Super Computing 18, MC04 Building your own mini-CORAL : Power Accelerated Computing Platform - IBM LSF & HPC User Group @ SC18 - IBM ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
IBM LSF & HPC User Group @ SC18
Super Computing 18, MC04 Building your own mini-
CORAL : Power Accelerated Computing Platform
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation
IBM LSF & HPC User Group @ SC18
Agenda
• IBM Power Accelerated Computing Platform requirements
• Structure of Power Accelerated Computing Platform
• Lessons learned deploying large CORAL HPC Clusters
• How to get started with Power Accelerated Computing Platform
• Discussion
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation 2
IBM LSF & HPC User Group @ SC18
IBM Power
Accelerated Computing Platform
IBM Power ACP gives clients their own AI
installation based upon the world’s most
powerful and smartest scientific
supercomputer
Supports
• High Performance Computing (HPC)
• Artificial Intelligence (AI)
• Machine Learning / Deep Learning
Based upon IBM CORAL!
Natural markets: Research Labs, Universities,
3
Government Labs, Military Research, Industry
IBM LSF & HPC User Group @ SC18
Questions?
Complete Solutions for AI and Modern HPC
– CORAL Servers (POWER9 – IBM Power System AC922)
How Do I Deploy AI
– Management Servers/Head Nodes
at my Company?
– Networking : Ethernet and IB
– Elastic Storage Server
I want to run Workloads and – Linux and Software Development tools
Experiments on Summit! – Pre-Sales/Install expert review by IBM Systems Lab
Services
– Hardware Configuration assembly in IBM facility
– Software Installation and Configuration by IBM before
I want to explore delivery
Quantum Computing
– Installation and connectivity support with IBM Systems Lab
Services
– Software Flexibility: HPC and/or PowerAI base or PowerAI
Enterprise, and/or H2O
Power AI Reference Architecture:
https://ibm.ent.box.com/s/8w75cdh6s4smgix7ckoh4yisn06h93iw
4
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation
IBM LSF & HPC User Group @ SC18
CORAL and Summit & Sierra
CORAL = Collaboration of Oak Ridge, Argonne & Lawrence Livermore
National Labs
Summit, Ascent and Peak are cluster names of Oak Ridge
Sierra, Lassen, Ansel and Butte are cluster names at Lawrence Livermore
Group Name / DOC ID / Month XX, 2018 / © 2018 IBM Corporation 5
IBM LSF & HPC User Group @ SC18
Group Name / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 6
IBM LSF & HPC User Group @ SC18
Group Name / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 7
IBM LSF & HPC User Group @ SC18
IBM LSF & HPC User Group @ SC18
IBM POWER SYSTEM
AC922
An Acceleration Superhighway
Unleash state of the art IO and
accelerated computing potential in
the post “CPU-only” era
Designed for the AI Era
Architected for the modern analytics
and AI workloads that fuel insights
Delivering Enterprise-
Enterprise-Class AI
Flatten the time to AI value curve
by accelerating the journey to build,
train, and infer deep neural networks
IBM LSF & HPC User Group @ SC18
The POWER9 processor
1stchip 2x Core performance
vs x86
~1TB/s with PCIe4
4GHZ
BW into PEAK performance
chip 7TB/s
On chip
FREQUENCY 1.5x vs POWER8
BW 8
>15
BILLION
TRANSISTORS
>24B 2x more memory
vs POWER8
MILES OF VIAS
WIRE 17 More memory
LEVELS
OF METAL
1.4x bandwidth vs x86IBM LSF & HPC User Group @ SC18
Watching Processors
Evolve!
HPC analyst Addison Snell (CEO of Intersect360
Research) ….commented by email.
“One, Power9 has excellent memory bandwidth
and performance.
Two, it is a great platform for attaching accelerators
or co-processors. It’s an odd statement of
direction, but maybe a visionary one,
essentially saying a processor isn’t
about computation per se, but rather
it’s about feeding data to other
computational elements.”
IBM and Business Partner Use OnlyIBM LSF & HPC User Group @ SC18
IBM Power System AC922 - POWER9 with increased GPU and IO bandwidth for differentiation
Realize unprecedented performance and application gains with POWER9 and NVLink 2.0
• 2 POWER9 CPUs and up to 4 “Volta” NVLink 2.0 GPUs in a versatile 2U Linux server
• PCIe Gen4 bus has double I/O Bandwidth vs. PCIe Gen3
• CPU (Turbo)/GPU (Boost) enabled for improved data center efficiency and performance to be
maintained at high levels (3.3 / 3.45ghz, air/water).
High level System Overview
2-Socket, 2U Packaging
32, 40 (air) or 36,44 (water) P9 Processor cores
4 NVIDIA Volta V100 NVLink2 GPUs
2 TB Memory (16x - 128GB DIMMs)
4 PCIe Gen4 Slots
2x SFF (HDD/SSD), SATA, Up to 7.7 TB storage
Supports 1.6, 3.2 and 6.4TB NVMe Adapters
Redundant Hot Swap Power Supplies and Fans
Default 3 year 9x5 warranty, 100% CRUIBM LSF & HPC User Group @ SC18
IBM Spectrum LSF Suites
Powerful Workload Management
The suite delivers:
• Enhanced Utilization of assets through effective
scheduling and sharing policies
• Enhancing User Productivity through ease of
use, accessibility and simplification
• Operational Efficiency through insight of how the
HPC environment is being used
Comprehensive GPU, Container and Hybrid Cloud
Support
The LSF Suite for HPC is available at no charge via
the IBM Academic Initiative
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation 13IBM LSF & HPC User Group @ SC18
AI Changes Everything for Data
Diversity of Data
– Local, HDFS, NFS, Posix, Cloud
Amount of Data
– A Petabyte is just a starting point
Delivery of Data
– Gigabytes/Sec/Server to feed GPU
14IBM LSF & HPC User Group @ SC18
IBM Spectrum Scale with Elastic Storage Server Family
The IBM ESS Family
The Storage Built for AI!
• Over 1000 ESS Installed
• Over 300 ESS customers
• Over 5,000 Spectrum Scale clients
Five 9’s Reliability!
IBM is the World Leader in Software Defined Storage
EnvironmentsIBM LSF & HPC User Group @ SC18
ESS Installation at ORNL
77 ESS Systems delivering:
• Single Namespace up to 250 Petabytes
• 2.5 TB/s large block sequential IO performance
• 2.6M file creates/sec for 32KB files in unique
directories
• 50K file creates/sec to single shared directory
• Spectrum Scale RAID with declustered erasure
coding
• 16 GB/Second of Data I/O to a Single ServerIBM LSF & HPC User Group @ SC18
IBM Elastic Storage Server (ESS) Family
Speed Hybrid Capacity
Model GS1S Model GL1S:
1 Enclosures, 9U Model GL6S:
24 SSD
82 NL-SAS, 2 SSD 6 Enclosures, 28U
502 NL-SAS, 2 SSD
ESS 5U84
Storage
ESS 5U84
Model GH24S: Storage
14 GB/s Model GH14S:
2 2U24 Enclosure SSD
4 5U84 Enclosure HDD Model GL4S:
1 2U24 Enclosure SSD 334 NL-SAS, 48 SSD 4 Enclosures, 20U
4 5U84 Enclosure HDD 334 NL-SAS, 2 SSD ESS 5U84
Storage
Model GS2S 334 NL-SAS, 24 SSD
48 SSD
6 GB/s ESS 5U84
Storage ESS 5U84
Storage
ESS 5U84 Storage ESS 5U84 Storage
Model GL2S:
2 Enclosures, 12U
ESS 5U84
166 NL-SAS, 2 SSD Storage
26 GB/s ESS 5U84 Storage ESS 5U84 Storage
ESS 5U84
Storage
ESS 5U84
Storage
Model GS4S
96 SSD
ESS 5U84
Storage ESS 5U84
Storage
ESS 5U84 Storage ESS 5U84 Storage
ESS 5U84 ESS 5U84
ESS 5U84
Storage Storage
ESS 5U84 Storage ESS 5U84 Storage Storage
40 GB/s 38 GB/s 40 GB/s 12 GB/s 24 GB/s 36 GB/s
IBM Systems | 17IBM LSF & HPC User Group @ SC18
New ESS C-Series
Maximum Density with Room to Upgrade and Grow!
New! Model GL6C:
6 Enclosures, 28U
634 NL-SAS, 2 SSD
New! Model GL4C
4 Enclosures, 16U
432 NL-SAS, 2 SSD 4U106
Storage
New! Model GL2C: 4U106
Storage
2 Enclosures, 12U
210 NL-SAS, 2 SSD 4U106
Storage 4U106
Storage
New! Model GL2C:
4U106
1 Enclosure, 8U Storage
104 NL-SAS, 2 SSD
4U106
Storage
4U106
Storage
4U106
Storage 4U106
4U106
Storage
Storage
4U106 4U106 4U106
Storage Storage Storage
1.0 PB Disk 2.0 PB Disk 4.2 PB Disk 6.3 PB Disk
© IBM Corporation 2018 18IBM LSF & HPC User Group @ SC18
Power Accelerated Computing Platform – Sample Building Block View
1-4 S42 Racks
Mellanox AC992: The World’s Premier AI Servers
Switches
• Featured in ORNL and LLNL CORAL Installs
• ExaOps of demonstrated AI Performance
Compute:
• Able to Process more than 20 GB/S of Data
AC922 • Add Servers as Workloads Grow!
2 or 4 GPUs
Management
L922 or AC922
IBM Elastic Storage Server for AI Workloads
• Density meets Performance
• High Density Petabytes in Minimum Space
Elastic Storage • Featured in ORNL and LLNL Installs
Server
• Grow Performance by Scaling Up or Out!
(5147 & 5148)
• Supports IB and Ethernet!
IBM SystemsIBM LSF & HPC User Group @ SC18
PowerAI Integrated & Supported AI Platform
Open-Source Based 3-4x Speedup for AI Training
Enterprise AI Platform Ease of Use Tools for Data Scientists
Developer Ease-of-Use Tools
Open Source Frameworks:
Supported Distribution
Caffe
SnapML
Faster Training Times via
HW & SW Performance Optimizations
GPU-Accelerated20 Storage
Power ServersIBM LSF & HPC User Group @ SC18
5x Faster Data Communication with Unique
CPU-GPU NVLink High-Speed Connection
Store Large Models in 1 TB 1 TB
System Memory Memory Memory
170GB/s 170GB/s
Fast Transfer POWER9 POWER9
via NVIDIA CPU CPU
NVLink NVLink NVLink
150 GB/s 150 GB/s
Operate on One V100 GPU V100 GPU V100 GPU V100 GPU
Layer at a Time
IBM Power System AC922
Deep Learning Server (4-GPU Config)
21IBM LSF & HPC User Group @ SC18
Auto-ML for Images & Video
PowerAI
Vision
PowerAI: Open Source ML Frameworks
PowerAI SnapML
Large Model Support (LMS) Deep Learning Impact
(DLI) Module
Distributed Deep Learning Auto-Hyper Data & Model
(DDL) Parameter Tuning Management, ETL,
PowerAI Visualize, Advise
Enterprise
Accelerated
Infrastructure
Accelerated Servers Storage
22IBM LSF & HPC User Group @ SC18
Increased
Faster Time Simplified Enterprise
Resource
to Results Management Solution
UtilizationIBM LSF & HPC User Group @ SC18
Power AI Enterprise Project Examples
Industry Scenario Industry Scenario
Credit Scoring Network cabling detection
Telcom
Banking Face Masking Detection Service halt handling
Stock Index Futures Prediction LED Panel defect inspection
Manufacturing
Research Exploration Steel quality classification
OCR recognition correction Wafer Flaw detection
Company Logo and name auto matching Energy Power transmission line safety detection
Securities
AI on cloud Healthcare Pathologic analysis
Hand writing recognition Retail Retail market analysis via image recognition
Insurance Work order auto clustering/handling
Public Satellite photo fault reorganization
Transportation Train & subway defect inspection
© 2017 IBM Corporation 24IBM LSF & HPC User Group @ SC18
PowerAI Vision: “Point-and-Click” AI for Images & Video
Label Image or Auto-Train AI Model Package & Deploy
Video Data AI ModelIBM LSF & HPC User Group @ SC18
PowerAI Vision Project
Examples
Defect Identification
•Wafer Fab Inspection – Electronics
•Cam Shaft Inspection – Automotive
•Seat Inspection – Automotive
•PCBA Inspection – Electronics
•Utility disk Inspection – Energy/Utilities
•Mainframe assemble inspection – Electronics
•Ceramic capacitor - Electronics
•Defective Components – Oil/Gas
Facial / Object Recognition
•Safety/Security - Transit, Banking, Gaming
•Building Infrastructure – Building/Construction
•Service – Retail, Food
•Traffic – Municipal
IBM Supply Chain Engineering / DOC ID / / © 2017 IBM Corporation 26IBM LSF & HPC User Group @ SC18
Power Accelerated Computing Platform – Building Blocks
1-4 S42 Racks
Hardware Building Blocks
IB TOR switch
IB and Ethernet Enet TOR switch
Mellanox Switches (Mellanox)
Switches (Shared w/ESS)
AC922 Compute Nodes
8335-GTG 4 – 15 Compute Servers*
2 or 4 GPUs
xCAT / Manager /
Login node
1–3
9008-22L or Management / Login Servers ESS mgmt. node or
8335-GTG (1st rack) protocol nodes
ESS
Elastic Storage 0-1 ESS per cluster
Server (optional, 1st rack)
(5147 & 5148)
* 7 max in 1st rack
15 max in 2nd - 4th
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation | 27
27IBM LSF & HPC User Group @ SC18
Power Accelerated Computing Platform
Configurable HW to simplify creation of “CORAL Like” scale out clusters
Storage Compute Management Switches Rack
Elastic Storage AC922 (2 or 4 GPU) L922+ and/or Mellanox One to four
Server 8335-GTG AC922 (0,2,4 GPU) 42U Racks
8335-GTG (S42)
Optional Air Cooled Only 100Gb If you really
Same Processors as in InfiniBand need more,
CORAL Servers 40Gb Ethernet let us know!
10Gb Ethernet
1Gb Ethernet
-
Configurable to support HPC, Power AI, and in the future, Quantum Simulator stacks
- Simplifies ability to configure complex configs for scale out infrastructure
- Software customization & fully rack integrated in IBM manufacturing
- Determined in IBM System Lab Services Implementation Design Workshop
- Optional On-Site network Integration and knowledge transfer available
- Option to assemble in Rochester, MN Pre-build lab if customer wants to use their own switches, racks or desire Water
Cooled AC922 Compute processors
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation 28IBM LSF & HPC User Group @ SC18
Software that can be customized at IBM Manufacturing *
Red Hat OS 7.5 (5639-RLE)
IBM Spectrum Scale Client Optional frameworks/levels as identified in the
Base Mellanox OFED driver (Mellanox) Implementation Design Workshop :
NVIDIA CUDA Software (Nvidia) Anaconda
Caffe
IBM Advanced Toolchain
PowerAI Base (5765-PAI) Jupyter Notebook
PowerAI Enterprise (5765-AIE) Keras
AI Spectrum Conductor Python
DL Impact Optional Open
PyTorch Source for P9
PowerAI TensorFlow
PowerAI Vision (5737-H10) xCAT
H2O Driverless AI (5639-AIH) XGBOOST (latest git code)
IBM Spectrum LSF Suite (5737-F30)
IBM Compilers – XLC/C++/Fortran, gcc
HPC ESSL (5765-L61)
IBM Spectrum MPI (5725-G83)
Performance Toolkit (5765-PD2)
xCAT support (5771-CAT)
* Assuming customer has required licenses (design workshop)
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation 29IBM LSF & HPC User Group @ SC18
How do I get started?
What use cases in my company will have payback?
Who can help my company customize the software?
Who can provide knowledge transfer to my personnel?
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation 30IBM LSF & HPC User Group @ SC18
Cognitive Discovery Workshop: Helping you identify the right cognitive use cases
Objective: To provide an overview of Cognitive technologies, explore potential uses cases and how they can
be deployed to provide business value. The key focus is to identify potential use cases for Proof
of Concept project.
How’s it Delivered ? A 4-6 hour Face to Face workshop at customer location delivered by a IBM
Cognitive Workshop team
What’s the output ? Potential use cases and an action plan to help team select an appropriate Cognitive
project.
Who should attend ? Key IT resources, Data Scientist/Customer Data Architect, LOB(Business Sponsor),
any others the customer team feels are important to the discussion.
Detailed abstract: This session typically includes discussions on:
Overview of industry and cross industry use cases
• Discussion of Open Source Cognitive technologies such as Tensorflow, Caffe, Theano, Torch,
• Discussion on data layer technologies such as Hadoop, NoSQL, NewSQL and relational DB technologies and
the Importance of End to End process (Governance and Data management)
• Discussion of Customer Specific use cases including feasibility assessment.
• Develop action plan to assist the customer to Identify and justify Cognitive use cases (ROI or or ROI factors)
• ID infrastructure actions necessary to support Cognitive project
Email: cssc@us.ibm.com
Submit Online Request: https://ibm.biz/BdFfcV
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation 31IBM LSF & HPC User Group @ SC18
Discovery Workshop
Time Topic Speaker Audience
9:00-9:15 am Introductions and Review Workshop Objectives All Execs, LOB, IT
Executive
Liaisons
Session
9:15-10:45 am Executive Session IBM Execs, LOB, IT
-What is AI Liaisons
-Art of the Possible
-Short Demo – H2O
10:45 – 11:00am Break
11:00 – 11:45 pm Introduce Use Case Workshop LOB, IT Liaisons
-Answering lingering Q&A
Use Case -Each LOB department mission overview & focus areas
Discovery 11:45 - 12:30 pm Industry Examples of Applied AI IBM/Client LOB, IT Liaisons
-Group Discussion on applicability to Customer
12:30 – 1:00 pm Lunch
1:00 – 2:30 pm Discussion and Identification of Use cases by LOB. IBM/Client LOB, IT Liaisons
-Feasibility and Impact of Use Cases
-Identify High Interest and Highest Value Use Cases for Customer
2:30 – 2:45 pm Break
Business Case 32
2:45 – 4:00 pm Develop Action Plan for Creation of Exec Proposal for High Value IBM/Client LOB, IT Liaison
Development
Use Cases for identified use
-Use Case Pay Back, Cognitive Work Flow, Timeline, Data Strategy cases
-Cognitive Skill Set, Data Strategy, POC/Trial Implementation stepsIBM LSF & HPC User Group @ SC18
Power ACP – IBM Systems Lab Services
Manufacturing Install
Implementation Design Hardware Racking, Network Integration &
Workshop Software Knowledge Transfer on
Customization in site
Manufacturing
- Develops information to enable - Install, Configure & Verify software - Optional network integration
majority of system implementation - Done on customer site
and tailoring to occur in IBM - Billable to customer
Manufacturing - Knowledge Transfer on solution
- Done on customer site configuration
Note: This step mandatory for enabling
manufacturing SW preload
Contact us today fdrobin@us.ibm.com
On the Web: www.ibm.com/it-infrastructure/services/lab-services PartnerWorld:
www.ibm.com/partnerworld/systems/services/lab-services Email us:
ibmsls@us.ibm.com
33
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM CorporationIBM LSF & HPC User Group @ SC18
IBM Systems Lab Services Implementation Design Workshop
Onsite customer workshop to enable a fast time-
time-to-
to-benefit implementation
- Develops information to enable majority of system implementation and tailoring to occur in
IBM Manufacturing
- Documents software and infrastructure required to enable customer use cases
- Includes:
- Data Center personnel to ensure client data center is ready for the Power
Accelerated Computing Platform implementation
- Customer personnel to determine customization of software like PowerAI
Enterprise or PowerAI Vision or H20
- Client networking team to document customization needed for networking (IPs,
VLANS, Uplinks, etc)
- Creation of the implementation documentation that will be used for customization at
IBM Manufacturing and for solution knowledge transfer
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation 34IBM LSF & HPC User Group @ SC18
End Result at the Data Center
This Not This
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation 35IBM LSF & HPC User Group @ SC18
Lessons learned with Summit on deploying large
HPC Clusters
IBM Systems Lab Services/ SC18 / November, 2018 / © 2018 IBM Corporation
36IBM LSF & HPC User Group @ SC18
Group Name / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 37IBM LSF & HPC User Group @ SC18
Deployment of Large HPC Clusters Lessons Learned
Architecture for scale is important. In our case, the network architecture was quite successful, and service
nodes were used to distribute provisioning workload across many nodes.
Most of the effort in deploying a large cluster is in the infrastructure racks
Switch-level discovery becomes critical for large-scale rapid deployment of racks. Cabling verification and
double-checking node positions became important.
It's important to establish a good, complete set of node-level diagnostics to run on every node in the cluster,
and to run this set of diagnostics on a continuous basis
Establish a process and mechanism to deploy updates continuously to the cluster, for both software and
firmware. This includes both stateful and stateless nodes.
Expect issues at scale with most tools
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation 38IBM LSF & HPC User Group @ SC18
Performance Testing as you go
One of the final objectives for the cluster deployment was a submission to the Top 500
For Sierra, HPL (Linpack) became an extraordinarily valuable tool for exercising a cluster, and finding and
diagnosing performance issues
We started small at the node level, and worked up to the rack level, row level and cluster level. In this way, we
could identify performance issues at the micro level, rather than the macro level. When tuned well, node level
and rack level performance was remarkably similar.
Node level HPL identifies CPU, GPU and memory performance issues
Rack-level HPL identifies Infiniband performance issues both at individual nodes and at the rack-level IB
switches
Row-level HPL identifies performance issues in some core IB switches. For example, we saw performance
issues in the eastern end of one row in Sierra
Cluster-level HPL identifies issues at very large scale, and provides opportunities for novel approaches to HPL
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation 39IBM LSF & HPC User Group @ SC18
Power Accelerated Computing Platform
Getting Started
• IBM Cognitive Systems Solution Center (CSSC)
Optional Discovery Workshop to identify use cases
• Email: cssc@us.ibm.com
• Submit Online Request: https://ibm.biz/BdFfcV
• IBM Systems Lab Services three Stage Approach
i. Implementation Design Workshop
ii. Manufacturing Customization
iii. Data Center Integration
• Email: ibmsls@us.ibm.com or
• Fred Robinson fdrobin@us.ibm.com
• Configurator: eConfig -> Power -> Solutions -> Power
ACP
IBM Systems Lab Services/ SC18 / November , 2018 / © 2018 IBM Corporation | 40IBM LSF & HPC User Group @ SC18
IBM Systems Lab Services
Proven expertise to help leaders plan, design, and implement the essential IT infrastructure for what comes next
Our team of 1,000+ consultants, engage
worldwide in pre and post sales
opportunities in:
Power Systems
Storage and Software Defined
Infrastructure
IBM Z and LinuxONE
HPC & Deep Learning
Systems Consulting
Migration Factory ibmsls@us.ibm.com
www.ibm.com/it-infrastructure/services/lab-services
Technical Training and Events Fred Robinson fdrobin@us.ibm.comIBM LSF & HPC User Group @ SC18
IBM Power
Accelerated Computing Platform
IBM Power ACP gives clients their own AI
installation based upon the world’s most
powerful and smartest scientific
supercomputer
Includes everything required for success!
• Networking
• Servers
• Storage
• Software
• Services
• Support
Leverage CORAL success TODAY! 42IBM LSF & HPC User Group @ SC18
Notices and disclaimers
• © 2018 International Business Machines Corporation. No part of • Performance data contained herein was generally obtained in a
this document may be reproduced or transmitted in any form without controlled, isolated environments. Customer examples are presented
written permission from IBM. as illustrations of how those
• U.S. Government Users Restricted Rights — use, duplication or • customers have used IBM products and the results they may have
disclosure restricted by GSA ADP Schedule Contract with IBM. achieved. Actual performance, cost, savings or other results in other
operating environments may vary.
• Information in these presentations (including information relating to
products that have not yet been announced by IBM) has been • References in this document to IBM products, programs, or services
reviewed for accuracy as of the date of initial publication and could does not imply that IBM intends to make such products, programs or
include unintentional technical or typographical errors. IBM shall services available in all countries in which IBM operates or does
have no responsibility to update this information. This document is business.
distributed “as is” without any warranty, either express or • Workshops, sessions and associated materials may have been
implied. In no event, shall IBM be liable for any damage arising prepared by independent session speakers, and do not necessarily
from the use of this information, including but not limited to, reflect the views of IBM. All materials and discussions are provided for
loss of data, business interruption, loss of profit or loss of informational purposes only, and are neither intended to, nor shall
opportunity. IBM products and services are warranted per the constitute legal or other guidance or advice to any individual
terms and conditions of the agreements under which they are participant or their specific situation.
provided.
• It is the customer’s responsibility to insure its own compliance
• IBM products are manufactured from new parts or new and used
with legal requirements and to obtain advice of competent legal
parts.
counsel as to the identification and interpretation of any relevant laws
In some cases, a product may not be new and may have been
and regulatory requirements that may affect the customer’s business
previously installed. Regardless, our warranty terms apply.”
and any actions the customer may need to take to comply with such
• Any statements regarding IBM's future direction, intent or laws. IBM does not provide legal advice or represent or warrant that its
product plans are subject to change or withdrawal without services or products will ensure that the customer follows any law.
notice.
© Copyright IBM Corporation 2018 43IBM LSF & HPC User Group @ SC18
Notices and disclaimers
continued
• Information concerning non-IBM products was obtained from the suppliers of • IBM, the IBM logo, ibm.com and [names of other referenced IBM
those products, their published announcements or other publicly available products and services used in the presentation] are trademarks
sources. IBM has not tested those products about this publication and cannot of International Business Machines Corporation, registered in
confirm the accuracy of performance, compatibility or any other claims related many jurisdictions worldwide. Other product and service names
to non-IBM products. Questions on the capabilities of non-IBM products should might be trademarks of IBM or other companies. A current list of
be addressed to the suppliers of those products. IBM does not warrant the IBM trademarks is available on the Web at "Copyright and
quality of any third-party products, or the ability of any such third-party products trademark information" at: www.ibm.com/legal/copytrade.shtml.
to interoperate with IBM’s products. IBM expressly disclaims all warranties, • .
expressed or implied, including but not limited to, the implied warranties
of merchantability and fitness for a purpose.
• The provision of the information contained herein is not intended to, and does
not, grant any right or license under any IBM patents, copyrights, trademarks or
other intellectual property right.
© Copyright IBM Corporation 2018 44You can also read
