From: edgar helmut <helmut.edgar100@gmail.com>
To: "Wiles, Keith" <keith.wiles@intel.com>
Cc: "users@dpdk.org" <users@dpdk.org>
Subject: Re: [dpdk-users] Dpdk poor performance on virtual machine
Date: Thu, 15 Dec 2016 15:32:04 +0200 [thread overview]
Message-ID: <CABc_bMDkX+JF7qQY6pRrJWY0A2if8E=0UUFdD97Z2SBE2mN78Q@mail.gmail.com> (raw)
In-Reply-To: <D2D84874-8E8B-4CB0-B3B5-859ADBE62E2F@intel.com>
I have one single socket which is Intel(R) Xeon(R) CPU E5-2640 v4 @ 2.40GHz.
I just made two more steps:
1. setting iommu=pt for better usage of the igb_uio
2. using taskset and isolcpu so now it looks like the relevant dpdk cores
use dedicated cores.
It improved the performance though I still see significant difference
between the vm and the host which I can't fully explain.
any further idea?
Regards,
Edgar
On Thu, Dec 15, 2016 at 2:54 PM, Wiles, Keith <keith.wiles@intel.com> wrote:
>
> > On Dec 15, 2016, at 1:20 AM, edgar helmut <helmut.edgar100@gmail.com>
> wrote:
> >
> > Hi.
> > Some help is needed to understand performance issue on virtual machine.
> >
> > Running testpmd over the host functions well (testpmd forwards 10g
> between
> > two 82599 ports).
> > However same application running on a virtual machine over same host
> > results with huge degradation in performance.
> > The testpmd then is not even able to read 100mbps from nic without drops,
> > and from a profile i made it looks like a dpdk application runs more than
> > 10 times slower than over host…
>
> Not sure I understand the overall setup, but did you make sure the NIC/PCI
> bus is on the same socket as the VM. If you have multiple sockets on your
> platform. If you have to access the NIC across the QPI it could explain
> some of the performance drop. Not sure that much drop is this problem.
>
> >
> > Setup is ubuntu 16.04 for host and ubuntu 14.04 for guest.
> > Qemu is 2.3.0 (though I tried with a newer as well).
> > NICs are connected to guest using pci passthrough, and guest's cpu is set
> > as passthrough (same as host).
> > On guest start the host allocates transparent hugepages (AnonHugePages)
> so
> > i assume the guest memory is backed with real hugepages on the host.
> > I tried binding with igb_uio and with uio_pci_generic but both results
> with
> > same performance.
> >
> > Due to the performance difference i guess i miss something.
> >
> > Please advise what may i miss here?
> > Is this a native penalty of qemu??
> >
> > Thanks
> > Edgar
>
> Regards,
> Keith
>
>
next prev parent reply other threads:[~2016-12-15 13:32 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <CABc_bMBYNa7jdhqu2SNZkwtoJskCwQWGOZrB71RYzBZx5-OTfw@mail.gmail.com>
[not found] ` <CABc_bMBMxhbkGSg82tw7CKLw7AiccEESLAJaHGA6PAaYAPCmTg@mail.gmail.com>
[not found] ` <CABc_bMB2vnXEpacfeiWiC5X-x8iuB7jO6fR67n5jj67Pspcf2g@mail.gmail.com>
[not found] ` <CABc_bMB7C2zRjEkFbtAyk7buaP=QWwxCSwz=60AVJT_rjoPZTQ@mail.gmail.com>
[not found] ` <CABc_bMDgU62=i5uE52d12VGd8wUzqHq9siOjYKSh8v811TYpug@mail.gmail.com>
[not found] ` <CABc_bMCn9LivEn_PJCh+dQy7pPypLdmE+eZFQZtzphsHEgr08g@mail.gmail.com>
[not found] ` <CABc_bMADQ3PaZz5ovLmGu=3_w2pTO3c6BvNPM6-BUzJ5BXjxLQ@mail.gmail.com>
[not found] ` <CABc_bMBJ+NZLuo___qw_cBB=z=5nzgVx6rMd15nMiSYf_q3WoQ@mail.gmail.com>
[not found] ` <CABc_bMC6qP3K-kqVQORc9XKfbcXX63UfN=AdZ+sksHUN+Bx5kw@mail.gmail.com>
[not found] ` <CABc_bMDowcZKMrKc8omf-JqpUW=uPn-fq8sfLYk=AktbH9-aNw@mail.gmail.com>
[not found] ` <CABc_bMACisgaHBZedG05ZzJ3wzmudgBeTHdRr93M3-QOGKDKNw@mail.gmail.com>
[not found] ` <CABc_bMCTRDYy+9ZO6py+KupRStR=Rc4md+J0NhPcyTqaZhKxTA@mail.gmail.com>
[not found] ` <CABc_bMAu8BkZXzBSZLWNs=R6AJgMAw9WrTki=cEzMzjC7Z8LAQ@mail.gmail.com>
[not found] ` <CABc_bMDtzYE4NvhAT8nqi3qZbzV4WauzJLW-tcY_Wi5i88F7yQ@mail.gmail.com>
[not found] ` <CABc_bMAr0EjgH4f7UT-oi353FBqYwxt-UgfP6candcP=jkuyLg@mail.gmail.com>
[not found] ` <CABc_bMCyJ4o94T=VRWPjfyhXP1T3uDmeYVsu+0OrXi1AqUkLrg@mail.gmail.com>
[not found] ` <CABc_bMBK=AQ31=mQX0f7HMXs6REqj_bkqQ9KPD96vbtoKKHWCw@mail.gmail.com>
[not found] ` <CABc_bMA8z6XBEtnxwPE_LDpngx_DYRathn17KqFxNx2V5DFbng@mail.gmail.com>
[not found] ` <CABc_bMBwFAL5rc5goMP9pt+2z=TOM=VNfp76AznZ3jend9aS_Q@mail.gmail.com>
[not found] ` <CABc_bMBr1c_Evd5zxB89SinNFVMUGHFTBDoLusVHdMySBCyaBA@mail.gmail.com>
[not found] ` <CABc_bMChD5hpWmj1zhvZ5tocsxDCFH0QdE34TKWXsdo_danH-Q@mail.gmail.com>
[not found] ` <CABc_bMCqu-4V4gn=JPqO491BF6Cnjj=4SaAey9qyTQcha134yw@mail.gmail.com>
2016-12-15 7:20 ` edgar helmut
2016-12-15 12:54 ` Wiles, Keith
2016-12-15 13:32 ` edgar helmut [this message]
2016-12-15 14:33 ` Hu, Xuekun
2016-12-15 17:17 ` Stephen Hemminger
2016-12-15 17:29 ` edgar helmut
2016-12-15 19:14 ` Stephen Hemminger
2016-12-15 19:29 ` Jes Nielsen
2016-12-15 17:24 ` edgar helmut
2016-12-16 1:14 ` Hu, Xuekun
2016-12-17 12:56 ` edgar helmut
2016-12-23 19:22 ` edgar helmut
2016-12-24 7:06 ` Hu, Xuekun
2016-12-24 8:06 ` edgar helmut
2016-12-24 15:52 ` edgar helmut
2016-12-26 0:52 ` Hu, Xuekun
2016-12-27 15:52 ` edgar helmut
2016-12-27 15:59 ` edgar helmut
2016-12-27 18:52 ` Stephen Hemminger
2016-12-28 8:09 ` edgar helmut
2017-01-04 6:44 ` edgar helmut
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='CABc_bMDkX+JF7qQY6pRrJWY0A2if8E=0UUFdD97Z2SBE2mN78Q@mail.gmail.com' \
--to=helmut.edgar100@gmail.com \
--cc=keith.wiles@intel.com \
--cc=users@dpdk.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).