From: Sunil Pandey <skpgkp2@gmail.com>
To: James Tirta Halim <tirtajames45@gmail.com>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH] sysdeps/memmem-avx2.c: add memmem-avx2.c
Date: Mon, 11 Dec 2023 09:44:32 -0800 [thread overview]
Message-ID: <CAMAf5_edePM69BxZ1tSFRBc6024O6KBaoFopjUj2qF0GnSpDDA@mail.gmail.com> (raw)
In-Reply-To: <20231211173320.702246-1-tirtajames45@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 2610 bytes --]
Can you please provide some context/comment on this patch?
--Sunil
On Mon, Dec 11, 2023 at 9:37 AM James Tirta Halim <tirtajames45@gmail.com>
wrote:
> ---
> sysdeps/x86_64/multiarch/memmem-avx2.c | 55 ++++++++++++++++++++++++++
> 1 file changed, 55 insertions(+)
> create mode 100644 sysdeps/x86_64/multiarch/memmem-avx2.c
>
> diff --git a/sysdeps/x86_64/multiarch/memmem-avx2.c
> b/sysdeps/x86_64/multiarch/memmem-avx2.c
> new file mode 100644
> index 0000000000..b0cced73aa
> --- /dev/null
> +++ b/sysdeps/x86_64/multiarch/memmem-avx2.c
> @@ -0,0 +1,55 @@
> +#include <immintrin.h>
> +#include <string.h>
> +#include <inttypes.h>
> +#include <libc-pointer-arith.h>
> +
> +void *
> +__memmem_avx2 (const void *hs, size_t hs_len, const void *ne, size_t
> ne_len)
> +{
> + if (ne_len == 1)
> + return (void *) memchr (hs, *(unsigned char *) ne, hs_len);
> + if (__glibc_unlikely (ne_len == 0))
> + return (void *) hs;
> + if (__glibc_unlikely (hs_len == ne_len))
> + return !memcmp (hs, ne, ne_len) ? (void *) hs : NULL;
> + if (__glibc_unlikely (hs_len < ne_len))
> + return NULL;
> + const __m256i nv = _mm256_set1_epi8 (*(char *) ne);
> + const unsigned char *h = (const unsigned char *) hs;
> + const unsigned char *n = (const unsigned char *) ne;
> + const unsigned char *const end = h + hs_len - ne_len;
> + const int c1 = *(n + 1);
> + n += 2, ne_len -= 2;
> + __m256i hv;
> + uint32_t i, m;
> + if (!PTR_IS_ALIGNED (h)) {
> + hv = _mm256_loadu_si256 ((const __m256i *) h);
> + m = (uint32_t) _mm256_movemask_epi8 (_mm256_cmpeq_epi8 (hv, nv));
> + for (; m; m = _blsr_u32 (m)) {
> + i = _tzcnt_u32 (m);
> + if (__glibc_unlikely (h + i > end))
> + return NULL;
> + if (*(h + i + 1) == c1 && !memcmp (h + i + 2, n, ne_len))
> + return (char *) h + i;
> + }
> + h += sizeof (__m256i);
> + if (__glibc_unlikely (h > end))
> + return NULL;
> + h = (const unsigned char *) PTR_ALIGN_UP (h, sizeof (__m256i));
> + }
> + for (;;) {
> + hv = _mm256_load_si256 ((const __m256i *) h);
> + m = (uint32_t) _mm256_movemask_epi8 (_mm256_cmpeq_epi8 (hv, nv));
> + for (; m; m = _blsr_u32 (m)) {
> + i = _tzcnt_u32 (m);
> + if (__glibc_unlikely (h + i > end))
> + return NULL;
> + if (*(h + i + 1) == c1 && !memcmp (h + i + 2, n, ne_len))
> + return (char *) h + i;
> + }
> + h += sizeof (__m256i);
> + if (__glibc_unlikely (h > end))
> + return NULL;
> + }
> + return NULL;
> +}
> --
> 2.43.0
>
>
next prev parent reply other threads:[~2023-12-11 17:45 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-11 17:33 James Tirta Halim
2023-12-11 17:44 ` Sunil Pandey [this message]
2023-12-11 18:11 ` Noah Goldstein
2023-12-15 17:03 ` James Tirta Halim
2023-12-15 19:53 ` Carlos O'Donell
2023-12-16 4:33 ` [PATCH] sysdeps/x86_64/multiarch/memmem-avx2.c: " James Tirta Halim
2023-12-18 14:12 ` Carlos O'Donell
2023-12-18 17:48 ` Noah Goldstein
2023-12-18 23:08 ` James
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=CAMAf5_edePM69BxZ1tSFRBc6024O6KBaoFopjUj2qF0GnSpDDA@mail.gmail.com \
--to=skpgkp2@gmail.com \
--cc=libc-alpha@sourceware.org \
--cc=tirtajames45@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).