eScience @ Microsoft : Parallel Computing

Adding quarters to the innovation machine

eScience — Tue, 09 Jun 2009 21:12:15 GMT

I really enjoyed Dan Reed’s latest posting - HPC: Making a Small Fortune - and the need for real innovation in this space, especially with the rise of new technologies – that’s part of the reason we looked at at combining HPC and Databases via the GrayWulf Project…

HPC: Making a Small Fortune

N.B. I also write for the Communications of the ACM (CACM). The following essay recently appeared on the CACM blog.

There is an old joke in the high-performance computing community that begins with a question, "How do you make a small fortune in high-performance computing?" There are several variations on the joke, but they all end with the same punch line, "Start with a large fortune and ship at least one generation of product. You will be left with a small fortune." Forty years of experience, with companies large and small, has confirmed the sad truth of this statement.

Reed's Ruminations: A Blog by Dan Reed: HPC: Making a Small Fortune

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

Windows HPC Server 2008 cracks Top 10 in Top 500 list

eScience — Tue, 18 Nov 2008 08:34:59 GMT

The latest Top 500 list was released at Supercomputing Conference (SC08) in Austin, Texas today and Windows HPC Server 2008 was part of the #10 placed machine [Dawning 5000A] from Shanghai Supercomputer Center. This is the top machine outside of US – amazing job by the Windows HPC team.

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

HPC Server 2008 Experience Counts Academic Program

eScience — Tue, 11 Nov 2008 00:49:34 GMT

Great opportunity to get a special academic discount on Windows HPC Server 2008.

Windows® HPC Server 2008 Experience Counts Academic Program

You know the value of high-performance computing (HPC). No matter the topic - the inner workings of the universe or the economy; the spawning of a hurricane or urban development - HPC has become a key driver of the type of research that leads to groundbreaking insights.

Microsoft Higher Education - Experience Counts

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

HPC - MPI.NET 1.0 Is Now Released...!

eScience — Wed, 08 Oct 2008 21:12:04 GMT

PhilPen has posted that the folks at Indiana University have released MPI.NET: High Performance C# library for Message Passing. The runtime and source code are available for download as well as a Tutorial. This should make it much easier for folks to use any .Net language to write MPI apps – I’m interested in seeing ones written with F# and even PowerShell.

MPI.NET 1.0 Is Now Released...!

MPI.NET is a high-performance, easy-to-use implementation of the Message Passing Interface (MPI) for Microsoft's .NET environment. MPI is the de facto standard for writing parallel programs running on a distributed memory system, such as a compute cluster, and is widely implemented. Most MPI implementations provide support for writing MPI programs in C, C++, and Fortran. MPI.NET provides support for all of the .NET languages (especially C#), and includes significant extensions (such as automatic serialization of objects) that make it far easier to build parallel programs that run on clusters.

MPI.NET has been developed by the research staff at Indiana University in collaboration with Microsoft. Developers leverage the "Windows HPC Server 2008 SDK" in tandem with the MPI.NET SDK to build MPI.NET applications. An MPI.NET runtime component must be installed onto Windows HPC Server 2008 based clusters to host MPI.NET applications.

Regarding Windows Server : MPI.NET 1.0 Is Now Released...!

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

Scientific Discovery through Advanced Computing Program (SciDAC) tutorials on Friday, July 18th

eScience — Sat, 12 Jul 2008 02:08:35 GMT

Microsoft Research is hosting the DOE 2008 Scientific Discovery through Advanced Computing Program (SciDAC) tutorials on Friday, July 18^th. If interested in attending, please register at the SciDAC Tutorials Registration.

SciDAC Tutorial Agenda

July 18 – Microsoft Research, Bldg 99.

9:00 - 9:30

Reception, checkin and orientation

9:30 - 12:00

First session

Visualization and Data Analysis with VisTrails

Introduction to Scientific Workflow Management and the Kepler System

Porting and Scaling Applications on BlueGene/P

Advanced CompuTational Software (ACTS) Collection Tutorial

12:00 - 1:00

Lunch in bldg. 99 Atrium
(attendees in bldg. 117 walk to bldg. 99 next door)

1:00 - 3:30

Second session

Advanced Visualization and Data Analysis with the VisIt Visualization System

ADIOS The ADaptible IO System

An Introduction to the Cray XT4 for Application Scientists

Advanced CompuTational Software (ACTS) Collection Tutorial

Tutorial Abstracts

<></>

Title:

Introduction to Scientific Workflow Management and the Kepler System

Contact:

Ilkay Altintas

Partners:

The Scientific Data Management Center (SDM)

Presenters: Ilkay Altintas, Scott Klasky, Norbert Podhorszki, Mladen Vouk

A scientific workflow combines data and processes into a configurable, structured set of steps that implement semi- automated computational solutions of a scientific problem. Scientific workflow systems provide a framework and often graphical user interfaces to combine different technologies along with efficient methods for using them. This reduces overhead and increases the efficiency of the scientists towards a scientific discovery. SciDAC SDM Center's Scientific Process Automation group develops scientific workflow tools (called Kepler) for automation of scientific data management processes in a reusable, robust, tractable, and recoverable fashion to enhance scientific exploration. This tutorial provides an introduction to scientific workflow construction and management using the Kepler system. It is intended for the SciDAC scientists and students with a computational science background. It will cover principles and foundations of scientific workflows, Kepler environment installation, workflow construction out of the available Kepler library components, and workflow execution management that uses Kepler-based facilities to provide process and data monitoring, provenance information, portal access, and high speed data movement solutions. This tutorial includes hands-on sessions and application examples from different scientific disciplines.

<></> <></>

Title:

ADIOS: The ADaptible IO System

Contact:

Chen Jin

Partners:

The Scientific Data Management Center (SDM) and The Center for Plasma Edge Simulation and Oak Ridge National Lab

Presenters: Chen Jin, Scott Klasky, Steve Hodson: ORNL, Hasan Abbasi, Jay Lofstead, and Matthew Wolf: Georgia Tech, and Manish Parashar Rutgers

The ADaptible IO System (Adios) is a componentization of the IO layer for high performance scientific computations. Currently, we have integrated ADIOS into Fusion, Combustion and other pioneering scientific applications and achieved excellent performance results. ADIOS provides a simple to use, common programming interface for different transport layer methods. The design goals are set to allow the application scientist to program a very easy to use IO system, which can choose between different IO implementations at runtime. This allows IO experts to tune their implementations on different systems without changing the IO implementation. We currently support MPI-IO, collective MPI-IO, POSIX, asynchronous MPI-IO, the Georgia Tech DataTap system based on RDMA, and the Rutgers DART implementation also based on RDMA. ADIOS separates out the metadata and methods, by using an external XML file. This allows users to add annotations outside of their F90/C/C++ codes and to change the implementation outside of their code. ADIOS 1.0 will be released before 12/1/2008 and will provide very fast writes of data. In this tutorial, we will first present an introduction of ADIOS, along with the current supported methods, and file converters. We will then provide a hands-on session which will allow users to work with both a F90 and C code, and change the IO to ADIOS, and change this to use both MPI-IO and Posix IO. We will then show them how to read in the data, and convert the output to HDF5, NetCDF, and ASCII.

<></> <></>

Title:

Porting and Scaling Applications on ALCF’s BlueGene/P

Contact:

Kalyan Kumaran, Argonne Leadership Computing Facility

Partners:

The Argonne Leadership Computing Facility

Presenters: Vitali Morozov and Ray Loy

The Argonne Leadership Computing Facility houses a half a petaflop BlueGene/P supercomputer. A number of petascale applications, from a variety of engineering domains, are run every day on this highly scalable architecture via the Department of Energy's INCITE (Innovative and Novel Computational Impact on Theory and Experiment) program. Nevertheless, porting and scaling applications onto petascale computers remains a challenging task. This tutorial will focus on the BlueGene/P architecture and present an overview of compilers, libraries, performance tools, debuggers available for porting and scaling applications.

<></> <></>

Title:

Advanced Visualization and Data Analysis with the VisIt Visualization System

Contact:

Hank Childs

Partners:

The Visualization and Analytics Center for Enabling Technology (VACET)

Presenters: Hank Childs and Sean Ahern

Understanding scientific datasets generated at DOE's high-performance computing facilities is becoming increasingly difficult as dataset sizes and complexity grow beyond the scale that is approachable by traditional analysis techniques. VACET is delivering scalable solutions to real-world visualization and analysis problems through the VisIt visualization system. VisIt is a turnkey application for data exploration, visualization, code assessment, and quantitative analysis suitable for use on SciDAC datasets of any size. This tutorial is targeted at all levels of users, from first timers on up to VisIt experts. The tutorial will present the basics of using VisIt on up to advanced visualization techniques.

<></> <></>

Title:

Visualization and Data Analysis with VisTrails

Contact:

Claudio Silva

Partners:

The Visualization and Analytics Center for Enabling Technology (VACET)

Presenters: Claudio Silva and Carlos Scheidegger

The demand for the construction of complex visualizations is growing in many disciplines of science, as scientists are faced with ever increasing volumes of data to analyze. We give an overview of VisTrails, an open-source provenance management system that provides infrastructure for data exploration and visualization. VisTrails transparently records detailed provenance (history) information for exploratory computational tasks, both for the derived data products and for the pipelines used to derive them. Besides enabling the reproducibility of results, this provenance information can be used to simplify the process of data exploration through visualization. We will present several mechanisms and intuitive interfaces provided by VisTrails that allow flexible re-use of pipelines; exploration of large parameter spaces; comparison of visualizations and their respective pipelines; and the creation and refinement of visualizations by analogy.

This tutorial is targeted to users with different levels of expertise, from novice to expert. We will cover a number of specific examples that reflect typical visualization needs of DOE applications.

<></> <></>

Title:

An Introduction to the Cray XT4 for Application Scientists

Contact:

Richard Gerber

Partners:

NERSC and ORNL

Presenters: Richard Gerber and Rebecca Hartman-Baker

This tutorial provides the basics to get up and running on the Cray XT4. Topics covered include compilers, libraries, and job management with a focus on parallel scaling and performance optimization of real world applications. The presenters are HPC consultant staff with hands on experience with the XT4 systems at ORNL and NERSC. This tutorial is targeted at researchers transitioning to the XT4 from serial or other parallel computing environments.

<></> <></>

Title:

Advanced CompuTational Software (ACTS) Collection Tutorial

Contact:

Tony Drummond

Partners:

ACTS and TAU

Presenters: Tony Drummond

This short course will be an introduction to a set of advanced computational software tools to leverage the development of high performance applications. The lectures will focus on the selection, installation and use of scalable and robust software tools. Functionalities implemented in these software tools include; numerical algorithms for the solution of large computational problems, performance monitoring and profiling, and automatic tuning. Participants should expect to learn about techniques used to solve common computational problems and monitor their performance. Participants are encouraged to bring laptop computers and follow live demonstrations through hand-on experiences. The software presented here is freely available and widely used by the computational sciences international community.

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

Windows HPC Server 2008 Beta 2 is Here

eScience — Fri, 23 May 2008 18:32:47 GMT

Windows HPC Server 2008 Beta 2 is available on Microsoft Connect and Ryan Waite goes into detail on much of the new stuff. I'm really excited to start seeing scientists utilize and evaluate the new version - I think many of the new pieces especially Network Direct, PowerShell integration, and the HPC Basic Profile will surprise users. The other piece I'm looking forward to is the release of the next Top500 list in June...

We completed a few great Top500 runs in the last few weeks. We can’t talk about the numbers until the International Supercomputing Conference in June but it looks like Beta 2’s new MPI stack and new Network Direct RDMA interface are starting to hum.
Windows HPC Server 2008 Beta 2 is Here

HPC Server 2008 Beta Technical Overview

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

SciDAC Tutorials - July 18 at Microsoft Research

eScience — Tue, 29 Apr 2008 21:44:42 GMT

On July 18th, Microsoft Research will be hosting the tutorials for the 2008 SciDAC Conference. SciDAC - Scientific Discovery through Advanced Computing is part of DOE's Office of Science research program to "support multidisciplinary SciDAC projects aimed at developing future energy sources, studying global climate change, accelerating research in designing new materials,improving environmental cleanup methods, and understanding physics from the tiniest particles to massive supernovae explosions". It brings together computational scientists, applied mathematicians, and computer scientists from across application domains and from universities and national laboratories across the United States.

Tutorials are open to all, and students and local industry are encouraged to attend - please register to get a spot. The list of tutorial topics are one the webpage.

SciDAC Tutorials

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

Silverlight-based cycle stealing

eScience — Mon, 28 Jan 2008 23:12:20 GMT

Savas and I had talked about this idea of using Silverlight-based cycle stealing and wondered how well it would work. It's good to see this article on CodeProject about Legion: Build your own virtual super computer with Silverlight by Daniel Vaughan.

Legion is a Grid Computing framework that uses the Silverlight CLR to execute user definable tasks. Legion uses an ASP.NET application and web services to download tasks, upload result data, and provide grid-wide thread-safe operations for web clients or agents. Multiple tasks can be hosted at once, with Legion managing the delegation of tasks to agents. Client performance metrics, such as bandwidth and processor speed, may be used to tailor jobs for clients. Legion provides a management service and WPF application that is used to monitor the Legion grid.
I have deployed Legion to a demonstration server here so you can see it in action.

I wonder if Daniel is aware of the previous Legion grid system by Andrew Grimshaw that turned into Avaki (now part of Sybase)

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)

Parallel Computing Developer Center

eScience — Mon, 28 Jan 2008 21:37:24 GMT

There's a Parallel Computing Developer Center up on MSDN that is great resource for finding ways to solve parallel computing challenges. I've highlighted a couple of the interesting articles that scientists/researchers will find interesting in dealing with manycore and clusters. One I'm really interested in is how PLINQ (Parallel Language Integrated Query) can help scientists get data from SQL or other datasources efficiently.

Parallel Computing

Microsoft’s Parallel Computing developer center is dedicated to providing information, ideas, community, and technology to developers to make it easier to write programs that perform and scale well on parallel hardware.

The Manycore Shift: Microsoft Parallel Computing Initiative Ushers Computing into the Next Era
The computational power of manycore processors, new programming models and platform, and advanced research in usability promise to change the way people interact with computers. This paper describes how Microsoft and industry partners are working together to enable businesses, software and hardware vendors, and individuals to take advantage of the “manycore shift.”

Download Microsoft Parallel Extensions to the .NET Framework 3.5 December 2007 Community Technology Preview
Parallel Extensions to the .NET Framework is a managed programming model for data parallelism, task parallelism, and coordination on parallel hardware unified by a common work scheduler. Parallel Extensions makes it easier for developers to write programs that scale to take advantage of parallel hardware—providing improved performance as the numbers of cores and processors increase—without having to deal with many of the complexities of today’s concurrent programming models. Parallel Extensions provides library-based support for introducing concurrency into applications written with any .NET language, including but not limited to C# and Visual Basic.
Parallel LINQ
“Declarative Data Parallelism” with the Parallel Extensions is achieved through Parallel LINQ. This video explains how to parallelise your LINQ queries, how it works under the covers and how to configure it for advanced scenarios.

Parallel Computing Developer Center

Cross Posted from Dan Fay's Blog (http://blogs.msdn.com/dan_fay)