ -131,6 +136,19 @@ test_main (void)
> >
> > json_array_begin (&json_ctx, "results");
> >
> > + for (i = 0; i <= 1; ++i)
> > + {
> > + do_test (&json_ctx, i, 1, 128, MIDDLE_CHAR);
> > + do_test (&json_ctx, i, 128, 1, MIDDLE_CHAR);
> > + do_test (&json_ctx, i, 1, 2, MIDDLE_CHAR);
> > + do_test (&json_ctx, i, 2, 1, MIDDLE_CHAR);
> > +
> > + do_test (&json_ctx, 32 + i, 1, 128, MIDDLE_CHAR);
> > + do_test (&json_ctx, 32 + i, 128, 1, MIDDLE_CHAR);
> > + do_test (&json_ctx, 32 + i, 1, 2, MIDDLE_CHAR);
> > + do_test (&json_ctx, 32 + i, 2, 1, MIDDLE_CHAR);
> > + }
> > +
> > for (i = 1; i < 8; ++i)
> > {
> > do_test (&json_ctx, 0, i, i - 1, MIDDLE_CHAR);
> > @@ -149,18 +167,49 @@ test_main (void)
> > {
> > do_test (&json_ctx, 0, 1 << i, 5000, MIDDLE_CHAR);
> > do_test (&json_ctx, 1, 1 << i, 5000, MIDDLE_CHAR);
> > + do_test (&json_ctx, 0, 5000, 1 << i, MIDDLE_CHAR);
> > + do_test (&json_ctx, 1, 5000, 1 << i, MIDDLE_CHAR);
> > }
> >
> > for (i = 1; i < 8; ++i)
> > - do_test (&json_ctx, 0, i, 5000, BIG_CHAR);
> > + {
> > + do_test (&json_ctx, 0, i, 5000, BIG_CHAR);
> > + do_test (&json_ctx, 0, 5000, i, BIG_CHAR);
> > + }
> >
> > for (i = 1; i < 8; ++i)
> > - do_test (&json_ctx, i, i, 5000, BIG_CHAR);
> > + {
> > + do_test (&json_ctx, i, i, 5000, BIG_CHAR);
> > + do_test (&json_ctx, i, 5000, i, BIG_CHAR);
> > + }
> >
> > for (i = 2; i <= 10; ++i)
> > {
> > do_test (&json_ctx, 0, 1 << i, 5000, BIG_CHAR);
> > do_test (&json_ctx, 1, 1 << i, 5000, BIG_CHAR);
> > + do_test (&json_ctx, 0, 5000, 1 << i, BIG_CHAR);
> > + do_test (&json_ctx, 1, 5000, 1 << i, BIG_CHAR);
> > + }
> > +
> > + for (i = (16 / sizeof (CHAR)); i <= (8192 / sizeof (CHAR)); i += i)
> > + {
> > + for (j = 0; j <= (704 / sizeof (CHAR)); j += (32 / sizeof (CHAR)))
> > + {
> > + do_test (&json_ctx, 0, 1 << i, (i + j), BIG_CHAR);
> > + do_test (&json_ctx, 0, i + j, i, BIG_CHAR);
> > +
> > + do_test (&json_ctx, 64, 1 << i, (i + j), BIG_CHAR);
> > + do_test (&json_ctx, 64, i + j, i, BIG_CHAR);
> > +
> > + if (j < i)
> > + {
> > + do_test (&json_ctx, 0, 1 << i, i - j, BIG_CHAR);
> > + do_test (&json_ctx, 0, i - j, i, BIG_CHAR);
> > +
> > + do_test (&json_ctx, 64, 1 << i, i - j, BIG_CHAR);
> > + do_test (&json_ctx, 64, i - j, i, BIG_CHAR);
> > + }
> > + }
> > }
> >
> > json_array_end (&json_ctx);
> > --
> > 2.34.1
> >
prev parent reply other threads:[~2022-09-28 20:42 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-28 19:20 Noah Goldstein
2022-09-28 19:25 ` H.J. Lu
2022-09-28 19:26 ` Sunil Pandey
2022-09-28 20:41 ` Noah Goldstein [this message]
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='CAFUsyf+n6q3XPGiQ=yH2wxR1=NJfLWtSFPUxvzwhNxr08R3Qpg@mail.gmail.com' \
--to=goldstein.w.n@gmail.com \
--cc=libc-alpha@sourceware.org \
--cc=skpgkp2@gmail.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 read-only IMAP folder(s) and NNTP newsgroup(s).