From: "Mcnamara, John" <john.mcnamara@intel.com>
To: "Zhang, Qi Z" <qi.z.zhang@intel.com>,
"Wu, Jingjing" <jingjing.wu@intel.com>
Cc: "Zhang, Helin" <helin.zhang@intel.com>, "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH] doc: add knowing issue for i40e VF performance
Date: Sun, 2 Jul 2017 19:10:03 +0000 [thread overview]
Message-ID: <B27915DBBA3421428155699D51E4CFE23ED1DE0D@IRSMSX104.ger.corp.intel.com> (raw)
In-Reply-To: <20170702120644.3510-1-qi.z.zhang@intel.com>
> -----Original Message-----
> From: Zhang, Qi Z
> Sent: Sunday, July 2, 2017 1:07 PM
> To: Mcnamara, John <john.mcnamara@intel.com>; Wu, Jingjing
> <jingjing.wu@intel.com>
> Cc: Zhang, Helin <helin.zhang@intel.com>; dev@dpdk.org; Zhang, Qi Z
> <qi.z.zhang@intel.com>
> Subject: [PATCH] doc: add knowing issue for i40e VF performance
>
> VF performance is limited by some kernel PCI setting.
> Update the document to explain the knowing issue and work around solution.
>
> Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
> ---
> doc/guides/nics/i40e.rst | 24 ++++++++++++++++++++++++
> 1 file changed, 24 insertions(+)
>
> diff --git a/doc/guides/nics/i40e.rst b/doc/guides/nics/i40e.rst index
> 4d3c7ca..557d83d 100644
> --- a/doc/guides/nics/i40e.rst
> +++ b/doc/guides/nics/i40e.rst
> @@ -447,3 +447,27 @@ It means if APP has set the max bandwidth for that
> TC, it comes to no effect.
> It's suggested to set the strict priority mode for a TC that is latency
> sensitive but no consuming much bandwidth.
> +
> +VF performance is impacted by PCI extended tag setting
> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> +
> +To reach maximum NIC performance. PCI extended tag is required to be
> enabled.
> +DPDK I40E PF drvier will set this feature during initialization, but
> +Kernel PF driver does not. So when running traffic on a VF which is
> +managed by kernel PF driver, we saw significent NIC performance
> +downgrade (for 64 bytes packet, there is about 25% linerate downgrade
> +for 25G device and about 35% for 40G device).
> +
> +Solution:
> +
> +For kernel version >= 4.11, kernel's PCI driver will enale extended tag
> +if it detects that device support extended tag. So by default, this is
> not an issue.
> +When extended tag is be disabled by occasionally, to re-enable it, see
> below.
> +
> +For kernel version < 4.11, use setpci command to enable PCI extended
> +flag
> +1) get current value of PCI configure register setpci -s <XX:XX.X> a8.w
> +2) set bit 8
> +value = value | 0x100
> +3) set PCI configure register with new value.
> +setpci -s <XX:XX.X> a8.w=<value>
To use a numbered list in the doc you should do it like this:
#. Get the current value of the PCI configure register::
setpci -s <XX:XX.X> a8.w
#. Set bit 8::
value = value | 0x100
#. Set the PCI configure register with new value::
setpci -s <XX:XX.X> a8.w=<value>
John
prev parent reply other threads:[~2017-07-02 19:10 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-02 12:06 Qi Zhang
2017-07-02 19:10 ` Mcnamara, John [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=B27915DBBA3421428155699D51E4CFE23ED1DE0D@IRSMSX104.ger.corp.intel.com \
--to=john.mcnamara@intel.com \
--cc=dev@dpdk.org \
--cc=helin.zhang@intel.com \
--cc=jingjing.wu@intel.com \
--cc=qi.z.zhang@intel.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).