From owner-freebsd-performance@FreeBSD.ORG Wed Oct 1 22:37:09 2014 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id F1CE6E01 for ; Wed, 1 Oct 2014 22:37:09 +0000 (UTC) Received: from nm14-vm5.access.bullet.mail.bf1.yahoo.com (nm14-vm5.access.bullet.mail.bf1.yahoo.com [216.109.115.20]) (using TLSv1 with cipher ECDHE-RSA-RC4-SHA (128/128 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 6089D831 for ; Wed, 1 Oct 2014 22:37:08 +0000 (UTC) Received: from [66.196.81.157] by nm14.access.bullet.mail.bf1.yahoo.com with NNFMP; 01 Oct 2014 22:33:42 -0000 Received: from [66.196.81.132] by tm3.access.bullet.mail.bf1.yahoo.com with NNFMP; 01 Oct 2014 22:33:42 -0000 Received: from [127.0.0.1] by omp1008.access.mail.bf1.yahoo.com with NNFMP; 01 Oct 2014 22:33:42 -0000 X-Yahoo-Newman-Property: ymail-3 X-Yahoo-Newman-Id: 280995.81749.bm@omp1008.access.mail.bf1.yahoo.com Received: (qmail 32626 invoked by uid 60001); 1 Oct 2014 22:33:41 -0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sbcglobal.net; s=s1024; t=1412202821; bh=8KN31nr9unSZfR72oDas2370zUEt9RJRucGbDSD4XZ0=; h=Message-ID:Date:From:Reply-To:Subject:To:MIME-Version:Content-Type; b=jEUPIS01i3el95HQ1YzvBNkiQbk8p2910ghtVVYPrayzt70uF+X4nCvhdImqLh7Butp+7ASroG7+0CcfW3k8h1xzUMeLcjoTJrQdP1yehXAevVXr723ywhvFseIIg2sXJS6SvNvjEhuyHguA9pT3xVHKzungYjkHtteHcwHoy3c= X-YMail-OSG: bAg75CoVM1njL6GXoP3ALn6XC7ceVnVp_8A85pCJTnQXRag tIbQNiksCKH8.mxqyjpfdzQZqageiSzRFuHfBwT0sMKeVSxUOwUlNyF3DAVz iFaidjeGo8aKLtJA_Clz46Gdv5vMZHBxW1MiU3vmND6q7ed.KR7d4vhX.yZV nQx.KUAHagmWea3nmKrDdbUIGNGipvPJ5fd7W8wGoc2pmQ21Uvtt5EzqMBEt p_yFibBgCv3nkrZNdcrWJ5prw1kkWCMaYbmdZAK02BToVSPVyWfinHcK0.S4 R2q8mHp5vlauDIa6VEBJT2mZNlZ33FbI3SiadVBknhWfeKOac_CN.SOiRVvh 5OLjCxaZIfuJIdi8c_2E2XdcOYzJI1y2DvA5dD.pNVTc.gYpwphRSMupIqlp bag9YhFbW1QiYipqJFhLEt_pNYNEicOlpRlf2v.Z7kUejXyaLXgPICYLMzY0 k0XzMBIMcEX_m.ZDpR7HLPfbDdva2qvIWQ20uqRrJsqROZ9M9.akVfk.bXdH 49GQhL_Z_1PrCW7N75VO0vY27Ja.JNq.1SVG9VKmWl_RWVkJcHZSKWs7Ved5 XYj7GTZdD0wNWs4Y- Received: from [162.239.0.170] by web180904.mail.ne1.yahoo.com via HTTP; Wed, 01 Oct 2014 15:33:40 PDT X-Rocket-MIMEInfo: 002.001, UnVuIHNvbWUgY29tcGFyaXNvbiB0ZXN0IG9uIDEwLjEtQkVUQTIgYW5kIHNlZSBzb21lIHBlcmZvcm1hbmNlIGRyb3AgZXNwZWNpYWxseSB0aGUgRk9SSyBhbmQgVVNCIDMgcmVhZC4KCkF0dGFjaGVkIGlzIGEgcGVyZm9ybWFuY2UgY29tcGFyaXNvbiBwYWdlLiBDb2x1bW4gMiBhbmQgMyBhcmUgdXNpbmcgdGhlIHNhbWUgaGFyZHdhcmUgKEFNRCBBOC02NjBLSyArIEFTVVMgQTg4WE0tQSBNL0IpIGJ1dCBydW5uaW5nIDguNCBhbWQ2NCBhbmQgMTAuMUJFVEEyIChoaWdobGlnaHRlZCBhcmVhKSBhbWQ2NCBrZXIBMAEBAQE- X-Mailer: YahooMailWebService/0.8.203.696 Message-ID: <1412202820.67636.YahooMailNeo@web180904.mail.ne1.yahoo.com> Date: Wed, 1 Oct 2014 15:33:40 -0700 From: Jin Guojun Reply-To: Jin Guojun Subject: fork and USB3 performance drop in 10.1-BETA2 To: "freebsd-performance@freebsd.org" MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Content-Filtered-By: Mailman/MimeDel 2.1.18-1 X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 01 Oct 2014 22:37:10 -0000 Run some comparison test on 10.1-BETA2 and see some performance drop especially the FORK and USB 3 read. Attached is a performance comparison page. Column 2 and 3 are using the same hardware (AMD A8-660KK + ASUS A88XM-A M/B) but running 8.4 amd64 and 10.1BETA2 (highlighted area) amd64 kernels. Row 3 in highlighted area shows multi-thread performance is still excellent but rest rows show performance drop especially everything below the multi-threading test (SMP MT). Very last row shows that USB3 reading performance dropped to 45-55% (writing is OK). This was tested with four different USB3 drives. Above that row, it shows fork performance -- 4 processors only gain 199% (24.6s) throughput (supposed close to 390+%), and two processors gained about 50% more (37.06s) compare single core another line above (48.28s). The single core performance is also drop compare to 8.4 kernel on left. Has anyone seen similar performance issue? or 10.1 requires some different turning trick? -Jin