[R-sig-hpc] Advice on HPC/R needs for advanced computing oversight committee

[David J. Vanness]

Dear Suzen,

A belated thank-you for your thoughtful reply and useful links.  I am
bookmarking your email for what is likely to be repeated future use!

Best regards,

Dave Vanness

-----Original Message-----
From: mehmet.suzen at gmail.com [mailto:mehmet.suzen at gmail.com] On Behalf Of
Suzen, Mehmet
Sent: Friday, January 11, 2013 4:10 PM
To: Etienne B. Racine
Cc: David J. Vanness; r-sig-hpc at r-project.org
Subject: Re: [R-sig-hpc] Advice on HPC/R needs for advanced computing
oversight committee

Hi David,

Updating packages or the base should not be an issue as long as deployment
and development platforms are identical. But that is a valid problem for any
software project, it is too generic to be Rs problem only.

>> nascent advanced computing initiative.  What hardware issues do you 
>> see as most critical?  Software?  What kinds of consulting resources 
>> would you want to have available to facilitate your R-related work?  
>> Any recommendations

Step 0 would be to first start exploring current HPC capabilities.
There is a very well
maintained classification in views:
http://cran.r-project.org/web/views/HighPerformanceComputing.html

Each solution address to different architecture. I would take each sub-title
there and generate "minimal" hardware requirement for each.
So It is all depends on your needs and applications.

Concerning consultancy, there are very successful R companies (I don't want
to advertise names; ask me off the list) that provides off the shelf
parallel R solutions, optimized packages/R-core and support.

>> for introductory material on high performance computing in R that I 
>> should read to get up to speed?  Thanks in advance for any
comments/advice.

Probably this book came up  in the list before;  Parallel R  :
http://shop.oreilly.com/product/0636920021421.do

Also this resource list is pretty good

http://rdatamining.wordpress.com/2012/05/06/online-resources-for-handling-bi
g-data-and-parallel-computing-in-r/

I have noticed that you work with MCMC. The parallel R solutions guru on
MCMC is Professor Wilkinson of Newcastle University. His posts are quite
pedagogical:

http://darrenjw.wordpress.com/2011/12/29/parallel-particle-filtering-and-pmc
mc-using-r-and-multicore/