From mboxrd@z Thu Jan 1 00:00:00 1970 From: zengxg14 Subject: Re: 1 core / 1 port receiving performance discrepancy Date: Sat, 13 Jul 2013 13:51:10 +0800 Message-ID: <201307131351011854082@gmail.com> References: Reply-To: zengxg14 Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_001_NextPart286070380570_=----" To: "jinho hwang" , dev Return-path: List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces-VfR2kkLFssw@public.gmane.org Sender: "dev" This is a multi-part message in MIME format. ------=_001_NextPart286070380570_=---- Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: base64 SGk6DQoNCldoYXQgaXMgeW91ciBidXJzdCBzaXplIGluIHJ0ZV9ldGhfcnhfYnVyc3Q/DQpDYW4g eW91IHByb3ZpZGUgdGhlIHNoZWxsIGNvbW1hbmQgeW91IHJ1biwgYm90aCB0ZXN0LXBtZCBhbmQg eW91ciBjb2RlPw0KDQoNCg0KDQoNCg0KRnJvbTogamluaG8gaHdhbmcNCkRhdGU6IDIwMTMtMDct MTIgMjM6NDYNClRvOiBkZXYNClN1YmplY3Q6IFtkcGRrLWRldl0gMSBjb3JlIC8gMSBwb3J0IHJl Y2VpdmluZyBwZXJmb3JtYW5jZSBkaXNjcmVwYW5jeQ0KSGkgR3V5cywNCg0KSSBhbSBjdXJyZW50 bHkgdGVzdGluZyBteSBjb2RlIHdpdGggdHdvIG1hY2hpbmVzKFg1NjUwIEAgMi42N0dIeiwNCjgy NTk5RUIpOiBvbmUgcnVucyBteSBjb2RlLCBhbmQgdGhlIG90aGVyIHJ1bnMgUGt0Z2VuLiBUaGUg Zmlyc3QgdGFzaw0KaXMgdG8gc2VlIGhvdyBteSBjb2RlIHBlcmZvcm1zIGNvbXBhcmluZyB0byB0 ZXN0LXBtZC4gQm90aCB1c2Ugb25seQ0Kb25lIHBvcnQgd2l0aCBvbmUgY29yZSB0byByZWNlaXZl IHBhY2tldHMuIHRlc3QtcG1kIHVzZXMgcnhvbmx5IG1vZGUuDQpNeSBjb2RlIG9ubHkgcmVjZWl2 ZXMgcGFja2V0cyBhbmQgZGlzY2FyZHMgdGhlbSByaWdodCBhd2F5LCBhbmQgdGhlcmUNCmFyZSBu byBvdGhlciBwcm9jZXNzaW5nIGF0IGFsbC4gRGlmZmVyZW50IGZyb20gd2hhdCBJIGV4cGVjdGVk LA0KdGVzdC1wbWQgc2hvd3MgMzNNIHBhY2tldHMgKGRvIG5vdCBjb3VudCBvbiB0aGUgc2Vjb25k IC0tLSBqdXN0IHRvdGFsDQpudW1iZXIgb2YgcGFja2V0cyBzZW50IGZyb20gUGt0Z2VuKSwgd2hl cmVhcyBtaW5lIHNob3dzIDE4TSBwYWNrZXRzDQp3aXRoIHRoZSBzYW1lIHRyYWZmaWMuDQoNCkkg aGF2ZSB0cmllZCB0byBjb25maWd1cmUgbWFueSB0aGluZ3Mgc3VjaCBhcyBpbmNyZWFzaW5nIHJ4 IHF1ZXVlDQpzaXplLCBpbmNyZWFzaW5nIG1heCBwYWNrZXQgcmVjZWl2ZSBidWZmZXIsIGJ1dCB0 aGUgbnVtYmVyIGRpZCBub3QNCm1ha2UgZGlmZmVyZW5jZXMuDQoNCkNhbiBJIGdldCBzb21lIGhl bHA/DQoNClRoYW5rIHlvdSwNCg0KSmluaG8= ------=_001_NextPart286070380570_=---- Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable =EF=BB=BF

Hi:

What is your burst size in rte_eth_rx_burst?

Can you provide the shell command you run, both test-pmd and you= r=20 code?

From: jinho=20 hwang

Date: 2013-07-12 23:46

To: dev

Subject: [dpdk-dev] 1 core / 1 port receiving performance= =20 discrepancy

Hi Guys,

I am currently testing my code with&nbs= p;two machines(X5650 @ 2.67GHz,

82599EB): one runs my code, and the&nbs= p;other runs Pktgen. The first task

is to see how my code performs com= paring to test-pmd. Both use only

one port with one core to receive = packets. test-pmd uses rxonly mode.

My code only receives packets and disca= rds them right away, and there

are no other processing at all. Differe= nt from what I expected,

test-pmd shows 33M packets (do not coun= t on the second --- just total

number of packets sent from Pktgen), wh= ereas mine shows 18M packets

with the same traffic.

I have tried to configure many things&n= bsp;such as increasing rx queue

size, increasing max packet receive buffer,&= nbsp;but the number did not

make differences.

Can I get some help?

Thank you,

Jinho

------=_001_NextPart286070380570_=------