DOCSLIB.ORG

Caffe Deep Dive Session

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Caffe Deep Dive Session May 22-25, 2017 | San Francisco Machine Learning/Deep Learning Track ▶ Caffe Deep Dive Session ▶ Increase your expertise on the Caffe framework through hands-on guided exercises. You will experience fast, scalable, easy machine learning with openPOWER, GPUs and Docker on the Nimbix public cloud. You’ll look at image classification with a trained model. Train a model of a small dataset with DIGITS and image classify with your own trained model. Compare different techniques and explore the ”Model Zoo”. ▶ Duration – 1.5 hours ▶ Day/Time – Thu 25 9:00AM – 10:30AM ▶ Leader/Trainers – Jason Furmanek, Franck Barillaud, Clarisse Taaffe-Hedglin May 22-25, 2017 San Francisco Agenda Deep Learning and Caffe Why Now Why Caffe Hands–on Lab: Caffe and Digits Inference using a trained model Training a new model May 22-25, 2017 San Francisco Machines Are Learning the Way We Learn…. ▪From "Texture of the Nervous System of Man and the Vertebrates" ▪Artificial Neural Networks by Santiago Ramón y Cajal. May 22-25, 2017 San Francisco 4 Deep Learning Across All Industries Automotive and Security and Public Consumer Web, Medicine and Biology Broadcast, Media and Transportation Safety Mobile, Retail Entertainment • Autonomous driving: • Video Surveillance • Image tagging • Drug discovery • Captioning • Pedestrian detection • Image analysis • Speech recognition • Diagnostic assistance • Search • Accident avoidance • Facial recognition and • Natural language • Cancer cell detection • Recommendations detection • Sentiment analysis • Real time translation Auto, trucking, heavy Local and national Hyperscale web Pharmaceutical, Medical Consumer facing equipment, Tier 1 police, public and companies, large equipment, Diagnostic companies with large suppliers (Hyundai, private safety/ security retail (Google photos, labs (Takeda, Asian streaming of existing Toyota, Komatsu, General (ADT, IViz, Pinkerton, Twitter, Woolworths, Pharma, Pfizer) media, or real time Motors, Volvo) Sentry) Aeon) content May 22-25, 2017 San Francisco Why Now? → Data → Compute → Technique May 22-25, 2017 San Francisco 6 Data Sources MNIST 0-9 Over 1000 datasets at: https://www.kaggle.com/datasets CIFAR-10 (RBG) ImageNet 10,000,000 labeled images 300,000 Labeled images depicting 10,000+ object categories https://quickdraw.withgoogle.com/data Learned filter for AlexNet, Krizhevsky et al. 2012 May 22-25, 2017 San Francisco Data & Compute at the Core of Training Inference Training Inference •Data intensive: historical •Enables the computer to data sets act in real time •Compute intensive: •Low Power 100% accelerated •Out at the edge •Develop a model for use on the edge as inference May 22-25, 2017 San Francisco Requires a Lot of Computational Resources Easy scale-out with: But training DL Models takes time and effort Training can take hours, days or weeks Input data and model sizes are becoming larger than ever (e.g. video input, billions of features etc.) Moore’s law is dying Real-time analytics with: Resulting in…. Unprecedented demand for offloaded computation, accelerators, and higher memory bandwidth systems May 22-25, 2017 San Francisco OpenPOWER: Open Hardware for High Performance Systems designed for big data analytics and superior cloud economics Traditional Intel x86 https://power.jarvice.com OpenPOWER https://mc.jarvice.com Upto: 12 cores per cpu 96 hardware threads per cpu ▪1 TB RAM 7.6Tb/s combined I/O Bandwidth GPUs and FPGAs coming… http://www.softlayer.com/POWER-SERVERS May 22-25, 2017 San Francisco 1 0 New Paradigm, New Chip, New Servers for AI New Chip New Power Accelerated AI “POWER8 with NVLink” Linux Servers S821LC: High Density 2-Socket 1U S822LC for Big Data Accelerator X POWER8 + coherent CAPI + novel NVlink for high BW coherent CPU/GPU acceleration S822LC for High Performance Computing May 22-25, 2017 M.Gschwind, Bringing the Deep Learning Revolution into the Enterprise San Francisco NVLink and P100 advantage NVLink reduces communication time and overhead Incorporating the fastest GPU for deep learning Data gets from GPU-GPU, Memory-GPU faster, for shorter training times IBM advantage: data communication and GPU performance POWER8 + 78 ms Tesla P100+NVLink x86 based 170 ms GPU system ImageNet / Alexnet: Minibatch size = 128 May 22-25, 2017 San Francisco Accelerated Compute with OpenPOWER and NVIDIA GPU Huge speed-ups with GPUs and OpenPOWER! Mesos supports GPUs IBM Spectrum Conductor includes enhanced support for fine grained GPU and CPU scheduling with Apache Spark and Docker ▪Credit: Kevin Klaues, Mesosphere May 22-25, 2017 San Francisco 1 3 Introducing PowerAI to Get Started Fast Package of Pre- Easy to install & get Optimized for Compiled Major started with Deep Performance Deep Learning Learning with Enterprise- To Take Advantage of Frameworks Class Support NVLink https://www.ibm.com/us-en/marketplace/deep-learning-platform Enabled by High Performance Computing Infrastructure May 22-25, 2017 San Francisco14 NVIDIA Docker Available on OpenPOWER • A docker wrapper and tools to package and GPU based apps • Uses drivers on the host • Manual GPU assignment • Good for single node Credit: https://github.com/NVIDIA/nvidia-docker May 22-25, 2017 San Francisco 1 5 DevOps for accelerated Deep Learning on POWER Build Servers VM VM Image Registry Developer Jane DockerHub Inference Micro Services in Docker containers POWER Bare metals with GPUs for model training Shared Data Store May 22-25, 2017 San Francisco 1 6 Technique Increasing in Complexity Convolutional Neural Networks (CNNs) are evolving May 22-25, 2017 San Francisco Caffe Framework Addresses Technique frontend: a language for any network, any task tools: And plenty With the visualization, profiling, network right tools debugging, etc. internal of examples representation backend: layer library: dispatch compute for fast implementations of learning and inference common functions and gradients framework From the Concrete to the Abstract May 22-25, 2017 San Francisco 1 Caffe Overview DL Framework developed by the Berkeley AI Research (BAIR) Formerly Berkeley Vision and Learning Center (BVLC) Yangqing Jia created the project during his PhD at UC Berkeley Caffe is released under the BSD 2-Clause license https://github.com/BVLC/caffe Strong developer/user community: 11,000+forks, 248 Contributors Written in C++, CUDA; Command line, Python and Matlab interfaces May 22-25, 2017 San Francisco Caffe Constructs Model Development name: "LogReg" Blob: how you represent your data (N-D Arrays) layer { name: "mnist" Layer: takes bottom blobs (input) to top blobs (output) and back with weights type: "Data" top: "data" and biases top: "label" Setup: initialize the layer and its connections data_param { source: "input_leveldb" Forward: given input from bottom compute the output and batch_size: 64 } send to the top } Backward: given the gradient w.r.t. the top output compute layer { name: "ip" the gradient w.r.t. to the input and send to the bottom. type: "InnerProduct" bottom: "data" Net: is a collection of layers organized as a directed graph top: "ip" Including a loss layer that defines the learning inner_product_param { num_output: 2 } Implementation/Optimization } layer { Solver: how you train your model with methods that address name: "loss" type: "SoftmaxWithLoss" the general optimization problem of loss minimization bottom: "ip" bottom: "label" top: "loss" Uses Google’s protocol buffer: you build a model schema: the code is built for you } May 22-25, 2017 San Francisco Caffe Use Cases and Model Zoo Primarily focused on vision but branching out: Sequences, reinforcement learning, speech + text Used by many companies including Facebook, Pinterest, Adobe at scale production implementations Provides Reference Models via Model Zoo May 22-25, 2017 San Francisco Caffe Model Zoo Not Just a static repository but a strong community when you can find model definitions, optimization parameters, solver configurations, pre-trained weights Table of Contents includes over 40 areas of interest https://github.com/BVLC/caffe/wiki/Model-Zoo Great starting point for a fast start or fine tuning May 22-25, 2017 San Francisco Fine Tuning Made Easy A trained model that classifies objects into a defined number of categories may Fine Tuning Tips not be sufficient to classify objects into a Start with the last layer first •Caffe layers have local learning rates: param { lr_mult: 1 } new set of categories based on the •Freeze all but the last layer for fast optimization and avoiding early divergence by setting lr_mult: 0 learned features. to fix a parameter. •Stop if good enough, or keep fine-tuning Fine tuning is a way to copy the layers of Reduce the learning rate •Drop the solver learning rate by 10x, 100x features you want to keep and modify the •Preserve the initialization from pre-training and avoid divergence model layers to learn new weights for new categories. May 22-25, 2017 San Francisco Why Caffe Active Community Fast and efficient Plenty of Reference Models to build upon May 22-25, 2017 San Francisco Tutorial Steps → Connect to the NIMBIX public cloud → Image Classification with Caffe and a trained model → Train a model on a small dataset with DIGITS → Image Classify with your own trained model in DIGITS https://www.slideshare.net/IndrajitPoddar/fast-scalable-easy-machine-learning-with-openpower-gpus-and-docker May 22-25, 2017 San Francisco In Summary →We reviewed the structure and advantages of the Caffe Deep Learning framework →We've used the NIMBIX Power container cloud running Docker containers with GPUs. →We built new Docker containers,
Load more

Recommended publications
  • Automated Analysis of Speculation Windows in Spectre Attacks
    Barbara Gigerl Automated Analysis of Speculation Windows in Spectre Attacks MASTER'S THESIS to achieve the university degree of Diplom-Ingenieurin Master's degree programme: Software Engineering and Management submitted to Graz University of Technology Supervisor: Daniel Gruss Institute for Applied Information Processing and Communication Graz, May 2019 i AFFIDAVIT I declare that I have authored this thesis independently, that I have not used other than the declared sources/resources, and that I have explicitly indicated all material which has been quoted either literally or by content from the sources used. The text document uploaded to TUGRAZonline is identical to the present master's thesis. Date Signature EIDESSTATTLICHE ERKLARUNG¨ Ich erkl¨arean Eides statt, dass ich die vorliegende Arbeit selbstst¨andig verfasst, andere als die angegebenen Quellen/Hilfsmittel nicht benutzt, und die den benutzten Quellen w¨ortlich und inhaltlich entnommenen Stellen als solche kenntlich gemacht habe. Das in TUGRAZonline hochgeladene Textdokument ist mit der vorliegenden Masterarbeit identisch. Datum Unterschrift Abstract Speculative execution is a feature integrated into most modern CPUs. Although intro- duced as a way to enhance the performance of processors, the release of Spectre attacks showed that it is a significant security risk. Since CPUs from various vendors, includ- ing Intel, AMD, ARM, and IBM, implement speculative execution, all different kinds of devices are affected by Spectre attacks, for example, desktop PCs and smartphones. Spectre attacks exploit the branch prediction mechanisms of the CPU and then use a cache covert channel to leak secret data. Several attack variants have been discovered since the release, including Spectre-PHT which targets the Pattern History Table of the CPU.
    [Show full text]
  • Defeating Invisible Enemies:Firmware Based
    Defeating Invisible Enemies: Firmware Based Security in OpenPOWER Systems — Linux Security Summit 2017 — George Wilson IBM Linux Technology Center Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation Agenda Introduction The Case for Firmware Security What OpenPOWER Is Trusted Computing in OpenPOWER Secure Boot in OpenPOWER Current Status of Work Benefits of Open Source Software Conclusion Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 2 Introduction Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 3 Disclaimer These slides represent my views, not necessarily IBM’s All design points disclosed herein are subject to finalization and upstream acceptance The features described may not ultimately exist or take the described form in a product Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 4 Background The PowerPC CPU has been around since 1990 Introduced in the RS/6000 line Usage presently spans embedded to server IBM PowerPC servers traditionally shipped with the PowerVM hypervisor and ran AIX and, later, Linux in LPARs In 2013, IBM decided to open up the server architecture: OpenPOWER OpenPOWER runs open source firmware and the KVM hypervisor with Linux guests Firmware and software designed and developed by the IBM Linux Technology Center “OpenPOWER needs secure and trusted boot!” Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 5 The Case for Firmware Security Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 6 Leaks Wikileaks Vault 7 Year 0 Dump NSA ANT Catalog Linux Security Summit / Defeating Invisible Enemies / September 14, 2017 / © 2017 IBM Corporation 7 Industry Surveys UEFI Firmware Rootkits: Myths and Reality – Matrosov Firmware Is the New Black – Analyzing Past Three Years of BIOS/UEFI Security Vulnerabilities – Branco et al.
    [Show full text]
  • Foundation Overview February 2014
    OpenPOWER Overview May 2015 Keith Brown Director, IBM Systems Technical Strategy & Product Security [email protected] http://openpowerfoundation.org/ © 2015 OpenPOWER Foundation What is the OpenPOWER Ecosystem? Cloud Software Existing ISV community of 800+ Standard Operating Open Environment Source All major Linux distros (System Mgmt) Software Communities Operating Open sourced Power8 System / KVM firmware stack New OSS Firmware OpenPOWER Resources for porting and Firmware Community optimizing on Hardware OpenPOWER OpenPOWERFoundation.org Technology 2 © 2015 OpenPOWER Foundation A Fast Start for OpenPOWER! The year • Collaborative solutions, standards, and reference designs available • Independent members solutions and systems ahead • Sector growth in technical computing and cloud • Global growth with increasing depth in all layers • Broad adoption across hardware, software, and end users 3 © 2015 OpenPOWER Foundation Fueling an Open Development Community 4 © 2015 OpenPOWER Foundation Critical workloads run on Linux on Power Web, Java Apps and Infrastructure Analytics & Research HPC applications for Life Sciences • Highly threaded • Compute intensive • Throughput oriented • High memory bandwidth • Scale out capable • Floating point • High quality of service • High I/O rates Business Applications Database • High quality of service • Handle peak workloads • Scalability • Scalability • Flexible infrastructure • High quality of service • Large memory footprint • Resiliency and security 5 © 2015 OpenPOWER Foundation IBM, Mellanox, and NVIDIA
    [Show full text]
  • Linux and Open Source: the View from IBM
    Linux @ IBM Linux and Open Source: The View From IBM Jim Elliott, Advocate, Strategic Growth Businesses IBM Canada Ltd. ibm.com/vm/devpages/jelliott SHARE Session 9200 February 28, 2005 © 2005 IBM Corporation Linux @ IBM Linux and Open Source: The View from IBM Session 9200 Linux and Open Source are game-changing technologies. Jim will provide a review of Linux and Open Source from IBM's point of view covering: – Overview, Value and Marketplace: A brief update on Linux and Open Source and the value to customers – Usage: How Linux and Open Source are being used by customers today and our view of the future – IBM and Open Source: How IBM is using Open Source software internally and IBM involvement in the Open Source community 2 SHARE Session 9200 February 28, 2005 Linux @ IBM Linux Overview, Value, and Marketplace “Linux will do for applications what the Internet did for networks.” Irving Wladawsky-Berger, IBM LinuxWorld, January 2000 SHARE Session 9200 February 28, 2005 © 2005 IBM Corporation Linux @ IBM Advancing Technology What if … ... everything is connected and intelligent? ... networking and transactions are inexpensive? ... computing power is unlimited? Adoption of Processor Storage Bandwidth Number of Interaction open standards speed networked costs devices 4 SHARE Session 9200 February 28, 2005 Linux @ IBM The road to On Demand is via Open Computing Open Source Open Architecture Open Standards 5 SHARE Session 9200 February 28, 2005 Linux @ IBM Open Source Software www.opensource.org What is Open Source? – Community develops, debugs, maintains – “Survival of the fittest” – peer review – Generally high quality, high performance software – Superior security – on par with other UNIXes Why does IBM consider Open Source important? – Can be a major source of innovation – Community approach – Good approach to developing emerging standards – Enterprise customers are asking for it 6 SHARE Session 9200 February 28, 2005 Linux @ IBM Freedom of Choice “Free software is a matter of liberty, not price.
    [Show full text]
  • HARDWARE INNOVATION REVEAL Openpower Foundation Summit
    HARDWARE INNOVATION REVEAL OpenPOWER Foundation Summit 2018 87 Solutions BPS-8201 Server The BPS-8201 is a High Density/Storage with IBM POWER8 Turismo SCM processor Platform , 2U/ (16) 3.5" SAS/SATA HDDs. System ADM-PCIE-9V3 - FPGA Accelerator Board—Latest FPGA accelerator board, is CAPI 2.0 and OpenCAPI enabled featuring a powerful Xilinx® Virtex® UltraScale+ ™ FPGA. The ADM-PCIE-9V3 is ideal for a variety of acceleration applications (Compute, Networking, Storage) packed Card into a small HHHL server friendly PCIe add-in card size. OpenCAPI FPGA Loop Back cable and OpenCAPI cable. These enable testing and OpenCAPI accelerators to be connected to standard PCIe while signaling to the host processor through sockets attached to the main system board. Cable Parallware Trainer aims to democratize access to HPC by providing sim- ple-to-use assistance in developing software for shared memory and accelerator technologies. Developed specifically to help software pro- grammers learn how to parallelise software, quickly and efficiently aids Software users in developing OpenMP and OpenACC enabled software. Escala E3-OP90 is an Open Power based server optimized for Deep Learning. The L3-OP90 features 2 Power9 sockets which are intercon- nected with up to 4 Nvidia Volta GPUs through NVlink 2.0. The architec- ture is designed for the implementation of large deep learning models System enabling highest levels of resolution / accuracy. HBA 9405W-16i Adapter—The x16 low-profile HBA9405W-16i is ide- al for high-performing, bandwidth-intense applications. The HBA enables internal communication to PCIe JBOFs and delivers the per- formance and scalability needed by critical applications such as video Card streaming, medical imaging and big data analytics.
    [Show full text]
  • Openpower Innovations for HPC IBM Research
    IWOPH workshop, ISC, Germany June 21, 2017 OpenPOWER Innovations for HPC IBM Research Christoph Hagleitner, [email protected] IBM Research - Zurich Lab IBM Research - Zurich . Established in 1956 . 45+ different nationalities . Open Collaboration – 43 funded projects and 500+ partners in Horizon 2020 – Binnig and Rohrer Nanotechnology Centre opened in 2011 (Public Private Partnership with ETH Zürich and EMPA) – 7 European Research Council Grants . Two Nobel Prizes: – 1986 for the scanning tunneling microscope (Binnig and Rohrer) – 1987 for the discovery of high- temperature superconductivity (Müller and Bednorz) 6/22/2017 IBM Research - Zurich Lab 2 Agenda . Cognitive Computing & HPC . HPC Software ecosystem . HPC system roadmap . HPC Processor . HPC Accelerators 6/22/2017 IBM Research - Zurich Lab 3 Cognitive Computing Applications Deep Computational Learning Complexity Graph Analytics O(N3) Classic HPC Applications Dim. Reduction HPC 2 Uncertainty Knowledge Graph O(N ) Quantification Creation Information Retrieval HADOOP O(N) Database queries Data MB GB TB PB Volume 6/22/2017 IBM Research - Zurich Lab 4 Cognitive Computing: Integration & Disaggregation . hadoop-style workloads . complex HPC-like workloads ... scale-out via network ... scale-up via high-speed buses . main metrics . main metrics – cost (capital, energy) – memory / accelerator / inter-node BW – compute density – optimal mix of heterogeneous resources – scalability (CPU / GPU / FPGA / HBM / DRAM / NVMe) homogeneous nodes – compute density, scalability (CPU / FPGA / NVMe plus compute) heterogeneous nodes datacenter disaggregation data centric design 6/22/2017 IBM Research - Zurich Lab 5 Dense, Energy Efficient Computing: Hyperscale FPGA . Cloud economics – density (>1000 nodes / rack) – integrated NICs – switch card (backplane, no cables) – medium to low-cost compute chips .
    [Show full text]
  • Summit Press
    Summit Press During the second annual OpenPOWER Summit, held April 5-8 in San Jose, the OpenPOWER Foundation made a set of announcements and showcased more than 50 new member innovations that advance its commitment to open technologies in the data center. The solutions and innovations showcased are built around OpenPOWER technology and the Open Compute reference architecture. OpenPOWER members Rackspace and the University of Michigan also announced the deployment of OpenPOWER systems. Media coverage from OpenPOWER Summit resulted in over 134 stories and reached an audience of over 285 million individuals based on the latest data provided by Cision. Highlights and full coverage are below. Google and Rackspace OpenCompute Collaboration o "Google also said it’s developing a data center server with cloud-computing company Rackspace Hosting Inc. that runs on a new IBM OpenPower chip called Power9, rather than Intel processors that go into most servers. The final design will be given away through Facebook Inc.’s Open Compute Project, so other companies can build their data center servers this way, too." – Bloomberg o "The search giant [Google] said on Wednesday that, along with cloud computing company Rackspace, it’s co-developing new server designs that are based on IBM chip technology." – Fortune Rackspace Barreleye o “Rackspace has already designed one Power-based server, called Barreleye, that it plans to put in production in the coming months, offering cloud services to its cloud customers.”-- CIO IBM OpenCompute o “Big Blue will also
    [Show full text]
  • Openpower Ready™ November 13, 2017 Revision 2.0 - PRD
    REVIEW DRAFT www.openpowerfoundation.org OpenPOWER Ready™ November 13, 2017 Revision 2.0 - PRD OpenPOWER Ready™: Definition and Criteria OpenPOWER Ready Work Group <[email protected]> OpenPower Foundation Revision 2.0 - PRD (November 13, 2017) Copyright © 2017 OpenPOWER Foundation All capitalized terms in the following text have the meanings assigned to them in the OpenPOWER Intellectual Property Rights Policy (the "OpenPOWER IPR Policy"). The full Policy may be found at the OpenPOWER website or are available upon request. REVIEW DRAFT - This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published, and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this section are included on all such copies and derivative works. However, this document itself may not be modified in any way, including by removing the copyright notice or references to OpenPOWER, except as needed for the purpose of developing any document or deliverable produced by an OpenPOW- ER Work Group (in which case the rules applicable to copyrights, as set forth in the OpenPOWER IPR Policy, must be followed) or as required to translate it into languages other than English. The limited permissions granted above are perpetual and will not be revoked by OpenPOWER or its successors or assigns. This document and the information contained herein is provided on an
    [Show full text]
  • The Openpower Initiative
    Foundation Overview April 2014 Gilad Shainer HPC Advisory Council, Lugano Switzerland, April 2014 © OpenPOWER Foundation 2014 OpenPOWER Overview • OpenPOWER Foundation was founded in 2013 • Open technical membership organization • To enable open data centers • To enable open architecture • To enable higher performance and efficiency Board of Directors Gordon MacKean, Chair Brad McCredie, President Michael Diamond, Vice President and Google IBM Treasurer, NVIDIA David Gamba, Director Zheng Jiang, Director Albert Mu, Director Altera Suzhou PowerCore Technology Tyan Gordon Patrick, Director Gilad Shainer, Director Mike Williams, Director 3 Micron Mellanox Samsung © OpenPOWER Foundation 2014 Driving industry innovation The goal of the OpenPOWER Foundation is to create an open ecosystem, using the POWER Architecture to share expertise, investment, and server-class intellectual property to serve the evolving needs of customers. – Opening the architecture to give the industry the ability to innovate across the full Hardware and Software stack • Simplify system design with alternative architecture • Includes SOC design, Bus Specifications, Reference Designs, FW OS and Open Source Hypervisor • Little Endian Linux to ease the migration of software to POWER – Driving an expansion of enterprise class Hardware and Software stack for the data center – Building a complete ecosystem to provide customers with the flexibility to build servers best suited to the Power architecture 4 © OpenPOWER Foundation 2014 OpenPOWER moves the industry forward • The
    [Show full text]
  • Linux and Open Source: the View from IBM
    Linux @ IBM Linux and Open Source: The View From IBM Jim Elliott Advocate – Infrastructure Solutions Manager – System z9 and zSeries Operating Systems IBM Canada Ltd. ibm.com/vm/devpages/jelliott SHARE 105 Session 9200 August 22, 2005 © 2005 IBM Corporation Linux @ IBM Linux and Open Source: The View from IBM Session 9200 Linux and Open Source are game-changing technologies. Jim will provide a review of Linux and Open Source from IBM's point of view covering: – Overview, Value and Marketplace: A brief update on Linux and Open Source and the value to customers – Usage: How Linux and Open Source are being used by customers today and our view of the future – IBM and Open Source: How IBM is using Open Source software internally and IBM involvement in the Open Source community 2 SHARE 105 Session 9200 August 22, 2005 Linux @ IBM Open Computing SHARE 105 Session 9200 August 22, 2005 © 2005 IBM Corporation Linux @ IBM The Principles of Open Computing Permit interoperability by using published specifications for APIs, protocols, and data and file formats The specifications must be published without restrictions that limit implementations, or require royalties or payments* Open Computing Open Standards Open Architecture Open Source * other than reasonable royalties for essential patents 4 SHARE 105 Session 9200 August 22, 2005 Linux @ IBM Adaptability is vital “It is not the strongest of the species that survives, nor the most intelligent; it is the one that is most adaptable to change.” – Charles Robert Darwin (1809-82) 5 SHARE 105 Session 9200 August 22, 2005 Linux @ IBM Open Computing Policy Roadmap 1.
    [Show full text]
  • IBM Power Systems HPC Solutions Brief
    IBM Power Systems August 2017 Solution Brief IBM Power Systems HPC solutions Faster time to insight with the high-performance, comprehensive portfolio designed for your workflows Architecting superior high performance computing (HPC) clusters Highlights requires a holistic approach that responds to performance at every level of the deployment. • IBM Power Systems HPC solutions are comprehensive clusters built for faster time ® to insight IBM high performance computing solutions, built with IBM Power Systems™, IBM® Spectrum™ Computing, IBM Spectrum Storage™, and • IBM POWER8 offers the performance, IBM Software technologies, provide an integrated platform to optimize cache and memory bandwidth to drive the best results from high performance your HPC workflows, resulting in faster time to insights and value. computing and high performance data analytics applications The industry’s most comprehensive, • IBM HPC software is designed to data-centric HPC solutions capitalize on the technical features Only IBM provides a total HPC solution, including optimized, of IBM Power Systems clusters best-of-breed components at all levels of the system stack. Comprehensive solutions ensure: • Rapid deployment • Clusters that deliver value immediately after acceptance IBM HPC solutions are built for data-centric computing, and delivered with integration expertise targeting performance optimization at the workflow level. Data-centric design minimizes data motion, enables compute capabilities across the system stack, and provides a modular, scalable architecture that is optimized for HPC. Data-centric HPC and CORAL Data-centric design was a primary reason the Department of Energy selected IBM for the CORAL deployment. Summit (Oak Ridge National Labora- tory) and Sierra (Lawrence Livermore National Laboratory) will become some of the largest, most groundbreaking, and most utilized installations in the world.
    [Show full text]
  • Presence of Europe in HPC-HPDA Standardisation and Recommendations to Promote European Technologies
    Ref. Ares(2020)1227945 - 27/02/2020 H2020-FETHPC-3-2017 - Exascale HPC ecosystem development EXDCI-2 European eXtreme Data and Computing Initiative - 2 Grant Agreement Number: 800957 D5.3 Presence of Europe in HPC-HPDA standardisation and recommendations to promote European technologies Final Version: 1.0 Authors: JF Lavignon, TS-JFL; Marcin Ostasz, ETP4HPC; Thierry Bidot, Neovia Innovation Date: 21/02/2020 D5.3 Presence of Europe in HPC-HPDA standardisation Project and Deliverable Information Sheet EXDCI Project Project Ref. №: FETHPC-800957 Project Title: European eXtreme Data and Computing Initiative - 2 Project Web Site: http://www.exdci.eu Deliverable ID: D5.3 Deliverable Nature: Report Dissemination Level: Contractual Date of Delivery: PU * 29 / 02 / 2020 Actual Date of Delivery: 21/02/2020 EC Project Officer: Evangelia Markidou * - The dissemination level are indicated as follows: PU – Public, CO – Confidential, only for members of the consortium (including the Commission Services) CL – Classified, as referred to in Commission Decision 2991/844/EC. Document Control Sheet Title: Presence of Europe in HPC-HPDA standardisation and Document recommendations to promote European technologies ID: D5.3 Version: <1.0> Status: Final Available at: http://www.exdci.eu Software Tool: Microsoft Word 2013 File(s): EXDCI-2-D5.3-V1.0.docx Written by: JF Lavignon, TS-JFL Authorship Marcin Ostasz, ETP4HPC Thierry Bidot, Neovia Innovation Contributors: Reviewed by: Elise Quentel, GENCI John Clifford, PRACE Approved by: MB/TB EXDCI-2 - FETHPC-800957 i 21.02.2020 D5.3 Presence of Europe in HPC-HPDA standardisation Document Status Sheet Version Date Status Comments 0.1 03/02/2020 Draft To collect internal feedback 0.2 07/02/2020 Draft For internal reviewers 1.0 21/02/2020 Final version Integration of reviewers’ comments Document Keywords Keywords: PRACE, , Research Infrastructure, HPC, HPDA, Standards Copyright notices 2020 EXDCI-2 Consortium Partners.
    [Show full text]