Come hear why customers are moving to the AWS Cloud for HPC Workloads at the ISC High Performance Conference.
In our booth at ISC – 2017 Amazon Web Services will be showcasing our portfolio of services that are suited towards HPC, these will include:
- Tuned Compute Services for HPC (EC2, F1, P2, AWS Batch, CFN Cluster)
- High Performance Data Analytics
- Data Lakes
- Machine Learning/AI
- Performance Storage options (S3, EFS, Lustre, BeeGFS etc)
- Remote Graphics (Nice DCV, Elastic Graphics etc)
- Near Serverless Infrastructure & Agile Deployment (Lambda, ECS, Cloud Formation)
AWS, along with several of our partners, will be providing mini sessions in our booth G-822 covering all aspects relevant to Cloud HPC.
03:00pm - 03:30pm |
OpenHydro, world leaders in Tidal Turbine Technology, use ANSYS CFD running on AWS clusters, to optimise turbine power generation in a range of tidal conditions. Company: Cadfem Speaker: Derek Sweeney, Managing Director |
04:00pm - 04:30pm |
AWS and FlyElephant platform for Data Science, simulation engineering, and rendering tasks. In the presentation, we will describe how the FlyElephant platform and AWS can be used for Data Science, simulation engineering, and rendering tasks. We will give a comprehensive insight into accelerating the creation, deployment, and maintenance of data science tasks in AWS using the FlyElephant platform. The experience of finding human resources for complex projects will be covered as well as the feature that FlyElephant provides to solve this problem. Company: Fly Elephant Speaker: Dmitry Spodarets, CEO & Founder at FlyElephant |
05:00pm - 05:30pm |
myHPC - Hybrid Cloud Solution for HPC/Technical Computing How can onsite HPC/Technical Computing workloads seamlessly extend into on-demand HPC cloud resources including transparent job and cluster control as well as data management? With the myHPC Hybrid HPC cloud solution based on EnginFrame HPC cloud-based HPC is one click away. Company: Nice Software Speaker: Karsten Gaier, Alliances Manager |
06:00pm - 06:30pm |
BeeGFS - The fast and flexible parallel file system With the ever-growing data sets that need to be analyzed in almost all areas of scientific research and industry, scaling I/O performance has become a big challenge. BeeGFS has been designed to tackle this problem for the most demanding applications, based on an extremely leight-weight and highly optimized architecture that exploits the specific capabilities of traditional disk-based storage just as well as modern all-flash storage. To enable data-intensive computing on AWS, BeeGFS is now available on the AWS Marketplace in different flavors, either directly for your on-demand storage cluster or integrated into the Alces Flight HPC environment. Due to its special focus on performance and flexibility, BeeGFS is being used all around the globe to provide extremely fast access to storage systems of all kinds and sizes, from small scale up to enterprise-class systems with thousands of hosts and including some of the fastests supercomputers in the world Company: BeeGFS Speaker: Sven Breuner, CEO |
09:00am - 09:30am |
Traditional HPC is focused on processing complex models which either read or produce massive datasets. Big Data is all about processing massive datasets, but often more focused on low quality or un/semi-structured data. In this talk, we'll look at how innovative business are unifying their application stacks around algorithmic modelling at scale, with massive post processing jobs running within the same environments. You'll learn how specific AWS technologies can be leveraged to create high performance applications which are both compute and data intensive, where instead of making large up front investments, you only pay for the capacity you consume. Company: AWS Speaker: Ian Meyers, Specialist Solution Architect, Sr Manager |
11:00am - 11:30am |
The next generation of grid computing on AWS What are the top requirements for the next generation grid computing solution:
In this talk, Teppo Tammisto will introduce how the Techila Distributed Computing Engine has enabled a customer to run their previously several months long simulations in two hours, and a secure and compliance-ready integration of AWS Cloud. He will also show a live demonstration how to run a MATLAB simulation using hundreds of Amazon EC2 instances without need to invest in MATLAB Parallel Computing Toolbox or MATLAB server licenses for the computing environment. Company: Techila Speaker: Teppo Tammisto, CTO |
12:00pm - 12:30pm |
Cost and time to solution of HPC workloads are important measures used to determine many HPC choices today. A clear understanding of the factors that determine these helps optimize both. We will show how Allinea Performance Reports provides data and insight to enable tuning of instance type selection, data layout and code that can reduce both cost and time of deploying HPC to Amazon Web Services. Company: ARM Speaker: Florent Lebeau, Staff Applications Engineer |
01:00pm - 01:30pm
|
Description: This talk describes how a Cycle Computing customer used HTCondor to manage Docker containers in AWS. The flexibility of Docker containers with HTCondor and the scalability of the public cloud made it possible for this customer to speed up a tedious and inflexible EDA model development process. This allowed them to increase productivity throughput, and reduce overall time to results dramatically Company: Cycle Computing Speaker: Tim Carroll, VP - Sales & Eco-System |
02:00pm - 02:30pm |
Data Lake Best Practices The flexibility of AWS enables you to tailor your approach for architecting a Data Lake. AWS provides you with secure infrastructure and offers a number of scalable, cost-effective storage, data management, and analytics services to turn heterogeneous data into new meaningful insights. Hosting your Data Lake on AWS gives you access to the most complete platform for Big Data. Furthermore, a Data Lake isn’t meant to be replace your existing Data Warehouses, but rather complement them. If you’re already using a Data Warehouse, or are looking to implement one, a Data Lake can be used as a source for both structured and unstructured data, which can be easily converted into a well-defined schema before ingesting it into your Data Warehouse. A Data Lake can also be used for ad hoc analytics with unstructured or unknown datasets, so you can quickly explore and discover new insights without the need to convert them into a well-defined schema. In this talk, we will cover best practices and reference architecture and cover the key tenants of a Data Lake such as: (a) Decoupling data from compute, (b) Collecting and storing any type of data, at any scale and at low costs, (c) Securing and protecting all of data stored in the central repository, (d) Searching and finding the relevant data in the central repository, (e) Quickly and easily performing new types of data analysis on datasets and (f) Querying the data by defining the data’s structure at the time of use (schema on read). Company: AWS Speaker: Pratim Das, Specialist SA – Analytics, EME |
02:30pm - 03:00pm |
Cloud 3.0, The Rise of Big Compute In 2017, the enterprise software industry is at an inflection point for ubiquitous cloud adoption as part of a $4 trillion enterprise IT market transformation. The reasons for enterprise IT to adopt cloud are compelling, and many assume this shift has already happened, but enterprise cloud adoption is only around 6% today and most businesses have yet to realize the full value of this transformation to their bottom line. In this presentation, we'll review the growth and increasing acceptance of cloud through 1.0, SaaS (software as a service), where companies have redefined enterprise categories by innovation and frictionless delivery, through 2.0, IaaS (infrastructure as a Service), where cloud scaling and connectivity has enabled big data analytics and finally to 3.0, big compute, where heavy duty cloud resources are starting to eliminate constraints on high-performance computing (HPC) and democratizing access across the enterprise Company: Rescale Speaker: Wolfgang Dreyer, EMEA Director |
03:00pm - 03:30pm |
Delivering powerful graphics-intensive applications from the AWS Cloud AWS provides unprecedented computational power for graphics-intensive applications in areas such as design, engineering simulations, and 3D content rendering. Together, Amazon EC2 Elastic GPUs and Amazon AppStream 2.0 provide the capabilities necessary for end users to access and run these applications. In this session, you learn more about Elastic GPUs and Amazon AppStream 2.0, and how you can run graphics-intensive applications on AWS. Company: AWS Speaker: Tosh Tambe, Head if Integration - Nice |
03:30pm - 04:00pm |
Aerospace HPC on EC2 Zenotech have extensive experience of running Aerospace HPC workloads on the AWS platform. We have a strong background in CFD applications but are also involved in the wider HPC workflows. In this talk we will discuss our approach and experience of doing this, present our view of how best to make use of cloud resources, and present use cases from some of our customers. Company: Zenotech Speaker: Mike Turner, Director |
04:00pm - 04:30pm |
AWS - HPC Building Blocks AWS breadth of services offers large opportunities to build innovative infrastructures. In this talk, we’ll look at the building blocks that can be leveraged to build a efficient, scalable simulation environment, taking into account long-term storage needs, data-migration, security, remote visualisation and integration to entreprise IT. Company: AWS Speaker: Arthur Petitpierre, Specialist Solution Architect |
04:30pm - 05:00pm |
Lessons learned in using Cloud BigCompute to transform research This talk will focus on how scientists, engineers, and researchers are leveraging the Big Compute capabilities of the public cloud to perform larger, more accurate, and more complete workloads than ever before. Real world use cases where big pharma, materials science, and manufacturing researchers accelerated science using 160 to 160,000 cores on cloud. Attendees will leave with a clear understanding of where cloud resources can, or can not, help their work. Company: Cycle Computing Speaker: Tim Carroll, VP - Sales & Eco-System |
05:00pm - 05:30pm |
I/O profiling: What to look for when migrating applications and data to a cloud environment Bad I/O patterns can sometimes be masked by fast storage or other bottlenecks, but become obvious when those variables change. When migrating applications to new compute environments such as the cloud, it's important to make sure that your applications are not hiding any I/O problems that can prevent you from taking advantage of the performance improvements or cost savings of the new system. We'll be looking at how the I/O profiles of big HPC benchmarks change when they are run on different compute environments. The talk will highlight possible improvements for those applications as well as what to look for in general when making HPC workloads future proof. We will be using data collected in collaboration with Alces Flight, which delivers scalable HPC resources that can augment a traditional on-premises HPC cluster, whilst maintaining the agility and cost-effectiveness of the AWS public cloud platform. Company: Ellexus Speaker: Dr Rosemary Francis, COO |
09:00am - 09:30am |
Simulation on Demand – eCADFEM meets AWS Simulation on Demand is a new initiative at CADFEM to expand the existing License- and Engineering-Offering (eCADFEM). Virtual product development is one of the key strategies and CADFEM is supporting customers with flexibility and deep knowledge transfer. The new implemented Hardware on Demand solution based on an integrated AWS-solution enables this by adding compute capacity dedicated to the customers individual needs. These solutions will be presented in detail, together with some CAE-application references. Company: Cadfem Speaker: Gerhard Zelder, Team Leader IT-Solutions |
09:30am - 10:00am |
It’s all about the money, money, money… Too often the arguments for and against (HPC in the) cloud are based on mis-understandings, fallacies and half-truths, built off an incomplete understanding of the total cost of ownership of the competing options. This talk aims to disrupt the fallacies and half-truths and shine some light into the dark corners of how much you are actually spending. This presentation will be especially helpful for those contemplating a move towards cloud for some or all of their HPC needs. Company: Red Oak Consulting Speaker: Owen Thomas, Partner |
10:30am - 11:00am
|
Convolutional Neural Networks (CNN) for image recognition. CNNs are extremely powerful tools to extract features from non-temporal input vectors such as images. This workshop focuses on describing architecture of one of the most famous networks for image recognition based on ImageNet challenge and describes its implementation in Apache MXNet. Company: AWS Speaker: Cyrus Vahid, Principal Solution Architect |
11:00am - 11:30am
|
Cloud computing: the scientific method embodied in computational form The scientific method is fundamentally about exploring novel ideas to find where they fail, so we can adjust our theories & models, test them again, and in doing so, improve our understanding of the world. In a cloud computing world where servers become software that are created with a script, computation itself becomes something that be be iteratively improved. A new exploration of an idea can be launched in a few minutes, scaled up or down based on needs (not limitations) and entirely transplanted to different physical environments if it turns out that the computation has different evolving needs to those we anticipated at the beginning. Success in science almost never involves guessing the right answer the first time, so why treat computation any differently? Company: AWS Speaker: Brendan Bouffler, Head of AWS Research Cloud |
11:30am - 12:00pm |
HPC made simple with NICE EnginFrame 2017 Discover how EnginFrame leverages key AWS technologies like CfnCluster and Amazon Elastic File System. EnginFrame helps HPC users to be more productive and keeps them focused on their applications, jobs and data, hiding the complexity of clusters, schedulers, storage and scripting, both on premises and on AWS. Company: Nice Software Speaker: Francesco Ruffino, Global Specialized HPC Solution Architect |
12:00pm - 12:30pm |
Scalable, shared site HPC facilities on AWS with Flight Enterprise The use of on-demand HPC in public cloud has seen much success from individual researchers, scientists, and engineers, but how do you expand this to a site-wide HPC facility for larger groups? The team at Alces have designed Flight Enterprise to deliver a persistent, scalable HPC resource that can augment a traditional on-premises HPC cluster, whilst maintaining the agility and cost-effectiveness of the AWS public cloud platform. Our demonstration will show you first hand how sites can use Flight Enterprise to deliver large-scale, high-performance HPC facilities for their end-users that leverage the security and accessibility of AWS. Company: Alces Flight Speaker: Wil Mayers, Director |
12:30 pm - 01:00pm |
Introduction to F1 FPGA Introduction and deep dive overview on the AWS F1 FPGA Instance Type. Company: AWS Speaker: Arthur Petitpierre, Specialist Solution Architect |
01:00pm - 01:30pm |
Common Myths Debunked for Engineering Simulation in the Cloud Although businesses are increasingly using cloud computing for engineering simulations, many misconceptions still surround cloud-based simulation. This presentation will debunk the leading misconceptions about cloud computing, and, in doing so, will assist engineering and IT staff as they make decisions regarding computing resources. While dispelling these misconceptions, resources and insight will be provided to help organizations steer around possible failure points as they consider cloud computing. Company: Ansys Speaker: Dr Wim Slagter, HPC & Cloud Alliances |
01:30pm - 02:00pm |
Computational Fluid Dynamics is virtually synominous with extreme high-performance computing. Still, this tightly-coupled, HPC application runs very well on AWS. From launching and running on 1000’s of cores, to monitoring convergence residuals with cloudwatch, and methods for taking advantage of spot pricing; this talk offers suggestions on how to get the best performance out of AWS. Many different CFD software pacakges are deployed to AWS routinely and examples are included. Company: AWS Speaker: Linda Hedges, Principal Solution Architect |
02:00pm - 02:30pm |
An Intuitive introduction to Artificial Neural Networks (ANN) ANNs have been around for decades with periodic surge in popularity. The latest started in 2012 and in 2015 the adoption of Deep Learning networks gained unprecedented momentum. This workshop intends to deliver a fundamental understanding of feed-forward networks, and intuitive understanding of how learning works, and use cases in AI that can be tackled using ANN along with sample implementations using Apache MXNet. Company: AWS Speaker: Cyrus Vahid, Principal Solution Architect |
02:30pm - 03:00pm |
Advanced Oil & Gas optimisation on AWS Zenotech recently completed a project to demonstrate a new way to conduct testing on an oil reservoir development using advanced optimisation technology, cutting-edge reservoir modelling software and computing resource from AWS. The demonstrator shows that significant economic benefits can be achieved using these techniques for relatively minor resource costs. This talk will describe the project, the tools and services used, and the key findings. Company: Zenotech Speaker: Mike Turner, Business Development |
03:00pm - 03:30pm |
Generating Insights from IoT Data AWS IoT Services help you collect and send data to the cloud, make it easy to load and analyze that information, and provide the ability to manage your devices, so you can focus on developing applications that fit your needs. In this session, you will learn how to use AWS IoT Service, its components and typical use cases. We will also look at various patterns of analyzing IoT data, including HPC use cases such as Deep Learning, to build smart IoT applications on AWS and how it allows scientists and engineers to solve complex, compute-intensive and data-intensive problems. HPC applications often require high network performance, fast storage, large amounts of memory, very high compute capabilities, or all of these. Company: AWS Speaker: Neeraj Kumar, Solution Architect |
03:30pm - 04:00pm |
Agile HPC on-demand clusters for Ansys Fluent/AWS EnginFrame/Alces Flight For CFD users getting on-demand access to high-performance compute cluster resources has long been considered a fantasy. Due to the high cost and often lengthy process needed to start their work users can find themselves frustrated before they even begin a project. Things are finally changing. Alces Flight, in partnership with AWS, EngineFrame, and the ANSYS Fluent teams have created a quick-launch cluster set designed specifically for CFD workloads certified for use with commercial applications. Leveraging the agility of public cloud, users can get the resources they need in minutes to manage their workflow quickly and easily. Our demonstration will show the speed of quick-launch and how a user can be up and running their workloads in a matter of minutes, bypassing the long acquisition process of old. Company: Alces Flight Speaker: Wil Mayers, Director |
04:00pm - 04:30pm |
Deep Learning at Cloud Scale with Intel MKL Optimised MXNET Intel will present an overview of deep learning and use-cases with worked examples and demonstration on fine-tuning for a deep network. Deliered via AWS EC2 and Intel tools. Company: Intel Speaker: Andres Rodriguez, Technical Specialist |
04:30pm - 05:00pm |
PBS Works – Enabling HPC in the Cloud We will discuss PBS Cloud as a tool to manage, control and provision Cloud HPC resources in AWS for an end user enterprise. Further we will cover PBS Access as a toolkit to provide easy access for Engineers to utilise compute resource in a transparent way independent to the resource being utilised. PBS Professional is the invisible glue that manages and makes the experience seamless for users and administrators Company: Altair Speaker: Ian Littlewood, Director of Enterprise Solutions |
05:00pm - 05:30pm |
3 Hidden Benefits of Cloud HPC Companies are moving to high-performance computing (HPC) in the cloud for access to flexible, varied, pay-per-use HPC resources. At the same time, they don’t have to manage these assets or spend a massive amount of capital and time to deploy them. Moving simulations to cloud HPC resources enables companies to take advantage of a wide variety of standard and specialty hardware at previously unimaginable scales. Utilizing the cloud can unlock the potential of engineering staff and drive innovation to the next level. As cloud HPC matures, these obvious benefits are driving movement to the cloud for HPC workloads. But often overlooked are other benefits that are less obvious harder to quantify. However, they represent the truly transformative power of the cloud. In this presentation, we will look at three often overlooked, but transformational benefits of moving to the cloud Company: Rescale Speaker: Wolfgang Dreyer, EMEA Director |
If you would like to schedule an in person meeting with an AWS team member, on any of the subjects being covered in our mini sessions, please contact us directly!
For more information we have provided some useful links below:
- High Performance Computing (HPC)
- AWS Batch
- CfnCluster
- Introducing Elastic Network Adapter (ENA), the next generation network interface for EC2 instances
- Coming Soon: Amazon EC2 C5 Instances, the next generation of Compute Optimized instances
