From issues-return-8315-archive-asf-public=cust-asf.ponee.io@systemml.apache.org Sat May 5 23:04:05 2018 Return-Path: X-Original-To: archive-asf-public@cust-asf.ponee.io Delivered-To: archive-asf-public@cust-asf.ponee.io Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by mx-eu-01.ponee.io (Postfix) with SMTP id 8B6D9180671 for ; Sat, 5 May 2018 23:04:04 +0200 (CEST) Received: (qmail 76242 invoked by uid 500); 5 May 2018 21:04:03 -0000 Mailing-List: contact issues-help@systemml.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: dev@systemml.apache.org Delivered-To: mailing list issues@systemml.apache.org Received: (qmail 76233 invoked by uid 99); 5 May 2018 21:04:03 -0000 Received: from pnap-us-west-generic-nat.apache.org (HELO spamd4-us-west.apache.org) (209.188.14.142) by apache.org (qpsmtpd/0.29) with ESMTP; Sat, 05 May 2018 21:04:03 +0000 Received: from localhost (localhost [127.0.0.1]) by spamd4-us-west.apache.org (ASF Mail Server at spamd4-us-west.apache.org) with ESMTP id 42583C0157 for ; Sat, 5 May 2018 21:04:03 +0000 (UTC) X-Virus-Scanned: Debian amavisd-new at spamd4-us-west.apache.org X-Spam-Flag: NO X-Spam-Score: -110.311 X-Spam-Level: X-Spam-Status: No, score=-110.311 tagged_above=-999 required=6.31 tests=[ENV_AND_HDR_SPF_MATCH=-0.5, RCVD_IN_DNSWL_MED=-2.3, SPF_PASS=-0.001, T_RP_MATCHES_RCVD=-0.01, USER_IN_DEF_SPF_WL=-7.5, USER_IN_WHITELIST=-100] autolearn=disabled Received: from mx1-lw-eu.apache.org ([10.40.0.8]) by localhost (spamd4-us-west.apache.org [10.40.0.11]) (amavisd-new, port 10024) with ESMTP id lWRIe1QqDcUl for ; Sat, 5 May 2018 21:04:02 +0000 (UTC) Received: from mailrelay1-us-west.apache.org (mailrelay1-us-west.apache.org [209.188.14.139]) by mx1-lw-eu.apache.org (ASF Mail Server at mx1-lw-eu.apache.org) with ESMTP id 4677A5F568 for ; Sat, 5 May 2018 21:04:01 +0000 (UTC) Received: from jira-lw-us.apache.org (unknown [207.244.88.139]) by mailrelay1-us-west.apache.org (ASF Mail Server at mailrelay1-us-west.apache.org) with ESMTP id 66646E010F for ; Sat, 5 May 2018 21:04:00 +0000 (UTC) Received: from jira-lw-us.apache.org (localhost [127.0.0.1]) by jira-lw-us.apache.org (ASF Mail Server at jira-lw-us.apache.org) with ESMTP id 1DAEA21296 for ; Sat, 5 May 2018 21:04:00 +0000 (UTC) Date: Sat, 5 May 2018 21:04:00 +0000 (UTC) From: "LI Guobao (JIRA)" To: issues@systemml.apache.org Message-ID: In-Reply-To: References: Subject: [jira] [Updated] (SYSTEMML-2085) Single-node parameter server primitives MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-JIRA-FingerPrint: 30527f35849b9dde25b450d4833f0394 [ https://issues.apache.org/jira/browse/SYSTEMML-2085?page=3Dcom.atlas= sian.jira.plugin.system.issuetabpanels:all-tabpanel ] LI Guobao updated SYSTEMML-2085: -------------------------------- Description: A single node parameter server acts as a data-parallel par= ameter server. And a multi-node model parallel parameter server will be dis= cussed if time permits.=C2=A0The=C2=A0idea is to run=C2=A0a single-node par= ameter server by maintaining a hashmap inside the CP (Control Program) wher= e the parameter as value accompanied with a defined key. For example, inser= ting the global parameter with a key named =E2=80=9Cworker-param-replica=E2= =80=9D allows the workers to retrieve the parameter replica. Hence, in the = context of local multi-threaded backend, workers can communicate directly w= ith this hashmap in the same process. And in the context of Spark distribut= ed backend, the CP firstly needs to fork a thread to start a parameter serv= er which maintains a hashmap. And secondly the workers can send intermediat= es and retrieve parameters by connecting to parameter server via TCP socket= . Since SystemML has good cache management, we only need to maintain the ma= trix reference pointing to a file location instead of real data instance in= the hashmap. If time permits, to be able to introduce the async and stalen= ess update strategies, we would need to implement the synchronization by le= veraging vector clock. (was: A single node parameter server acts as a data= -parallel parameter server. And a multi-node model parallel parameter serve= r will be discussed if time permits.=C2=A0) > Single-node parameter server primitives > --------------------------------------- > > Key: SYSTEMML-2085 > URL: https://issues.apache.org/jira/browse/SYSTEMML-2085 > Project: SystemML > Issue Type: Sub-task > Reporter: Matthias Boehm > Assignee: LI Guobao > Priority: Major > > A single node parameter server acts as a data-parallel parameter server. = And a multi-node model parallel parameter server will be discussed if time = permits.=C2=A0The=C2=A0idea is to run=C2=A0a single-node parameter server b= y maintaining a hashmap inside the CP (Control Program) where the parameter= as value accompanied with a defined key. For example, inserting the global= parameter with a key named =E2=80=9Cworker-param-replica=E2=80=9D allows t= he workers to retrieve the parameter replica. Hence, in the context of loca= l multi-threaded backend, workers can communicate directly with this hashma= p in the same process. And in the context of Spark distributed backend, the= CP firstly needs to fork a thread to start a parameter server which mainta= ins a hashmap. And secondly the workers can send intermediates and retrieve= parameters by connecting to parameter server via TCP socket. Since SystemM= L has good cache management, we only need to maintain the matrix reference = pointing to a file location instead of real data instance in the hashmap. I= f time permits, to be able to introduce the async and staleness update stra= tegies, we would need to implement the synchronization by leveraging vector= clock. -- This message was sent by Atlassian JIRA (v7.6.3#76005)