Microsoft Windows HPC Server 2008 R2 for the Cluster Developer - 50291

Course Outline

Module 1: Introduction to High-Performance Computing and HPC Server 2008 R2

This module introduces the field of high-performance computing, the product Microsoft Windows HPC Server 2008 R2, and developing software for HPCS-based clusters.

Lessons

 Motivation for HPC
 Brief product history of CCS, HPCS, and HPCS 2008 R2
 Brief overview of HPC Server 2008 R2 — components, job submission, scheduler
 Product differentiators
 Measuring performance — linear speedup
 Predicting performance — Amdahl’s law

Lab : Introduction to HPC And Windows HPC Server 2008

 Submitting and monitoring jobs
 Running an HPC app
 Measuring performance
 Measuring the importance of data locality

Module 2: Developing Software for HPC Server 2008 R2

This module presents designs, technologies, and challenges when developing HPC software. It also presents the use of Parametric Sweep to solve a real-world HPC problem.

Lessons

 Design challenges
 Design patterns
 Common problem decompositions
 Common communication patterns
 Computation vs. communication
 Available HPC technologies: multi-threading, GPUs, MPI, SOA, etc.
 Data-mining and Parametric Sweep

Module 3: The HPCS Job Scheduler

This module introduces the heart of HPCS-based clusters — the Job Scheduler.

Lessons

 Throughput vs. performance
 Nodes vs. sockets vs. cores
 Jobs vs. Tasks
 Job and task states
 Default scheduling policies
 The impact of job priorities and job preemption
 Job resources and dynamic growing / shrinking
 Submission and activation filters

Lab : Working with the Job Scheduler

 Environment variables in HPC Server 2008 R2
 Exit codes and denoting success / failure
 Checkpointing in case of failure
 Multi-task jobs and task dependences

Module 4: Multicore for Performance

This module provides an overview of Microsoft’s multicore libraries available for C# and C++.

Lessons

 Parallel, multicore programming for responsiveness and performance
 Structured, fork-join parallelism
 Multicore in C# using the Task Parallel Library in .NET 4
 Multicore in VC++ using OpenMP and the Parallel Patterns Library
 Scheduling parallel, multicore apps on Windows HPC Server

Lab : OPTIONAL: Multicore Programming in C# using Task Parallel Library

 Creating a parallel, multicore app in C# and .NET 4
 Running and measuring performance locally
 Running and measuring performance on the cluster

Module 5: Interfacing with HPCS-based Clusters

This module demonstrates the various ways you can interface with Windows HPC Server 2008 R2, in particular using the HPC Server 2008 API.

Lessons

 Cluster Manager
 Job Manager
 Job Description Files
 clusrun
 Console window
 PowerShell
 Scripts
 Programmatic access via HPCS API v2.0
 Showing job progress
 Implementing job fault tolerance

Lab : Interfacing with Windows HPC Server 2008

 Clusrun is your friend
 Scripting
 Using the HPCS API to submit and monitor a job

Module 6: Introduction to MPI

This module introduces *the* most common approach to developing cluster-wide, high-performance applications: the Message-Passing Interface.

Lessons

 Shared-memory vs. distributed-memory
 The essence of MPI programming — message-passing SPMD
 Microsoft MPI
 Using MSMPI in Visual Studio with VC++
 Execution model
 MPI Send and Receive
 mpiexec
 Scheduling MPI apps on Windows HPC Server

Lab : Introduction to MPI

 Creating a simple MPI app using Send and Receive
 Running and measuring performance locally
 Running and measuring performance on the cluster

Module 7: MPI on the Microsoft Platform

This module discusses MSMPI and data parallelism, in particular how best to build data parallel MPI apps using its collective operations.

Lessons

 MSMPI: Microsoft MPI
 Data parallelism in MPI
 A real world example
 Broadcast
 Scatter
 Gather
 Barriers
 Reductions
 Defining your own reduction operator
 Common pitfalls

Lab : Data Parallelism with MPI’s Collective Operations

 Parallelizing an existing MPI application
 Mapping Sends and Receives to Broadcast, Scanner, Gather, and All_reduce
 Running and measuring performance locally
 Running and measuring performance on the cluster

Module 8: MPI Debugging, Tracing, and Performance Tuning

This module dives into the practical realities of using MPI — debugging, tracing, and performance tuning.

Lessons

 Local MPI debugging with Visual Studio 2010
 Remote MPI debugging with Visual Studio 2010
 General MPI tracing
 Tracing with ETW (Event Tracing for Windows)
 Trace visualization
 Other tools for MPI developers: perfmon, resmon, and xperf
 Common performance problems in MPI

Module 9: MPI Application Design

This module presents the most common design issues facing MPI developers.

Lessons

 Hiding latency by overlapping computation and communication
 Non-blocking communication
 Safety: detecting and avoiding deadlock
 MPI.NET
 Hybrid designs involving both MPI and OpenMP
 Buffering, error handling, I/O, and large datasets
 Remote memory access

Module 10: Intro to SOA with HPC Server 2008 R2

This module presents one of the most interesting and unique features of Windows HPC Server 2008 R2 — service-oriented HPC.

Lessons

 Service-oriented architectures
 SOA and WCF
 Mapping SOA onto Jobs and the Job Scheduler
 Private vs. shared sessions
 Secure vs. insecure sessions
 Volatile vs. durable sessions

Lab : Consuming a HPC-based SOA Service

 Deploying a SOA service
 Building a desktop client to communicate with a SOA service
 Working with volatile SOA sessions
 Working with durable SOA sessions

Module 11: Create SOA-based Apps with HPC Server 2008 R2

This module presents the details of building a SOA-based HPC app, from start to finish.

Lessons

 Service-side programming
 Service configuration
 Client-side programming options
 Proxies: Async WCF vs. HPC BrokerClient, Enumeration vs. Callback

Lab : SOA-based HPC with HPCS and WCF

 Creating a SOA service from scratch
 Deploying a SOA service
 Develop a client-side app to communicate using WCF proxy
 Develop a client-side app to communicate using HPC BrokerClient proxy
 Call a service with multiple entry points, concurrently

Module 12: SOA Debugging, Tracing, and Performance Tuning

This module discusses various performance tuning strategies on Windows for parallel apps.

Lessons

 Local SOA debugging with Visual Studio 2010
 Remote SOA debugging with Visual Studio 2010
 Low-level tracing with WCF
 Enabling WCF tracing via HPC Cluster Manager
 Common performance problems with SOA-based HPC apps
 Troubleshooting SOA services

Module 13: HPC Services for Excel 2010

This module presents techniques for bringing the potential of high-performance computing to the world of spreadsheets.

Lessons

 Excel as a computation engine
 Performing Excel computations on Windows HPC Server 2008 R2
 Using HPC Services for Excel 2010 to run workbooks on the cluster
 Using Excel UDFs to run parallel computations on the cluster

Module 14: Designing for Workstation and Azure nodes

This module discusses the design of applications that take advantage of on-premise Win7 workstations, as well as off-premise Azure nodes.

Lessons

 Additional compute resources: on-premise workstations and off-premise Azure nodes
 Taking advantage of Workstation nodes
 Taking advantage of Azure nodes

Module 15: Supporting and Emerging Technologies

This module presents brief overview of supporting and emerging technologies around HPC Server 2008 R2.

Lessons

 GPU computing with CUDA and HPC Server 2008 R2
 Virtual Shared Memory with AppFabric Caching
 Large data processing with Dryad and DryadLINQ
 Cluster data reporting via the HPC Server API
 
Additional Reading

To help you prepare for this class, review the following resources:

 http://www.microsoft.com/hpc