qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Juan Quintela <quintela@redhat.com>
To: Richard Henderson <richard.henderson@linaro.org>
Cc: qemu-devel@nongnu.org,  peter.maydell@linaro.org,
	 Peter Xu <peterx@redhat.com>,
	 Leonardo Bras <leobras@redhat.com>
Subject: Re: [PATCH 7/9] migration/xbzrle: Use i386 cacheinfo.h
Date: Thu, 18 May 2023 11:44:20 +0200	[thread overview]
Message-ID: <87v8gqdle3.fsf@secure.mitica> (raw)
In-Reply-To: <20230518044058.2777467-8-richard.henderson@linaro.org> (Richard Henderson's message of "Wed, 17 May 2023 21:40:56 -0700")

Richard Henderson <richard.henderson@linaro.org> wrote:
> Perform the function selection once, and only if CONFIG_AVX512_OPT
> is enabled.  Centralize the selection to xbzrle.c, instead of
> spreading the init across 3 files.
>
> Remove xbzrle-bench.c.  The benefit of being able to benchmark
> the different implementations is less important than peeking into
> the internals of the implementation.

Agreed.  If AVX512 is not better than a plain C implementation, better
to not have it O:-)

> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Juan Quintela <quintela@redhat.com>

Not queued.  It needs your other patches, I think it is better that all
teh series go through your tree.


> +static void __attribute__((constructor)) init_accel(void)
> +{
> +    unsigned info = cpuinfo_init();
> +    if (info & CPUINFO_AVX512BW) {
> +        accel_func = xbzrle_encode_buffer_avx512;
> +    } else {
> +        accel_func = xbzrle_encode_buffer_int;
> +    }
> +}

Wow.
Comparing it with previous implementation that did it by hand with asm.



  reply	other threads:[~2023-05-18  9:45 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-05-18  4:40 [PATCH 0/9] Host-specific includes, begin cpuinfo.h Richard Henderson
2023-05-18  4:40 ` [PATCH 1/9] util: Introduce host-specific cpuinfo.h Richard Henderson
2023-05-18  9:30   ` Juan Quintela
2023-05-18  4:40 ` [PATCH 2/9] util: Add cpuinfo-i386.c Richard Henderson
2023-05-18  9:35   ` Juan Quintela
2023-05-18 12:45     ` Richard Henderson
2023-05-18  4:40 ` [PATCH 3/9] util: Add i386 CPUINFO_ATOMIC_VMOVDQU Richard Henderson
2023-05-18 15:52   ` Peter Maydell
2023-05-18  4:40 ` [PATCH 4/9] tcg/i386: Use cpuinfo.h Richard Henderson
2023-05-18 15:53   ` Peter Maydell
2023-05-18  4:40 ` [PATCH 5/9] util/bufferiszero: Use i386 cpuinfo.h Richard Henderson
2023-05-18  9:49   ` Juan Quintela
2023-05-18 12:48     ` Richard Henderson
2023-05-18  4:40 ` [PATCH 6/9] migration/xbzrle: Shuffle function order Richard Henderson
2023-05-18  9:19   ` Juan Quintela
2023-05-18  4:40 ` [PATCH 7/9] migration/xbzrle: Use i386 cacheinfo.h Richard Henderson
2023-05-18  9:44   ` Juan Quintela [this message]
2023-05-18  4:40 ` [PATCH 8/9] migration: Build migration_files once Richard Henderson
2023-05-18  9:20   ` Juan Quintela
2023-05-18  4:40 ` [PATCH 9/9] util: Add cpuinfo-aarch64.c Richard Henderson
2023-05-18 15:55   ` Peter Maydell

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=87v8gqdle3.fsf@secure.mitica \
    --to=quintela@redhat.com \
    --cc=leobras@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=peterx@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    /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).