Hi Vaijanath I think you should investigate the 2D blocked algorithms for efficient operation of matrices. (Hama may follow the general interface designed by other matrices operation packages, such as, scalapack, octave, etc.)
And, I think we need to focus on the least network cost model than the least time complexity of operation model because the conventional approach was based on a few high-end machines, yet we are trying to achieve the performance by serveral thousands of computers. - Edward On Wed, Jul 23, 2008 at 2:03 PM, Vaijanath N. Rao <[EMAIL PROTECTED]> wrote: > Hi Ian, > > I think that would be the best place to start with. Also if I can know what > needs to be learned to know Hama better other than map-reduce, it would be > an great asset for me. > > --Thanks and Regards > Vaijanath > > > Ian Holsman wrote: >> >> Hey Vaijanath. >> >> good to see a fellow AOL'er ;-) >> another idea would be to integrate this into some of the clustering >> implemenations you are working on. >> >> Regards >> Ian >> >> Chanwit Kaewkasi wrote: >>> >>> Hello Vaijanath, >>> >>> Basically, if you'd like to contribute to the core of Hama, you need >>> to understand how map-reduce works, and how to implement map-reduce >>> algorithms. >>> >>> Also, you need some basic background on Hadoop and Hbase. >>> But I'm sure if you get your hands dirty with Hama code for a while, >>> it'll help you understand those two as well. >>> >>> However, there are also other parts of the project, for example >>> contributing a GUI / a shell for Hama, etc. >>> Any other cool ideas also welcome. >>> >>> Best regards, >>> >>> Chanwit >>> >>> 2008/7/23 Vaijanath N. Rao <[EMAIL PROTECTED]>: >>> >>>> >>>> Hi All, >>>> >>>> What would be the best way to start contributing and volunteering for >>>> this >>>> project. I would also like to know what all tools and knowledge would be >>>> required for Hama. >>>> >>>> --Thanks and Regards >>>> Vaijanath >>>> >>>> >>> >>> >> >> > > -- Best regards, Edward J. Yoon [EMAIL PROTECTED] http://blog.udanax.org
