From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ig0-f182.google.com (mail-ig0-f182.google.com [209.85.213.182]) by dpdk.org (Postfix) with ESMTP id BE2D820F for ; Sat, 17 Oct 2015 20:43:37 +0200 (CEST) Received: by igbdj2 with SMTP id dj2so33825959igb.1 for ; Sat, 17 Oct 2015 11:43:37 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:date:message-id:subject:from:to:content-type; bh=jHjAuhaKKFo3+K1bvrzw2vKP8eFsPWkS9jglsf1vgOQ=; b=bs5YniFVLb+EJyYB9u9tbeHvNqaHU5JJJtyRfEjDEhtBcRq6CNRkztMk19qAMLRPaN MbIZG4TiT4l+LCUd4x67G4wNPeSGUbp/ne9X8X6mq1bqNsuihpUsxg8Qb/6DPf7LO7aA TWLjvbUYuCuIie6iXmllW/E+DzdJOrHr94FbBQEgjD/bhzzbjHUXP5jov4hzcTjjINRR kwEakD4gr/1mLVLxtqRgy3JDTBoiYRsgYN1pjjgYN0HT6J5d2oHN77W0hC3K55sBc9pd SHwvGaOmXx8wLhw3sEEdhLjqNroMFX5iFa0kcShHXakREbRU/lpr9wR2vfzp9LcTLJMj iqgg== MIME-Version: 1.0 X-Received: by 10.50.111.79 with SMTP id ig15mr11453496igb.51.1445107417041; Sat, 17 Oct 2015 11:43:37 -0700 (PDT) Received: by 10.107.31.142 with HTTP; Sat, 17 Oct 2015 11:43:37 -0700 (PDT) Date: Sat, 17 Oct 2015 15:43:37 -0300 Message-ID: From: Victor Detoni To: dev@dpdk.org Content-Type: text/plain; charset=UTF-8 X-Content-Filtered-By: Mailman/MimeDel 2.1.15 Subject: [dpdk-dev] l3fwd performance drops with +3 rx cores X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 17 Oct 2015 18:43:38 -0000 hi guys, I'm trying l3fwd to use all of my 8 cores to reach better distributed performance, but I could reach only 3 cores with 2 ports. My environment is vmware 6.0 with vmxnet3 (vmware driver) and my processor is E5-2420-v2. My virtual machine has 8 cores and 6GB of memory. Please, I'd ike to know if is it possible to use all of my cores on rx queues? When I use below command, it works fine, reaching 1.3Mpps: ./l3fwd -c 0xff -n 1 -- -p 0x3 --config=(0,0,0),(0,1,1),(1,0,2)" If I use ./l3fwd -c 0xff -n 1 -- -p 0x3 --config=(0,0,0),(0,1,1),(0,2,2),(1,0,3)" the program shows the below error: Case: rte_eth_dev_start: err=-1, port=0 if I use ./l3fwd -c 0xff -n 1 -- -p 0x3 --config=(0,0,0),(0,1,1),(0,2,2),(0,3,3),(1,0,4)" the program works, but reach only 800kpps My virtual machine has 1 node (node 0) on numa support. Please, Someone knows why is it impossible to reach 1.3Mpps with +4 cores on rx queues? thanks Victor