AI lab using IBM Power Systems
- 1. © 2017 IBM Corporation AI Lab using IBM Power 9 Systems Fastest AI Supercomputer built using Power 9 systems 95 AI Use cases around 24 Industries Workshops/Bootcamp Big Data, AI , HPC , Cloud and Block Chain Curriculums/Courses
- 2. © 2017 IBM Corporation In the future, all communication between machines and humans will be powered by enterprise systems and operational AI
- 3. © 2017 IBM Corporation
- 4. © 2017 IBM Corporation IBM is leading the way IBM is teaming with universities, startups, ISV’s and industries to help develop further the impact of artificial intelligence for solutions for real-world opportunities
- 5. © 2017 IBM Corporation5 Background and Motivation The IBM AI Lab will play a major role in the research and development commercial and industrial development of emerging AI technologies There is a strong need for research and development activity in these domains: – Encouraging academic-industry partnerships – Cross-disciplinary and collaborative research – Making AI accessible to non-technical business students – Enabling faculty-technologist interaction and learning – Enabling startups , ISVs and industries to use the platform to innovate in ways that improve the World condition .
- 6. © 2017 IBM Corporation Technologies and Partners The AI Lab will include IBM and other corporate sponsors, coupled with open source technologies to accelerate results 6
- 7. © 2017 IBM Corporation7 CoE Charter and Objectives 1. Conduct research on rapidly advancing AI technologies 2. Enable and facilitate industry-academia partnerships in research and development, and foster relationships through collaborative projects 3. Encourage cross-disciplinary research in applied computing, in critical scientific and industrial domains, via research proposal submissions to funding agencies 4. Provide a state-of-the-art R&D facility for students, faculty and collaborators 5. Offer a comprehensive and meaningful computing environment for education by: 1. complementing the theoretical coursework in CC with appropriate laboratory coursework for students, and 2. encouraging team participation and cross-disciplinary problem solving
- 8. © 2017 IBM Corporation IBM’s AI Lab OpenPOWER System for Data Analytics with Accelerators (GPU) Collaborative technical projects Access to IBM Academic Initiative Toolkit Graduate, Ph.D. and Post-Doctoral research Webinars and Technical Workshops Projects related to make smart cities and smart villages
- 9. © 2017 IBM Corporation OpenPOWER System for Data Analytics with Accelerators (GPU) Collaborative technical projects Access to IBM Academic Initiative Toolkit Graduate, Ph.D. and Post-Doctoral research Webinars and Technical Workshops Projects related to make smart cities and smart villages
- 10. © 2017 IBM Corporation10 Proposed AI cloud setup and specifications - Hardware College Ethernet Network 4 4 College Lan Network College Ethernet Network 10 Desktops/Laptops 2 Jetson Nano Edge Devices
- 11. © 2017 IBM Corporation11 AI Lab users AI Lab Software Components
- 12. University Use Cases and Scenarios of Proposed AI Lab AI Cloud at Universities
- 13. © 2017 IBM Corporation13 Use Case 1 : Students (daily use) requests for compute resource Basic ML/DL exercises Login to web portal with “Student” profile; browse service catalog. Select and request for desired image, and usage period eg. MS Office with Windows for 2 hours. Login and access Docker Container (Remote Desktop) AI Cloud Portal AI Cloud Infrastructure User / profile authentication Service request processing & approval VM & storage created according to request OS deployed into Docker Container Application image deployed into Docker Container Login info sent to user via email Docker Container with PowerAI image f Downloads completed work into laptop and logs off. Resources made available to students for daily use will be restricted. The restriction will be enforced through profile management on the cloud portal. Students Students login from anywhere within the UM LAN. Cloud portal is accessed via a web browser. Application and OS images have to be preconfigured by the cloud admin before use.
- 14. © 2017 IBM Corporation14 Use Case 2 : Final Year Students requests for compute resource for AI Projects Login to webportal with “FY” profile; browse service catalog. Select and request for desired image, and usage period e Login and access Docker Container (Remote Desktop) AI Cloud Portal Cloud Infrastructure User / profile authentication Service request processing & approval VM & storage created according to request OS deployed into VM Application image deployed into VM Login info sent to user via email VM deprovisioned back into the cloud Downloads completed work into laptop and logs off. Resources made available to final year students will be restricted. The restriction will be enforced through profile management on the cloud portal. FY Students Students login from anywhere within the UM LAN. Cloud portal is accessed via a web browser. IP address for VM deployed within same subnet. Students access from laptop. VM and Storage size : 2-4 cores, 4GB RAM, 10GB storage RHEL Jetson Nano VM for the FY student will be operational until the expiration date stated in his request.
- 15. © 2017 IBM Corporation15 Use Case 3 : Final Year Students creates own application image, and shares image with other FY students. Student to seek approval from Cloud Admin to create new app image in cloud infra New image is displayed in the service catalog Other FY students proceed to request, access and use new application (as per Use Case 2) Ai Cloud Admin AI Cloud Infrastructure To ensure proper cloud operations, only the cloud administrator is allowed to manage image offereings in the cloud. FY Students In order to allow other FY students to have access to the new application image for their own project, the originator of the application has to work with the cloud admin to package the app as an image offering in the cloud. Cloud Portal Provisioning manager packages app image with OS Image is registered with service automation manager and portal User / profile authentication Service request processing VM & storage created according to request OS deployed into VM Application image deployed into VM Login info sent to user via email VM deprovisioned back into the cloud
- 16. © 2017 IBM Corporation16 During 2 hr class, provides VM login information to 40 students in class / exam Use Case 4 : Lecturers prebooking seats for AI/ML/DL class or exam AI Cloud Portal AI Cloud Infrastructure User / profile authentication Service request processing VM & storage created according to request OS deployed into VM Application image deployed into VM Login info sent to lecturer via email VMs deprovisioned back into the cloud Resources made available to students for daily use will be restricted. The restriction will be enforced through profile management on the cloud portal. Lecturers VM and Storage size : 40 VMs 2 core, 4GB RAM, 5GB storage RHEL PowerAI Vision Watson Machine Learning Accelerator Lecturer proceed to request for VMs with “Lecturer” profile. Select and request for desired image, and future usage period eg. 40 VMs of SPSS with LInux for 2 hours. Students access VMs from laptop / PC / workstations Students download work at end of class Application and OS images have to be preconfigured by the cloud admin before use. IP address for VM deployed within same subnet.
- 17. © 2017 IBM Corporation17 Use Case 5 : Researchers adding compute capacity with own applications through the AI cloud AI Cloud Portal Ai Cloud InfrastructureResearchers Researchers proceed to request, access VM and install own application (as per Use Case 2) User / profile authentication Service request processing VM & storage created according to request OS deployed into VM Application image deployed into VM Login info sent to user via email VM deprovisioned back into the cloud VM and Storage size : 8 cores, 16GB RAM, 250GB storage RHEL
- 18. © 2017 IBM Corporation 2 Year Developmental Timeline a) IBM POWER Academic Initiative partnership b) OpenPOWER system and Accelerator for Deep Learning and Machine Learning c) Technical Projects deployment d) Review of progress in technical projects, lab coursework e) Big data and AI curriculums
- 19. © 2017 IBM Corporation AI Cloud (On Premise) PowerAI makes deep learning, machine learning and AI more accessible and more performant By combining this software platform for deep learning with IBM Power Systems, enterprises and Institutions can rapidly deploy a fully optimized and supported platform for machine learning frameworks and their dependencies. And it is built for easy and rapid deployment PowerAI runs on the IBM Power System AC922 for High Performance Computer server infrastructure
- 20. © 2017 IBM Corporation Advantages for Your Faculty and Students Talent and Skills: (Remote Interns; Skills and Training) Students and Research scholars will start working on the advanced technologies will enable them to work on many applications Publications and Mindshare: (Press releases, Articles, and Publications; Conferences and Events) 1. Conference Paper on software-based application research /development in 6 months Intellectual Capital: (Patents, Open source; Prototypes, Demos; Curriculum; Student projects, Theses) 1. Prototype building of many research problems using software-centric approach (hardware-centric baseline implementation almost getting completed) 2. Potential to file disclosures Opportunities: (Seed revenue; Leverage other funding; Build ecosystems; Build government/client relationships) 1. Once software-centric solution available with comparable performance using latest technologies , your team would create prototypes which can be demonstrated to several colleges
- 21. © 2017 IBM Corporation 21 Ganesan Narayanasamy ganesana@in.ibm.com OpenPOWER leader in Education and Research WW IBM Systems Thank you!