From: "Xu, Qian Q" <qian.q.xu@intel.com>
To: "Liu, Yong" <yong.liu@intel.com>,
"Ding, HengX" <hengx.ding@intel.com>,
"dts@dpdk.org" <dts@dpdk.org>
Cc: "Ding, HengX" <hengx.ding@intel.com>
Subject: Re: [dts] [PATCH] l3fwd/testpmd performance: optimize command for testpmd and comment some cases no need to run.
Date: Fri, 22 Apr 2016 01:30:16 +0000 [thread overview]
Message-ID: <82F45D86ADE5454A95A89742C8D1410E0324F19E@shsmsx102.ccr.corp.intel.com> (raw)
In-Reply-To: <86228AFD5BCD8E4EBFD2B90117B5E81E14F0643A@SHSMSX103.ccr.corp.intel.com>
See below comments.
Thanks
Qian
-----Original Message-----
From: dts [mailto:dts-bounces@dpdk.org] On Behalf Of Liu, Yong
Sent: Friday, April 22, 2016 9:07 AM
To: Ding, HengX; dts@dpdk.org
Cc: Ding, HengX
Subject: Re: [dts] [PATCH] l3fwd/testpmd performance: optimize command for testpmd and comment some cases no need to run.
Hi Heng, some comments below.
> -----Original Message-----
> From: dts [mailto:dts-bounces@dpdk.org] On Behalf Of Ding Heng
> Sent: Wednesday, April 20, 2016 10:17 AM
> To: dts@dpdk.org
> Cc: Ding, HengX
> Subject: [dts] [PATCH] l3fwd/testpmd performance: optimize command for
> testpmd and comment some cases no need to run.
>
> l3fwd: comment 2 cases no need to run in nightly performance test. Fix
> issues for run rfc2544 case with different nic. Add 72 byte for
> performance test.
>
> testpmd: change report format for generate nightly performance test.
> Use core mask instead of all cores for performance test.
>
> framework/pmd_output: optimize command for performance, run testpmd in
> vector mode instead of Xmit mode. Optimize core configuration for
> testpmd because run testpmd with all cores will get a bad performance
> result.
>
> Signed-off-by: Ding Heng <hengx.ding@intel.com>
>
> diff --git a/framework/pmd_output.py b/framework/pmd_output.py index
> df8db12..05e4e04 100644
> --- a/framework/pmd_output.py
> +++ b/framework/pmd_output.py
> @@ -101,14 +101,18 @@ class PmdOutput():
>
> def start_testpmd(self, cores, param='', eal_param='', socket=0):
> # in dpdk2.0 need used --txqflags param to open hardware
> features
> + # use 0xf01 to enable vector mode and get better performance.
> if "--txqflags" not in param:
> - param += " --txqflags=0"
Please not changed this default value for other suites depend on this. You can appoint this value in performance suite.
> + param += " --txqflags=0xf01"
>
> if type(cores) == list:
> core_list = cores
> elif cores == "Default":
> core_list = self.dut.get_core_list(self.default_cores)
> else:
> + # testpmd could not run with single core, at least 2
> + cores
> are needed.
> + if cores == "1S/1C/1T":
> + cores = "1S/2C/1T"
That's strange logic, please change input core setting as required.
[Qian] Yes, I am also confused first to see it, I know the cores setting about 2C is testpmd used cores since testpmd will have a master core will is not used in packet fwd. So, you can name different, such as testpmd_cores and fwd_cores.
prev parent reply other threads:[~2016-04-22 1:30 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-20 2:16 Ding Heng
2016-04-22 1:07 ` Liu, Yong
2016-04-22 1:30 ` Xu, Qian Q [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=82F45D86ADE5454A95A89742C8D1410E0324F19E@shsmsx102.ccr.corp.intel.com \
--to=qian.q.xu@intel.com \
--cc=dts@dpdk.org \
--cc=hengx.ding@intel.com \
--cc=yong.liu@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).