Mailing-List: contact hama-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hama-dev@incubator.apache.org
Message-ID: <1167122244.1228400144388.JavaMail.jira@brutus>
Date: Thu, 4 Dec 2008 06:15:44 -0800 (PST)
From: "Samuel Guo (JIRA)" <jira@apache.org>
To: hama-dev@incubator.apache.org
Subject: [jira] Commented: (HAMA-121)  default maximum block size
In-Reply-To: <1414468690.1228375664624.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HAMA-121?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12653308#action_12653308 ] 

Samuel Guo commented on HAMA-121:
---------------------------------

I think we may consider the cpu's cache size to decide the default maximum block size.
A suitable block size will reduce cpu cache hit misses during *SubMatrix* multiplication. Or we may change the *mult* method's implementation of *SubMatrix* to match a cpu's cache working pattern.

I think we may do some experiments over *SubMatrix* to know a suitable size.

>  default maximum block size
> ---------------------------
>
>                 Key: HAMA-121
>                 URL: https://issues.apache.org/jira/browse/HAMA-121
>             Project: Hama
>          Issue Type: Improvement
>          Components: implementation
>            Reporter: Edward J. Yoon
>             Fix For: 0.1.0
>
>
> Using  'mapred.child.java.opts', we can calculate the default maximum block size.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.