From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Monjalon Subject: Re: [PATCH] app/test: reduced duration of red_autotest Date: Wed, 08 Jun 2016 18:22:52 +0200 Message-ID: <2506582.jAQRcmA1gs@xps13> References: <1465307358-115427-1-git-send-email-tomasz.kantecki@intel.com> <5309764.cBvyyiAfAy@xps13> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7Bit Cc: dev@dpdk.org To: "Kantecki, Tomasz" Return-path: Received: from mail-wm0-f45.google.com (mail-wm0-f45.google.com [74.125.82.45]) by dpdk.org (Postfix) with ESMTP id 67017BDB4 for ; Wed, 8 Jun 2016 18:22:54 +0200 (CEST) Received: by mail-wm0-f45.google.com with SMTP id m124so24526334wme.1 for ; Wed, 08 Jun 2016 09:22:54 -0700 (PDT) In-Reply-To: List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" 2016-06-08 15:56, Kantecki, Tomasz: > Hi Thomas, > > > > 12s is long for a functional test. We are used to have tests of less than 1s. > > > Do you think this magnitude order can be achieved for red tests? > > > > This is a real challenge :) Let me take another look into it. > > I had a 2nd look into this test suite and reducing time for functional tests is doable but labor intensive. Let me explain: > RED/WRED algorithm is time based (packet queue events in time). For tracking time the test uses TSC (x86 time stamp counter). This could be replaced with some fake time stamp in functional tests and result in significant reduction of execution time. It needs couple of days to carefully replace TSC with a fake time stamp solution without breaking test logic. Sorry, I haven't dived into the code, but I don't understand how we get some seconds delays in a code for fast processing on nano-second CPU. Are you sure you cannot have the same result with less iterations and shorter timeouts?