Sebastian Renzi’s Blog

En estos ultimos meses estuve trabajando conjuntamente con Angel "HPC" Lopez en "High Performance Computing". Especificamente con Microsoft HPC Server 2008.

A mediados de Enero estaremos dando una charla gratuita en el MUG.

Lugar: Auditorio del MUG, Rivadavia 1479 1º A, Ciudad de Buenos Aires.

Fecha y Horario: Jueves 15 de enero de 2009, en el horario de 9.30 a 12 hs.

Abstract:

Hay aplicaciones que necesitan procesar gran cantidad de información, o que tienen requerimientos de velocidad exigente.
Exploraremos en la charla algunas tecnologías y estrategías que tenemos disponibles como:
- Programación Paralela: usando múltiples threads, TPL (Task Parallel Library).
- Programación Distribuida: enviando el trabajo a realizar a distintas máquinas, usando WCF o DSS/CCR
- High Performance Computing: ejecutando job, tasks, y programa

HPC Server 2008 es el producto de Microsoft que permite armar un cluster de máquinas, pudiendo llegar a tener cientos de nodos.
Es la tendencia actual en supercomputación.

Mostraremos ejemplos de código de cada uno.

I would like to thank all the people who attended the web cast.

As we told you, you can find all the code examples shown during the session and also the presentation in the link below.

http://cid-477b63f6f5b695d3.skydrive.live.com/browse.aspx/Public/programacionEnHPCServer

Tomorrow, Angel "Java" Lopez and Sebastian Renzi will present an HPC 2008 live web cast (spanish). We will show applications running inside the cluster, using MPI.NET.

When:
2008/12/18
2.00 PM (GMT: -02:00. Buenos Aires)

Attendees URL:
http://msevents.microsoft.com/CUI/WebCastEventDetails.aspx?EventID=1032390232&EventCategory=4&culture=es-AR&CountryCode=AR

Abstract:

Gracias a Windows HPC Server 2008, tenemos un cluster de computadoras donde ejecutar aplicaciones que necesiten algo rendimientos. Esas aplicaciones se pueden desarrollar con distintos lenguajes y tecnologías. En este webcast, exploraremos las opciones de: Acceso programático a un cluster, Envío de Tasks, Jobs, MPI (Message Passing Interface), MPI.NET (librería .NET de acceso a MPI). Comentaremos ejemplos con código usando Visual Studio 2008, y MPI.NET.

See you there

Thanks to all who attended the webcast. I also want to thank @ajlopez and Southies for their support.

Here you will find the presentation we used during the webcast. Feel free to email me any doubts or questions.

  (Click the image to download the deck)

More HPC webcasts coming

December 3: HPC 2008: Instalación y Despliegue - 10 hs GMT-5 (1:00 p.m Buenos Aires)

http://msevents.microsoft.com/CUI/WebCastEventDetails.aspx?EventID=1032390230&EventCategory=4&culture=es-AR&CountryCode=AR

December  4: HPC 2008: Ejecución y Administración - 15:30 GMT-5 (6:30 p.m Buenos Aires)

http://msevents.microsoft.com/CUI/WebCastEventDetails.aspx?EventID=1032390234&EventCategory=4&culture=es-AR&CountryCode=AR

Tomorrow, Angel "Java" Lopez and Sebastian Renzi will present an HPC 2008 live web cast. We will show an introduction to Windows HPC Server 2008 and show the benefits of this product.

When:
2008/12/02
1.00 PM (GMT: -02:00. Buenos Aires)

Attendees URL:
http://msevents.microsoft.com/CUI/WebCastEventDetails.aspx?EventID=1032390223&EventCategory=4&culture=es-AR&CountryCode=AR

Abstract:
El termino HPC (High Performance Computing) se refiere a la necesidad de contar con gran potencia de computo para la resolución de problemas. Con el avance de la tecnología y el abaratamiento de los costos en hardware, hoy en día no es necesario contar con una supercomputadora e invertir grandes cantidades de dinero para obtener gran poder de computo. A partir de la aparición de "clusters", conformados por máquinas convencionales, podemos explotar al máximo la capacidad de cálculo de cada una de ellas mediante la utilización de Windows HPC Server 2008. En esta oportunidad veremos una introducción al tema de HPC, conjuntamente con las principales características de Windows HPC Server 2008.

See you there

I was really surprised when I saw HTC using Twitter for important product announcements like:

“sad news, US. we looked into it- by the time we could bring Touch HD to the states, it would be old news. we do have other cool stuff coming”

After researching a little bit I realized that HTC is not the only big company using Twitter as an official communication channel.

Comcast (USA) is giving technical support, CNN interacts with listeners through Twitter and many others like airlines, retailers and media companies.

Here you will find a list of companies using twitter in different areas, like customer support, marketing, flight status and technical support among others.

Media
Reuters http://twitter.com/reuters
CNN http://twitter.com/cnnbrk
BBC (Off-site List) http://menti.net/?p=17
NY Times http://twitter.com/nytimes 
ZDNet http://twitter.com/zdnett
CNET News http://twitter.com/CNETNews
Times OnLine http://twitter.com/TimesWorldNews
CBC News http://twitter.com/cbcnews

Airlines/Travel
Carnival Cruise Lines http://twitter.com/CarnivalCruise
Delta http://twitter.com/deltaairlines
Jetblue http://twitter.com/JetBlue
Southwest Airlines http://twitter.com/SouthwestAir
Mahalo Travel http://twitter.com/mahalotravel

Retailers
Dell Outler http://twitter.com/DellOutlet
Direct2Dell http://twitter.com/Direct2Dell
Direct2Dell (China): http://twitter.com/Direct2Dell_CN
Studio Dell http://twitter.com/StudioDell
WOOT: http://twitter.com/woot
Apple: http://twitter.com/appleinc 
Amazon.com Deals: http://twitter.com/amazondeals
Zappos http://twitter.com/zappos
Buy.com http://twitter.com/Buy_com 

Manufacturing
GM http://twitter.com/GMblogs
Ford http://twitter.com/FordCustService

If you want to start Twittering visit: http://www.twitter.com

Sebastian Renzi http://www.twitter.com/sebarenzi

What is High Performance Computing?

HPC is a branch of computer science that focuses on the processing of large amounts of information and the software that runs on it.
To achieve this aim HPC uses supercomputers or clusters.

Supercomputers:

A supercomputer is a computer with a high processing capacity, famous supercomputers are Cray or IBM RoadRunner.
Supercomputers have more performance than a cluster, especially in I/O operations. But also they are more expensive.

Cluster:

Cluster computing is the technique of linking 2 or more computer into a network to take advantage of the parallel processing.
Clusters are cheaper than supercomputers as they can be built with common computers,  maintenance is also more economical because the spare parts are common and they are available at very low cost.

As a consequence of this low costs, new scenarios for cluster usage have appeared.

Scenarios

From the beginning of HPC, the most common scenarios were related to gas/oil, climatology, finance, research, etc.
Now a day’s new industries are adopting HPC as a solution for their business needs. Digital cinema, 3D renderings are examples of these new scenarios.
Also companies like Ferrari are implementing HPC for predicting and preventing defects and gathering real-time information from vehicles on the road. You can find more information about Ferrari’s implementation here.

At the beginning, HPC implementations were very expensive and only governments or huge companies could cope that investment.
As I mentioned before, the cluster approach is cheaper and as a consequence of this, new scenarios are appearing. Good examples of these new scenarios are game servers and test environments.

Imagine your build server running in a clustered environment

Conclusions

Now we know a little bit more about HPC, the history and we also saw some of the typical scenarios and the new ones coming.
Implement an HPC cluster is not easy, neither impossible.

Next week I will post about Windows HPC Server 2008. Windows HPC Server 2008 is designed for high-end applications that require high performance computing clusters.

So stay tuned ..  coming soon …

Thanks !

Today i’m returning home from Merrill Lynch campus in New Jersey.
I’ve been working there for 1 week with Adrian Martinez.
We started integrating the application we are building with Merrill Lynch security framework.
Adrian will be working on this task for the next 2 weeks.
As always, it was an excellent week, we did an outstanding job as a team.

On June Ariel and I have been working for 2 weeks in Merrill Lynch IT campus located at Hopewell, NJ, USA.
One of the main goals for this trip, was the closure of the version 2 of the project and the
planning of version 3. During these two weeks we helped Devi’s team to showcase the tools to others
players within Merrill.

I would like to congratulate the whole team for the commitment and for the great job they
did during the 18 iterations.

We are building a great team.