DPDK patches and discussions
 help / color / mirror / Atom feed
From: Aman Kumar <aman.kumar@vvdntech.in>
To: Thomas Monjalon <thomas@monjalon.net>
Cc: dpdk-dev <dev@dpdk.org>, Slava Ovsiienko <viacheslavo@nvidia.com>,
	 Anatoly Burakov <anatoly.burakov@intel.com>,
	"Song, Keesang" <keesang.song@amd.com>,
	 Jerin Jacob <jerinjacobk@gmail.com>,
	konstantin.ananyev@intel.com,  bruce.richardson@intel.com
Subject: Re: [dpdk-dev] [PATCH v3 3/3] lib/eal: add temporal store memcpy support on AMD platform
Date: Wed, 27 Oct 2021 12:04:54 +0530	[thread overview]
Message-ID: <CAEe4q1OJq4S_u+aW6-La05BTW2Nnqc80Q9DqwCS1uSOCGAyt1Q@mail.gmail.com> (raw)
In-Reply-To: <2148097.ar7J4MBmm8@thomas>

On Tue, Oct 26, 2021 at 9:44 PM Thomas Monjalon <thomas@monjalon.net> wrote:

> 26/10/2021 17:56, Aman Kumar:
> > This patch provides a rte_memcpy* call with temporal stores.
> > Use -Dcpu_instruction_set=znverX with build to enable this API.
> >
> > Signed-off-by: Aman Kumar <aman.kumar@vvdntech.in>
> > ---
> >  config/x86/meson.build           |   2 +
> >  lib/eal/x86/include/rte_memcpy.h | 114 +++++++++++++++++++++++++++++++
>
> It looks better as C code.
> Do you achieve the same performance as the asm version?
>

In a few corner cases assembly performed better, but overall we have very
similar perf observations.

> > +#if defined RTE_MEMCPY_AMDEPYC
> [...]
> > +static __rte_always_inline void *
> > +rte_memcpy_aligned_tstore16_generic(void *dst, void *src, int len)
>
> So to be clear, an application will benefit of this optimization if
> 1/ DPDK is specifically compiled for AMD
> 2/ the application is compiled with above DPDK build (because of
> inlinining)
>
> I guess there is no good way to benefit from the optimization
> without specific compilation, because of inlining constraint.
> Another design, with less constraint but less performance,
> would be to have a function pointer assigned at runtime based on the CPU.
>

You're right. We need to build DPDK and apps with this flag enabled to get
the benefit.
In future versions, we will try to adapt in a more dynamic way. Thanks.

  reply	other threads:[~2021-10-27  6:35 UTC|newest]

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-23  8:44 [dpdk-dev] [PATCH 1/2] lib/eal: add amd epyc2 memcpy routine to eal Aman Kumar
2021-08-23  8:44 ` [dpdk-dev] [PATCH 2/2] net/mlx5: optimize mprq memcpy for AMD EPYC2 platforms Aman Kumar
2021-10-13 16:53   ` Thomas Monjalon
2021-10-19 10:52     ` Aman Kumar
2021-08-23 15:21 ` [dpdk-dev] [PATCH 1/2] lib/eal: add amd epyc2 memcpy routine to eal Jerin Jacob
2021-08-30  9:39   ` Aman Kumar
2021-10-19 10:47 ` [dpdk-dev] [PATCH v2 " Aman Kumar
2021-10-19 10:47   ` [dpdk-dev] [PATCH v2 2/2] net/mlx5: optimize mprq memcpy for AMD EPYC2 plaform Aman Kumar
2021-10-19 12:31   ` [dpdk-dev] [PATCH v2 1/2] lib/eal: add amd epyc2 memcpy routine to eal Thomas Monjalon
2021-10-19 15:35     ` Stephen Hemminger
2021-10-21 17:10     ` Song, Keesang
2021-10-21 17:40       ` Ananyev, Konstantin
2021-10-21 18:12         ` Song, Keesang
2021-10-21 18:41           ` Thomas Monjalon
2021-10-21 19:03             ` Song, Keesang
2021-10-21 19:50               ` Thomas Monjalon
2021-10-21 20:14   ` Thomas Monjalon
2021-10-22  8:45     ` Bruce Richardson
2021-10-26 15:56   ` [dpdk-dev] [PATCH v3 1/3] config/x86: add support for AMD platform Aman Kumar
2021-10-26 15:56     ` [dpdk-dev] [PATCH v3 2/3] doc/guides: add dpdk build instruction for AMD platforms Aman Kumar
2021-10-26 16:07       ` Thomas Monjalon
2021-10-27  6:30         ` Aman Kumar
2021-10-26 15:56     ` [dpdk-dev] [PATCH v3 3/3] lib/eal: add temporal store memcpy support on AMD platform Aman Kumar
2021-10-26 16:14       ` Thomas Monjalon
2021-10-27  6:34         ` Aman Kumar [this message]
2021-10-27  7:59           ` Thomas Monjalon
2021-10-26 21:10       ` Stephen Hemminger
2021-10-27  6:43         ` Aman Kumar
2021-10-26 16:01     ` [dpdk-dev] [PATCH v3 1/3] config/x86: add support for " Thomas Monjalon
2021-10-27  6:26       ` Aman Kumar
2021-10-27  7:28     ` [dpdk-dev] [PATCH v4 1/2] " Aman Kumar
2021-10-27  7:28       ` [dpdk-dev] [PATCH v4 2/2] lib/eal: add temporal store memcpy " Aman Kumar
2021-10-27  8:13         ` Thomas Monjalon
2021-10-27 11:03           ` Van Haaren, Harry
2021-10-27 11:41             ` Mattias Rönnblom
2021-10-27 12:15               ` Van Haaren, Harry
2021-10-27 12:22                 ` Ananyev, Konstantin
2021-10-27 13:34                   ` Aman Kumar
2021-10-27 14:10                     ` Van Haaren, Harry
2021-10-27 14:31                       ` Thomas Monjalon
2021-10-29 16:01                         ` Song, Keesang
2021-10-27 14:26                     ` Ananyev, Konstantin
2021-10-27 11:33         ` Mattias Rönnblom

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=CAEe4q1OJq4S_u+aW6-La05BTW2Nnqc80Q9DqwCS1uSOCGAyt1Q@mail.gmail.com \
    --to=aman.kumar@vvdntech.in \
    --cc=anatoly.burakov@intel.com \
    --cc=bruce.richardson@intel.com \
    --cc=dev@dpdk.org \
    --cc=jerinjacobk@gmail.com \
    --cc=keesang.song@amd.com \
    --cc=konstantin.ananyev@intel.com \
    --cc=thomas@monjalon.net \
    --cc=viacheslavo@nvidia.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).