Soumettre la recherche
Mettre en ligne
BSC LMS DDL
•
2 j'aime
•
722 vues
G
Ganesan Narayanasamy
Suivre
Large Model Support and Distributed Deep Learning features
Lire moins
Lire la suite
Technologie
Signaler
Partager
Signaler
Partager
1 sur 21
Télécharger maintenant
Télécharger pour lire hors ligne
Recommandé
CFD on Power
CFD on Power
Ganesan Narayanasamy
2018 bsc power9 and power ai
2018 bsc power9 and power ai
Ganesan Narayanasamy
SNAP MACHINE LEARNING
SNAP MACHINE LEARNING
Ganesan Narayanasamy
AI OpenPOWER Academia Discussion Group
AI OpenPOWER Academia Discussion Group
Ganesan Narayanasamy
OpenPOWER Webinar
OpenPOWER Webinar
Ganesan Narayanasamy
TAU E4S ON OpenPOWER /POWER9 platform
TAU E4S ON OpenPOWER /POWER9 platform
Ganesan Narayanasamy
Covid-19 Response Capability with Power Systems
Covid-19 Response Capability with Power Systems
Ganesan Narayanasamy
MIT's experience on OpenPOWER/POWER 9 platform
MIT's experience on OpenPOWER/POWER 9 platform
Ganesan Narayanasamy
Recommandé
CFD on Power
CFD on Power
Ganesan Narayanasamy
2018 bsc power9 and power ai
2018 bsc power9 and power ai
Ganesan Narayanasamy
SNAP MACHINE LEARNING
SNAP MACHINE LEARNING
Ganesan Narayanasamy
AI OpenPOWER Academia Discussion Group
AI OpenPOWER Academia Discussion Group
Ganesan Narayanasamy
OpenPOWER Webinar
OpenPOWER Webinar
Ganesan Narayanasamy
TAU E4S ON OpenPOWER /POWER9 platform
TAU E4S ON OpenPOWER /POWER9 platform
Ganesan Narayanasamy
Covid-19 Response Capability with Power Systems
Covid-19 Response Capability with Power Systems
Ganesan Narayanasamy
MIT's experience on OpenPOWER/POWER 9 platform
MIT's experience on OpenPOWER/POWER 9 platform
Ganesan Narayanasamy
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
Indrajit Poddar
OpenPOWER Webinar on Machine Learning for Academic Research
OpenPOWER Webinar on Machine Learning for Academic Research
Ganesan Narayanasamy
OpenPOWER/POWER9 AI webinar
OpenPOWER/POWER9 AI webinar
Ganesan Narayanasamy
WML OpenPOWER presentation
WML OpenPOWER presentation
Ganesan Narayanasamy
IBM HPC Transformation with AI
IBM HPC Transformation with AI
Ganesan Narayanasamy
Ac922 cdac webinar
Ac922 cdac webinar
Ganesan Narayanasamy
OpenPOWER/POWER9 Webinar from MIT and IBM
OpenPOWER/POWER9 Webinar from MIT and IBM
Ganesan Narayanasamy
PowerAI Deep dive
PowerAI Deep dive
Ganesan Narayanasamy
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
Ganesan Narayanasamy
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
Ganesan Narayanasamy
Deeplearningusingcloudpakfordata
Deeplearningusingcloudpakfordata
Ganesan Narayanasamy
Summit workshop thompto
Summit workshop thompto
Ganesan Narayanasamy
IBM BOA for POWER
IBM BOA for POWER
Ganesan Narayanasamy
JMI Techtalk: 한재근 - How to use GPU for developing AI
JMI Techtalk: 한재근 - How to use GPU for developing AI
Lablup Inc.
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Ganesan Narayanasamy
Large Model support and Distribute deep learning
Large Model support and Distribute deep learning
Ganesan Narayanasamy
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Lablup Inc.
Transparent Hardware Acceleration for Deep Learning
Transparent Hardware Acceleration for Deep Learning
Indrajit Poddar
A Primer on FPGAs - Field Programmable Gate Arrays
A Primer on FPGAs - Field Programmable Gate Arrays
Taylor Riggan
AMD It's Time to ROC
AMD It's Time to ROC
inside-BigData.com
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
DataWorks Summit/Hadoop Summit
IBM Cloud Paris Meetup - 20190520 - IA & Power
IBM Cloud Paris Meetup - 20190520 - IA & Power
IBM France Lab
Contenu connexe
Tendances
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
Indrajit Poddar
OpenPOWER Webinar on Machine Learning for Academic Research
OpenPOWER Webinar on Machine Learning for Academic Research
Ganesan Narayanasamy
OpenPOWER/POWER9 AI webinar
OpenPOWER/POWER9 AI webinar
Ganesan Narayanasamy
WML OpenPOWER presentation
WML OpenPOWER presentation
Ganesan Narayanasamy
IBM HPC Transformation with AI
IBM HPC Transformation with AI
Ganesan Narayanasamy
Ac922 cdac webinar
Ac922 cdac webinar
Ganesan Narayanasamy
OpenPOWER/POWER9 Webinar from MIT and IBM
OpenPOWER/POWER9 Webinar from MIT and IBM
Ganesan Narayanasamy
PowerAI Deep dive
PowerAI Deep dive
Ganesan Narayanasamy
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
Ganesan Narayanasamy
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
Ganesan Narayanasamy
Deeplearningusingcloudpakfordata
Deeplearningusingcloudpakfordata
Ganesan Narayanasamy
Summit workshop thompto
Summit workshop thompto
Ganesan Narayanasamy
IBM BOA for POWER
IBM BOA for POWER
Ganesan Narayanasamy
JMI Techtalk: 한재근 - How to use GPU for developing AI
JMI Techtalk: 한재근 - How to use GPU for developing AI
Lablup Inc.
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Ganesan Narayanasamy
Large Model support and Distribute deep learning
Large Model support and Distribute deep learning
Ganesan Narayanasamy
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Lablup Inc.
Transparent Hardware Acceleration for Deep Learning
Transparent Hardware Acceleration for Deep Learning
Indrajit Poddar
A Primer on FPGAs - Field Programmable Gate Arrays
A Primer on FPGAs - Field Programmable Gate Arrays
Taylor Riggan
AMD It's Time to ROC
AMD It's Time to ROC
inside-BigData.com
Tendances
(20)
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
Optimizing Hortonworks Apache Spark machine learning workloads for contempora...
OpenPOWER Webinar on Machine Learning for Academic Research
OpenPOWER Webinar on Machine Learning for Academic Research
OpenPOWER/POWER9 AI webinar
OpenPOWER/POWER9 AI webinar
WML OpenPOWER presentation
WML OpenPOWER presentation
IBM HPC Transformation with AI
IBM HPC Transformation with AI
Ac922 cdac webinar
Ac922 cdac webinar
OpenPOWER/POWER9 Webinar from MIT and IBM
OpenPOWER/POWER9 Webinar from MIT and IBM
PowerAI Deep dive
PowerAI Deep dive
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
SCFE 2020 OpenCAPI presentation as part of OpenPWOER Tutorial
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
OpenCAPI-based Image Analysis Pipeline for 18 GB/s kilohertz-framerate X-ray ...
Deeplearningusingcloudpakfordata
Deeplearningusingcloudpakfordata
Summit workshop thompto
Summit workshop thompto
IBM BOA for POWER
IBM BOA for POWER
JMI Techtalk: 한재근 - How to use GPU for developing AI
JMI Techtalk: 한재근 - How to use GPU for developing AI
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Xilinx Edge Compute using Power 9 /OpenPOWER systems
Large Model support and Distribute deep learning
Large Model support and Distribute deep learning
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Backend.AI Technical Introduction (19.09 / 2019 Autumn)
Transparent Hardware Acceleration for Deep Learning
Transparent Hardware Acceleration for Deep Learning
A Primer on FPGAs - Field Programmable Gate Arrays
A Primer on FPGAs - Field Programmable Gate Arrays
AMD It's Time to ROC
AMD It's Time to ROC
Similaire à BSC LMS DDL
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
DataWorks Summit/Hadoop Summit
IBM Cloud Paris Meetup - 20190520 - IA & Power
IBM Cloud Paris Meetup - 20190520 - IA & Power
IBM France Lab
OpenPOWER Boot camp in Zurich
OpenPOWER Boot camp in Zurich
Ganesan Narayanasamy
Innovation with ai at scale on the edge vt sept 2019 v0
Innovation with ai at scale on the edge vt sept 2019 v0
Ganesan Narayanasamy
Enabling a hardware accelerated deep learning data science experience for Apa...
Enabling a hardware accelerated deep learning data science experience for Apa...
DataWorks Summit
Cloud nativecomputingtechnologysupportinghpc cognitiveworkflows
Cloud nativecomputingtechnologysupportinghpc cognitiveworkflows
Yong Feng
Tuning For Deep Learning Inference with Intel® Processor Graphics | SIGGRAPH ...
Tuning For Deep Learning Inference with Intel® Processor Graphics | SIGGRAPH ...
Intel® Software
Enabling a hardware accelerated deep learning data science experience for Apa...
Enabling a hardware accelerated deep learning data science experience for Apa...
Indrajit Poddar
Introduction to Machine Learning on IBM Power Systems
Introduction to Machine Learning on IBM Power Systems
David Spurway
WML SNAP ML
WML SNAP ML
Ganesan Narayanasamy
Ibm symp14 referentin_barbara koch_power_8 launch bk
Ibm symp14 referentin_barbara koch_power_8 launch bk
IBM Switzerland
Deep Learning Frameworks Using Spark on YARN by Vartika Singh
Deep Learning Frameworks Using Spark on YARN by Vartika Singh
Data Con LA
IBM Power Systems - enabling cloud solutions
IBM Power Systems - enabling cloud solutions
David Spurway
AI in Health Care using IBM Systems/OpenPOWER systems
AI in Health Care using IBM Systems/OpenPOWER systems
Ganesan Narayanasamy
AI in Healh Care using IBM POWER systems
AI in Healh Care using IBM POWER systems
Ganesan Narayanasamy
Open power ddl and lms
Open power ddl and lms
Ganesan Narayanasamy
[AWS Dev Day] 인공지능 / 기계 학습 | AWS 기반 기계 학습 자동화 및 최적화를 위한 실전 기법 - 남궁영환 AWS 솔루션...
[AWS Dev Day] 인공지능 / 기계 학습 | AWS 기반 기계 학습 자동화 및 최적화를 위한 실전 기법 - 남궁영환 AWS 솔루션...
Amazon Web Services Korea
Demystify OpenPOWER
Demystify OpenPOWER
Anand Haridass
2689 - Exploring IBM PureApplication System and IBM Workload Deployer Best Pr...
2689 - Exploring IBM PureApplication System and IBM Workload Deployer Best Pr...
Hendrik van Run
AI Scalability for the Next Decade
AI Scalability for the Next Decade
Paula Koziol
Similaire à BSC LMS DDL
(20)
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
How to Optimize Hortonworks Apache Spark ML Workloads on Modern Processors
IBM Cloud Paris Meetup - 20190520 - IA & Power
IBM Cloud Paris Meetup - 20190520 - IA & Power
OpenPOWER Boot camp in Zurich
OpenPOWER Boot camp in Zurich
Innovation with ai at scale on the edge vt sept 2019 v0
Innovation with ai at scale on the edge vt sept 2019 v0
Enabling a hardware accelerated deep learning data science experience for Apa...
Enabling a hardware accelerated deep learning data science experience for Apa...
Cloud nativecomputingtechnologysupportinghpc cognitiveworkflows
Cloud nativecomputingtechnologysupportinghpc cognitiveworkflows
Tuning For Deep Learning Inference with Intel® Processor Graphics | SIGGRAPH ...
Tuning For Deep Learning Inference with Intel® Processor Graphics | SIGGRAPH ...
Enabling a hardware accelerated deep learning data science experience for Apa...
Enabling a hardware accelerated deep learning data science experience for Apa...
Introduction to Machine Learning on IBM Power Systems
Introduction to Machine Learning on IBM Power Systems
WML SNAP ML
WML SNAP ML
Ibm symp14 referentin_barbara koch_power_8 launch bk
Ibm symp14 referentin_barbara koch_power_8 launch bk
Deep Learning Frameworks Using Spark on YARN by Vartika Singh
Deep Learning Frameworks Using Spark on YARN by Vartika Singh
IBM Power Systems - enabling cloud solutions
IBM Power Systems - enabling cloud solutions
AI in Health Care using IBM Systems/OpenPOWER systems
AI in Health Care using IBM Systems/OpenPOWER systems
AI in Healh Care using IBM POWER systems
AI in Healh Care using IBM POWER systems
Open power ddl and lms
Open power ddl and lms
[AWS Dev Day] 인공지능 / 기계 학습 | AWS 기반 기계 학습 자동화 및 최적화를 위한 실전 기법 - 남궁영환 AWS 솔루션...
[AWS Dev Day] 인공지능 / 기계 학습 | AWS 기반 기계 학습 자동화 및 최적화를 위한 실전 기법 - 남궁영환 AWS 솔루션...
Demystify OpenPOWER
Demystify OpenPOWER
2689 - Exploring IBM PureApplication System and IBM Workload Deployer Best Pr...
2689 - Exploring IBM PureApplication System and IBM Workload Deployer Best Pr...
AI Scalability for the Next Decade
AI Scalability for the Next Decade
Plus de Ganesan Narayanasamy
Chip Design Curriculum development Residency program
Chip Design Curriculum development Residency program
Ganesan Narayanasamy
Basics of Digital Design and Verilog
Basics of Digital Design and Verilog
Ganesan Narayanasamy
180 nm Tape out experience using Open POWER ISA
180 nm Tape out experience using Open POWER ISA
Ganesan Narayanasamy
Workload Transformation and Innovations in POWER Architecture
Workload Transformation and Innovations in POWER Architecture
Ganesan Narayanasamy
OpenPOWER Workshop at IIT Roorkee
OpenPOWER Workshop at IIT Roorkee
Ganesan Narayanasamy
Deep Learning Use Cases using OpenPOWER systems
Deep Learning Use Cases using OpenPOWER systems
Ganesan Narayanasamy
OpenPOWER System Marconi100
OpenPOWER System Marconi100
Ganesan Narayanasamy
OpenPOWER Latest Updates
OpenPOWER Latest Updates
Ganesan Narayanasamy
POWER10 innovations for HPC
POWER10 innovations for HPC
Ganesan Narayanasamy
AI in healthcare and Automobile Industry using OpenPOWER/IBM POWER9 systems
AI in healthcare and Automobile Industry using OpenPOWER/IBM POWER9 systems
Ganesan Narayanasamy
AI in healthcare - Use Cases
AI in healthcare - Use Cases
Ganesan Narayanasamy
Poster from NUS
Poster from NUS
Ganesan Narayanasamy
SAP HANA on POWER9 systems
SAP HANA on POWER9 systems
Ganesan Narayanasamy
Graphical Structure Learning accelerated with POWER9
Graphical Structure Learning accelerated with POWER9
Ganesan Narayanasamy
AI in the enterprise
AI in the enterprise
Ganesan Narayanasamy
Robustness in deep learning
Robustness in deep learning
Ganesan Narayanasamy
Perspectives of Frond end Design
Perspectives of Frond end Design
Ganesan Narayanasamy
A2O Core implementation on FPGA
A2O Core implementation on FPGA
Ganesan Narayanasamy
OpenPOWER Foundation Introduction
OpenPOWER Foundation Introduction
Ganesan Narayanasamy
Open Hardware and Future Computing
Open Hardware and Future Computing
Ganesan Narayanasamy
Plus de Ganesan Narayanasamy
(20)
Chip Design Curriculum development Residency program
Chip Design Curriculum development Residency program
Basics of Digital Design and Verilog
Basics of Digital Design and Verilog
180 nm Tape out experience using Open POWER ISA
180 nm Tape out experience using Open POWER ISA
Workload Transformation and Innovations in POWER Architecture
Workload Transformation and Innovations in POWER Architecture
OpenPOWER Workshop at IIT Roorkee
OpenPOWER Workshop at IIT Roorkee
Deep Learning Use Cases using OpenPOWER systems
Deep Learning Use Cases using OpenPOWER systems
OpenPOWER System Marconi100
OpenPOWER System Marconi100
OpenPOWER Latest Updates
OpenPOWER Latest Updates
POWER10 innovations for HPC
POWER10 innovations for HPC
AI in healthcare and Automobile Industry using OpenPOWER/IBM POWER9 systems
AI in healthcare and Automobile Industry using OpenPOWER/IBM POWER9 systems
AI in healthcare - Use Cases
AI in healthcare - Use Cases
Poster from NUS
Poster from NUS
SAP HANA on POWER9 systems
SAP HANA on POWER9 systems
Graphical Structure Learning accelerated with POWER9
Graphical Structure Learning accelerated with POWER9
AI in the enterprise
AI in the enterprise
Robustness in deep learning
Robustness in deep learning
Perspectives of Frond end Design
Perspectives of Frond end Design
A2O Core implementation on FPGA
A2O Core implementation on FPGA
OpenPOWER Foundation Introduction
OpenPOWER Foundation Introduction
Open Hardware and Future Computing
Open Hardware and Future Computing
Dernier
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
Results
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
Anna Loughnan Colquhoun
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
Delhi Call girls
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Drew Madelung
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
naman860154
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
apidays
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Service
giselly40
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
Michael W. Hawkins
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Roshan Dwivedi
Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024
The Digital Insurer
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
Delhi Call girls
Slack Application Development 101 Slides
Slack Application Development 101 Slides
praypatel2
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
Earley Information Science
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
gurkirankumar98700
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
debabhi2
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Miguel Araújo
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Neo4j
Developing An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of Brazil
V3cube
Dernier
(20)
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Service
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
Slack Application Development 101 Slides
Slack Application Development 101 Slides
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Developing An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of Brazil
BSC LMS DDL
1.
© 2018 IBM
Corporation IBM Cognitive Systems IBM OpenPOWER foundation Barcelona Supercomputing Center Power9 AI differentiators Ander Ochoa – ander.ochoa.gilo@ibm.com Cognitive Systems Technical Architect for SPGI https://es.linkedin.com/in/anderotxoa @AnderOtxoa IBM 2018
2.
© 2018 IBM
Corporation IBM Cognitive Systems 2 Agenda • Large Memory Support (LMS) • Distributed Deep Learning (DDL) • PowerAI Vision
3.
© 2018 IBM
Corporation IBM Cognitive Systems 3 Large Memory Support (LMS) Objective: Overcome GPU Memory Limitations in DL Training. Increase the Batch Size and/or increase the resolution of the features. LMS enables processing of high definition images, large models, and higher batch sizes that doesn’t fit in GPU memory today (Maximum GPU memory available in Nvidia P100 and V100 GPUs is 16/32GB). Available for - Caffe - TensorFlow - Chainer https://www.sysml.cc/doc/127.pdf GPU RAM System RAM NVLinkv2.0 2 TB 16/32 GB Accelerated by NVLink Dataset
4.
© 2018 IBM
Corporation IBM Cognitive Systems LMS advantages running in Power architecture • LMS DL workload in Power vs LMS DL workload in x86 • NVLink provides up to 380% (depends on every workload/dataset) • NO LMS DL workload in Power vs NO LMS DL workload in x86 • NVLink provides 30% advantage (depends on every workload/dataset) NVMe Storage GPU RAM System RAM NVMeover PCIeV4 @32-140GB/sN TB 2 TB 16/32 GB SAS Storage GPU RAM System RAM SASover PCIeV3 @12Gb/s PCIeV3 @16GB/s N TB 1 TB 16/32 GB GPU GPU NVLinkv2.0 @150GB/s HBM2 HBM2 Power9x86 DDR4 DDR4
5.
© 2018 IBM
Corporation IBM Cognitive Systems 5 AC922 System buses and components diagram 32 -140+GB/s 64GB/s
6.
© 2018 IBM
Corporation IBM Cognitive Systems https://www.linkedin.com/pulse/deep-learning-high-resolution-images-large-models-sumit-gupta/ This comparison is between a AC922 server with four NVIDIA Tesla V100 GPUs versus a server with two Intel Xeon 2640v4 CPUs with four NVIDIA Tesla V100 GPUs
7.
© 2018 IBM
Corporation IBM Cognitive Systems 7 LMS in Caffe $caffe train -solver solver.prototxt -bvlc -gpu 0,1,2,3 -lms 10000 -lms_frac 0.5 • -lms 10000. Any memory chunk allocation larger than 10000KB will be done in CPU memory, and fetched to GPU memory only when needed for computation. • -lms_frac 0.5. LMS doesn’t kick in until more than at least 50% of GPU memory is expected to be utilized. Note that configuring the “lms” and “lms_frac” values depends on the below factors: • Batch size used • Model used • Number of GPUs used • System memory available Arriving at an optimal configuration requires understanding of the above and experimentation based on that. A general guideline is that the optimal configuration should utilize GPU memory close to fullest.
8.
© 2018 IBM
Corporation IBM Cognitive Systems 8 Demo https://developer.ibm.com/linuxonpower/2017/09/22/realizing-value-large-model-support-lms-powerai-ibm-caffe/ ssh bsc18651@plogin2.bsc.es srun -N 1 --exclusive --gres="gpu:4" --exclusive --pty $SHELL #Set the cpu to performance mode lscpu; ppc64_cpu –smt; ppc64_cpu --smt=2 cpupower -c all frequency-set -g performance #check gpu status nvidia-smi ; nvidia-smi -i 0 –q; nvidia-smi -ac 877,1530 #Activate caffe cd /gpfs/scratch/bsc18/bsc18040/lms source /opt/DL/caffe/bin/caffe-activate #show solver cat solver.prototxt #check in model batchsize = 1 vi models/googlenet_big.prototxt caffe train -solver solver.prototxt -bvlc -gpu 0,1,2,3 #change model batchsize = 5 #will give error (out of memory) vi models/googlenet_big.prototxt caffe train -solver solver.prototxt -bvlc -gpu 0,1,2,3 #check in model batchsize = 5 vi models/googlenet_big.prototxt caffe train -solver solver.prototxt -bvlc -gpu 0,1,2,3 -lms 10000 -lms_frac 0.5
9.
© 2018 IBM
Corporation IBM Cognitive Systems 9 Agenda • Large Memory Support (LMS) • Distributed Deep Learning (DDL) • PowerAI Vision
10.
© 2018 IBM
Corporation IBM Cognitive Systems 10
11.
© 2018 IBM
Corporation IBM Cognitive Systems 11 Distributed Deep Learning Objective: Overcome the server boundaries of some DL frameworks. How: Scaling. Using “ddlrun” applied to Topology aware distributed frameworks. Our software does deep learning training fully synchronously with very low communication overhead. The overall goal of ddlrun is to improve the user experience DDL users. To this end the primary features of ddlrun are: • Error Checking/Configuration Verification • Automatic Rankfile generation • Automatic mpirun option handling Available for: • Tensorflow • IBM Caffe • Torch How to code Topology Aware Distributed Models: • https://netweblog.wordpress.com/2018/04/10/distributed-tensorflow-sample-code-and-how-it-works/ • https://arxiv.org/pdf/1704.04560.pdf https://www.sysml.cc/doc/127.pdf Good for: • Speed • Accuracy
12.
© 2018 IBM
Corporation IBM Cognitive Systems Distributed Deep Learning (DDL) for Training phase Using the Power of 100s of Servers August 8, 2017 16 Days Down to 7 Hours: Near Ideal Scaling to 256 GPUs and Beyond 1 System 64 Systems 16 Days 7 Hours ResNet-101, ImageNet-22K, Caffe with PowerAI DDL, Running on Minsky (S822Lc) Power System 58x Faster https://www.ibm.com/blogs/research/2017/08/distributed-deep-learning/
13.
© 2018 IBM
Corporation IBM Cognitive Systems Demo 13 ssh bsc18651@plogin2.bsc.es #Slurm login node # You should have a ~/data dir with the dataset downloaded or internet conection to download it #Edit and include the following line in ~/.bashrc export TMPDIR=/tmp/ # To pass all the variables, like activate ...., you may need to write a simple submission script: Run as “sbatch script.sh” [bsc18651@p9login2 ~]$ cat script.sh #!/bin/bash #SBATCH -J test #SBATCH -D . #SBATCH -o test_%j.out #SBATCH -e test_%j.err #SBATCH -N 2 #SBATCH --ntasks-per-node=4 #SBATCH --gres="gpu:4" #SBATCH --time=01:00:00 module purge module load anaconda2 powerAI source /opt/DL/ddl-tensorflow/bin/ddl-tensorflow-activate export TMPDIR="/tmp/" export DDL_OPTIONS="-mode b:4x2" NODE_LIST=$(scontrol show hostname $SLURM_JOB_NODELIST | tr 'n' ',') NODE_LIST=${NODE_LIST%?} cd examples/mnist ddlrun -n 8 -H $NODE_LIST python mnist-init.py --ddl_options="-mode b:4x2" --data_dir /home/bsc18/bsc18651/examples/mnist/data [bsc18651@p9login2 ~]$ sbatch script.sh https://developer.ibm.com/linuxonpower/2018/05/01/improved-ease-use-ddl-powerai/
14.
© 2018 IBM
Corporation IBM Cognitive Systems 14 Agenda • Large Memory Support (LMS) • Distributed Deep Learning (DDL) • PowerAI Vision
15.
© 2018 IBM
Corporation IBM Cognitive Systems PowerAI Vision v1.1 15 https://www-01.ibm.com/common/ssi/cgi- bin/ssialias?infotype=an&subtype=ca&appname=gpateam&supplier=877&letternum=ENUSZP18-0143 PowerAI Vision V1.1 can help provide robust end-to-end workflow support for deep learning models related to computer vision. This enterprise-grade software provides a complete ecosystem to label raw data sets for training, creating, and deploying deep learning-based models. PowerAI Vision is designed to empower subject matter experts with no skills in deep learning technologies to train models for AI applications. It can help train highly accurate models to classify images and detect objects in images and videos. PowerAI Vision is built on open source frameworks • User interface-driven interaction to configure and manage lifecycles of data sets and models • A differentiated capability where trained deep learning models automatically detect objects from videos • Preconfigured deep learning models specialized to classify and detect objects • Preconfigured hyper-parameters optimized to classify and detect objects • Training visualization and runtime monitoring of accuracy • Integrated inference service to deploy models in production • Scalable architecture designed to run deep learning, high-performance analytics, and other long-running services and frameworks on shared resources
16.
© 2018 IBM
Corporation IBM Cognitive Systems PowerAI Vision demo 16 http://9.172.154.29:9080/powerai-vision/index.html
17.
© 2018 IBM
Corporation IBM Cognitive Systems Artificial Intelligence Power8 System Available NOW! ü To Facilitate and Expedite PoCs, Demos, Workshops… ü To offer to clients/BPs/partners so they can test it by themselves ü To show it to the world J Located in the IBM Client Center (Madrid) IBM “9” network Accesible from: (on demand) Internet Tech specs: - S822LC Power8 server (20 P8 cores) - 2x Nvidia P100 GPUs (7168 cuda cores) - 2x 500GB SDD + 1.6 TB NVMe Storage - 256 GB RAM - Ubuntu 16.04 - PowerAI v1.4 + VisionAI TP4
18.
© 2018 IBM
Corporation IBM Cognitive Systems IBM ICP + IBM DSX + IBM AI. Available NOW! ü To Facilitate and Expedite PoCs, Demos, Workshops… ü To offer to clients/BPs/partners so they can test it by themselves ü To show it to the world J Located in the IBM TEC (Madrid) IBM “9” network Accesible from: (for the time being) Tech specs: - S822L Power8 server (10 P8 cores) - 2 LPARs - 2x 500GB HDD - 512GB RAM - Ubuntu 16.04 - IBM Cloud Private - IBM Data Science Experience - S822L Power8 server (10 P8 cores) - 2x Nvidia P100 GPUs (7168 cuda cores) - 2x 500GB SDD + 1.6 TB NVMe Storage - 256 GB RAM - Ubuntu 16.04 - PowerAI tools ICP https://9.172.229.247:8443/console/ DSX https://9.172.229.247:31843
19.
© 2018 IBM
Corporation IBM Cognitive Systems THANK YOU!
20.
© 2018 IBM
Corporation IBM Cognitive Systems Notice and disclaimers ü Copyright © 2017 by International Business Machines Corporation (IBM). No part of this document may be reproduced or transmitted in any form without written permission from IBM. ü U.S. Government Users Restricted Rights — use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM. ü Information in these presentations (including information relating to products that have not yet been announced by IBM) has been reviewed for accuracy as of the date of initial publication and could include unintentional technical or typographical errors. IBM shall have no responsibility to update this information. This document is distributed “as is” without any warranty, either express or implied. In no event shall IBM be liable for any damage arising from the use of this information, including but not limited to, loss of data, business interruption, loss of profit or loss of opportunity. IBM products and services are warranted according to the terms and conditions of the agreements under which they are provided. ü IBM products are manufactured from new parts or new and used parts. In some cases, a product may not be new and may have been previously installed. Regardless, our warranty terms apply.” ü Any statements regarding IBM's future direction, intent or product plans are subject to change or withdrawal without notice. ü Performance data contained herein was generally obtained in a controlled, isolated environments. Customer examples are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual performance, cost, savings or other results in other operating environments may vary. ü References in this document to IBM products, programs, or services does not imply that IBM intends to make such products, programs or services available in all countries in which IBM operates or does business. ü Workshops, sessions and associated materials may have been prepared by independent session speakers, and do not necessarily reflect the views of IBM. All materials and discussions are provided for informational purposes only, and are neither intended to, nor shall constitute legal or other guidance or advice to any individual participant or their specific situation. ü It is the customer’s responsibility to insure its own compliance with legal requirements and to obtain advice of competent legal counsel as to the identification and interpretation of any relevant laws and regulatory requirements that may affect the customer’s business and any actions the customer may need to take to comply with such laws. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the customer is in compliance with any law.
21.
© 2018 IBM
Corporation IBM Cognitive Systems Notice and disclaimers continued Information concerning non-IBM products was obtained from the suppliers of those products, their published announcements or other publicly available sources. IBM has not tested those products in connection with this publication and cannot confirm the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM does not warrant the quality of any third-party products, or the ability of any such third-party products 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 particular, 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. IBM, the IBM logo, ibm.com, AIX, BigInsights, Bluemix, CICS, Easy Tier, FlashCopy, FlashSystem, GDPS, GPFS, Guardium, HyperSwap, IBM Cloud Managed Services, IBM Elastic Storage, IBM FlashCore, IBM FlashSystem, IBM MobileFirst, IBM Power Systems, IBM PureSystems, IBM Spectrum, IBM Spectrum Accelerate, IBM Spectrum Archive, IBM Spectrum Control, IBM Spectrum Protect, IBM Spectrum Scale, IBM Spectrum Storage, IBM Spectrum Virtualize, IBM Watson, IBM z Systems, IBM z13, IMS, InfoSphere, Linear Tape File System, OMEGAMON, OpenPower, Parallel Sysplex, Power, POWER, POWER4, POWER7, POWER8, Power Series, Power Systems, Power Systems Software, PowerHA, PowerLinux, PowerVM, PureApplica- tion, RACF, Real-time Compression, Redbooks, RMF, SPSS, Storwize, Symphony, SystemMirror, System Storage, Tivoli, WebSphere, XIV, z Systems, z/OS, z/VM, z/VSE, zEnterprise and zSecure are trademarks of International Business Machines Corporation, registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at: www.ibm.com/legal/copytrade.shtml. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates.
Télécharger maintenant