SlideShare ist ein Scribd-Unternehmen logo
1 von 17
Downloaden Sie, um offline zu lesen
1
Graph Hardware Architecture
Enterprise graphs deserve great hardware!
Dan McCreary, Distinguished Engineer, Optum
Nikhil Deshpande, Director, AI & HPC, Intel Corp.
Graph + AI World 2020
September 29, 2020
About Dan
2
• Distinguished Engineer at Optum Healthcare
(330K employees and 32K technical staff)
• Focused on AI enterprise knowledge graphs
• Help create the worlds largest healthcare
graph
• Coauthor of book "Making Sense of NoSQL”
• Worked at Bell Labs as a VLSI circuit designer
• Worked for Steve Jobs at NeXT
Clinical Data Fits Well With Graph Database Models
3
Clinical
Data
Graph
• There is a natural fit between clinical data and graph technologies
• Clinical data is both complex and shows high-variability
• Graphs are ideal for storing complex and highly-variable data
Business Impact of Moving from RDBMS to Graph
4
Relational Graph Business Impact of Using Graph for
Clinical Data
Relationships calculated at
query time
Relationships calculated at
load time
Relationship-intensive clinical data is
queried faster
Fixed data model Flexible data model It is easier to add new clinical data to a
graph at any time
Limited analytical algorithms Rich library of analytical
algorithms such as pathfinding,
PageRank, clustering, random
walk and similarity
It is easy to execute patient clustering,
similarity and care path recommendation
engines
Difficult to build custom
hardware to optimize
algorithms
Easy to build custom hardware
to optimize algorithms
Hardware for optimizing algorithms is
low cost
The Real Time Clinical Decision Support Challenge
5
• A new patient walks into the ER, urgent care or our clinic
• We gather patient data into the Electronic Medical Record (EMR)
• How quickly can we compare this patient with the best outcomes of 250
million other patients? Goal: 200 msec
250M Patients
Patient Data
Recommended
Care Path
How We Got Here: Pointer Hopping
• Several years ago we started working on graph scaling
problems
• Graph traversal is they key function that must scale in
parallel
• Graph traversal is all about pointer hopping
• Having a VLSI circuit design background help me
visualize how narrow the instruction sets were needed to
do this quickly
• I realized that both the instruction set and the memory
access patterns were not optimized with modern
general purpose CISC hardware
• We needed a new distributed graph architecture with
many cores and faster memory access patterns
6
Example: Learning Embeddings Via Random Walk
• NLP technologies have shown us that we can
create low-dimensional embeddings for every
word by looking at the sequence of words in text.
• Using Random Walk algorithms we can create a
similar sequence of items to create embeddings
for primary Vertices (Customer, Product etc.)
• We can use these embeddings for 50 msec
comparisons over 100M items if we do the
calculations in parallel
• Random Walks are very CPU intensive and
require traversal of hundreds of millions of
edges
7
Embedding: 200 32-bit integer
8Intel Confidential 8
RoleofHardware...
q Graph based analytics is the next frontier
q Architectures need to evolve to support next-gen graph
usages in healthcare
q Intel® Programmable Integrated Unified Memory
Architecture (PIUMA)
q ~1,000x Speedup on graphs!
Copyright © 2020, Intel Corporation
9Intel Confidential 9
Regular versus GraphApplications
Branch
Prediction
Locality
Data Access
Regular Graph
Branches have regular
pattern
Same or neighboring
data likely used
Same operation on
neighboring data
Branch outcome is
data dependent
(“Pointer hopping”)
Data is randomly
scattered
Operations on
scattered data
Copyright © 2020, Intel Corporation
10Intel Confidential 10
TraditionalComputeBottlenecksforGraph
AdvancedNIC
AdvancedNIC
Core level: Cache
Socket level: DRAM
limitations
Network Level: high
latency/SW layers
AdvancedNIC
AdvancedNIC
Rack Network
Infrastructure Level: designed
for big messages
Copyright © 2020, Intel Corporation
11Intel Confidential 11
Predictive Analytics / Graph Characteristics
Data Size
Communications
Intensity
Search, Hadoop
Web services,
IT, CRM, E-
commerce,
Gaming,
Security,
SaaS, IaaS
Predictive
Analytics in
Graphs
Scalability
Wall
Dense Matrix Multiplication
Random walk
Sub-graph search
Page Rank
The scalability wall is
the limit to
performance scaling
Copyright © 2020, Intel Corporation
12Intel Confidential 12
ImpactofHitting Scalability Wall
See backup for configuration details. For more complete information about performance and benchmark results,
visit www.intel.com/benchmarks.
GTEPS: Giga Traversed Edges
Per Second
GTEPS(measureaffectedbycomms.intensity)
Random Walks – OpenMP
Input Data Size (GB)
0
20
40
60
80
100
120
140
160
16 80 125 226 318 429 459 744
GFLOPS
Input Data Size (MB)
Page Rank / SpMV
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 4 8 12 24
28
Copyright © 2020, Intel Corporation
13Intel Confidential 13
Intel®PIUMA Design Principles
• New architecture to address
scalability wall
• Balanced I/O, memory and
compute
• General programmable
accelerator
• Real time query on large
graph datasets constantly
being updated
Intel® PIUMA
Compute Memory I/O
Traditional
Optim
ized
for
TOPS
Optim
ized
for
TOPS
and
TTEPS
TOPS: Tera-Operations Per Second
TTEPS: Tera-Traversed Edges Per Second
Copyright © 2020, Intel Corporation
14Intel Confidential 14
Intel®PIUMA core
Cores are designed for throughput or single-thread
performance
Novel instruction set with native 64-bit physical
addresses and graph operation optimizations
Distributed global address space (GAS) for all memory
visible by all cores
Memory and fabric subsystems optimized for both 8B
and 64B transfers
Copyright © 2020, Intel Corporation
15Intel Confidential 15
Intel® PIUMA System
Intel® PIUMA
HW
Runtime/Tools
C/C++/Python
Graph
Scalable SW and
Hardware
• Extremely high bandwidth, low energy and latency
direct fabric (TB/s Vs GB/s)
• Optimized for Random Accesses to PB of memory
• Balanced Compute to I/O Ratio
C
M
M
M
C
M
C
M
M
C
M
C
M
M
C
Intel® PIUMA
M
Multi-threaded
compute
Memory optimized
for TEPs
C CC
C
Compute Nodes
Today’s Architecture
• Limited I/O BW & High Latency Networks
• Low bandwidth to Memory
• High Compute to I/O Ratio (Unbalanced)
C M
C
M
M
M
C
M
C
M
C
M
C
M
C
M
C
M
C
C
Traditional Compute Memory optimized
for FLOPs
Compute Nodes
Traditional Switch
Copyright © 2020, Intel Corporation
16Intel Confidential 16
Intel®PIUMA:SpeedUpGraphTraversalby>1,000x
Performance
Graph Search (Random Walks)
>1,000x
No. of Communicating Systems
See backup for configuration details. For more complete information about performance and benchmark results,
visit www.intel.com/benchmarks.
Intel® PIUMA
System and Fabric
Traditional Fabric
Linear Scaling
Copyright © 2020, Intel Corporation
Accelerated
Enterprise
Knowledge Graph
Queries
Conclusion: Think BIG about your EKG!
• We encourage everyone to continue to
“think big” when doing strategic planning
around your Enterprise Knowledge Graphs
• Many organizations besides Google,
Amazon, Facebook etc. will be building 100B
to 1TB vertex graphs in the next few years to
better serve their customers
• New hardware will soon be arriving that will
dramatically accelerate today’s graph
queries
• Real time product recommendation systems
that take complex content into account will
be cost effective a accelerate data-driven
decision making in healthcare and other
industries
17

Weitere ähnliche Inhalte

Was ist angesagt?

RAPIDS cuGraph – Accelerating all your Graph needs
RAPIDS cuGraph – Accelerating all your Graph needsRAPIDS cuGraph – Accelerating all your Graph needs
RAPIDS cuGraph – Accelerating all your Graph needs
Connected Data World
 

Was ist angesagt? (20)

Using Graph Algorithms For Advanced Analytics - Part 4 Similarity 30 graph al...
Using Graph Algorithms For Advanced Analytics - Part 4 Similarity 30 graph al...Using Graph Algorithms For Advanced Analytics - Part 4 Similarity 30 graph al...
Using Graph Algorithms For Advanced Analytics - Part 4 Similarity 30 graph al...
 
Graph Databases and Machine Learning | November 2018
Graph Databases and Machine Learning | November 2018Graph Databases and Machine Learning | November 2018
Graph Databases and Machine Learning | November 2018
 
Fraud prevention is better with TigerGraph inside
Fraud prevention is better with  TigerGraph insideFraud prevention is better with  TigerGraph inside
Fraud prevention is better with TigerGraph inside
 
Machine Learning Feature Design with TigerGraph 3.0 No-Code GUI
Machine Learning Feature Design with TigerGraph 3.0 No-Code GUIMachine Learning Feature Design with TigerGraph 3.0 No-Code GUI
Machine Learning Feature Design with TigerGraph 3.0 No-Code GUI
 
Graph Gurus 21: Integrating Real-Time Deep-Link Graph Analytics with Spark AI
Graph Gurus 21: Integrating Real-Time Deep-Link Graph Analytics with Spark AIGraph Gurus 21: Integrating Real-Time Deep-Link Graph Analytics with Spark AI
Graph Gurus 21: Integrating Real-Time Deep-Link Graph Analytics with Spark AI
 
How to Build An AI Based Customer Data Platform: Learn the design patterns fo...
How to Build An AI Based Customer Data Platform: Learn the design patterns fo...How to Build An AI Based Customer Data Platform: Learn the design patterns fo...
How to Build An AI Based Customer Data Platform: Learn the design patterns fo...
 
Graph + AI World Opening Keynote
Graph + AI World Opening KeynoteGraph + AI World Opening Keynote
Graph + AI World Opening Keynote
 
Graph + AI World 2020: Opening Day Keynote
Graph + AI World 2020: Opening Day KeynoteGraph + AI World 2020: Opening Day Keynote
Graph + AI World 2020: Opening Day Keynote
 
Knowledge Graph for Machine Learning and Data Science
Knowledge Graph for Machine Learning and Data ScienceKnowledge Graph for Machine Learning and Data Science
Knowledge Graph for Machine Learning and Data Science
 
Graph Gurus Episode 1: Enterprise Graph
Graph Gurus Episode 1: Enterprise GraphGraph Gurus Episode 1: Enterprise Graph
Graph Gurus Episode 1: Enterprise Graph
 
Deep Learning for Recommender Systems with Nick pentreath
Deep Learning for Recommender Systems with Nick pentreathDeep Learning for Recommender Systems with Nick pentreath
Deep Learning for Recommender Systems with Nick pentreath
 
Graph Gurus Episode 17: Seven Key Data Science Capabilities Powered by a Nati...
Graph Gurus Episode 17: Seven Key Data Science Capabilities Powered by a Nati...Graph Gurus Episode 17: Seven Key Data Science Capabilities Powered by a Nati...
Graph Gurus Episode 17: Seven Key Data Science Capabilities Powered by a Nati...
 
Using Graph Algorithms for Advanced Analytics - Part 2 Centrality
Using Graph Algorithms for Advanced Analytics - Part 2 CentralityUsing Graph Algorithms for Advanced Analytics - Part 2 Centrality
Using Graph Algorithms for Advanced Analytics - Part 2 Centrality
 
Supply Chain and Logistics Management with Graph & AI
Supply Chain and Logistics Management with Graph & AISupply Chain and Logistics Management with Graph & AI
Supply Chain and Logistics Management with Graph & AI
 
Graph Gurus Episode 27: Using Graph Algorithms for Advanced Analytics Part 2
Graph Gurus Episode 27: Using Graph Algorithms for Advanced Analytics Part 2Graph Gurus Episode 27: Using Graph Algorithms for Advanced Analytics Part 2
Graph Gurus Episode 27: Using Graph Algorithms for Advanced Analytics Part 2
 
Hybrid Transactional/Analytics Processing: Beyond the Big Database Hype
Hybrid Transactional/Analytics Processing: Beyond the Big Database HypeHybrid Transactional/Analytics Processing: Beyond the Big Database Hype
Hybrid Transactional/Analytics Processing: Beyond the Big Database Hype
 
Spark DC Interactive Meetup: HTAP with Spark and In-Memory Data Grids
Spark DC Interactive Meetup: HTAP with Spark and In-Memory Data GridsSpark DC Interactive Meetup: HTAP with Spark and In-Memory Data Grids
Spark DC Interactive Meetup: HTAP with Spark and In-Memory Data Grids
 
Real-time Microservices and In-Memory Data Grids
Real-time Microservices and In-Memory Data GridsReal-time Microservices and In-Memory Data Grids
Real-time Microservices and In-Memory Data Grids
 
Counting Unique Users in Real-Time: Here's a Challenge for You!
Counting Unique Users in Real-Time: Here's a Challenge for You!Counting Unique Users in Real-Time: Here's a Challenge for You!
Counting Unique Users in Real-Time: Here's a Challenge for You!
 
RAPIDS cuGraph – Accelerating all your Graph needs
RAPIDS cuGraph – Accelerating all your Graph needsRAPIDS cuGraph – Accelerating all your Graph needs
RAPIDS cuGraph – Accelerating all your Graph needs
 

Ähnlich wie Graph Hardware Architecture - Enterprise graphs deserve great hardware!

Kognitio overview jan 2013
Kognitio overview jan 2013Kognitio overview jan 2013
Kognitio overview jan 2013
Kognitio
 

Ähnlich wie Graph Hardware Architecture - Enterprise graphs deserve great hardware! (20)

IBM Data Centric Systems & OpenPOWER
IBM Data Centric Systems & OpenPOWERIBM Data Centric Systems & OpenPOWER
IBM Data Centric Systems & OpenPOWER
 
Webinar: The Modern Streaming Data Stack with Kinetica & StreamSets
Webinar: The Modern Streaming Data Stack with Kinetica & StreamSetsWebinar: The Modern Streaming Data Stack with Kinetica & StreamSets
Webinar: The Modern Streaming Data Stack with Kinetica & StreamSets
 
Lecture_IIITD.pptx
Lecture_IIITD.pptxLecture_IIITD.pptx
Lecture_IIITD.pptx
 
Real time machine learning proposers day v3
Real time machine learning proposers day v3Real time machine learning proposers day v3
Real time machine learning proposers day v3
 
FPGA Hardware Accelerator for Machine Learning
FPGA Hardware Accelerator for Machine Learning FPGA Hardware Accelerator for Machine Learning
FPGA Hardware Accelerator for Machine Learning
 
Kognitio overview jan 2013
Kognitio overview jan 2013Kognitio overview jan 2013
Kognitio overview jan 2013
 
Kognitio overview jan 2013
Kognitio overview jan 2013Kognitio overview jan 2013
Kognitio overview jan 2013
 
CC LECTURE NOTES (1).pdf
CC LECTURE NOTES (1).pdfCC LECTURE NOTES (1).pdf
CC LECTURE NOTES (1).pdf
 
42 grid computing
42 grid computing42 grid computing
42 grid computing
 
42 grid computing
42 grid computing42 grid computing
42 grid computing
 
AWS Summit Berlin 2013 - Big Data Analytics
AWS Summit Berlin 2013 - Big Data AnalyticsAWS Summit Berlin 2013 - Big Data Analytics
AWS Summit Berlin 2013 - Big Data Analytics
 
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc..."An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
 
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc..."An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
"An introduction to Kx Technology - a Big Data solution", Kyra Coyne, Data Sc...
 
EDGE devices_ ERTOS_ IOT_ presentation_P.pptx
EDGE devices_ ERTOS_ IOT_ presentation_P.pptxEDGE devices_ ERTOS_ IOT_ presentation_P.pptx
EDGE devices_ ERTOS_ IOT_ presentation_P.pptx
 
IoT meets Big Data
IoT meets Big DataIoT meets Big Data
IoT meets Big Data
 
Pro Tips: Designing and Deploying End-to-End HPC and AI Solutions
Pro Tips: Designing and Deploying End-to-End HPC and AI SolutionsPro Tips: Designing and Deploying End-to-End HPC and AI Solutions
Pro Tips: Designing and Deploying End-to-End HPC and AI Solutions
 
Penguin computing designing and deploying end to end HPC and AI Solutions
Penguin computing designing and deploying end to end HPC and AI SolutionsPenguin computing designing and deploying end to end HPC and AI Solutions
Penguin computing designing and deploying end to end HPC and AI Solutions
 
Penguin Computing Designing and Deploying End to End HPC and AI Solutions
Penguin Computing Designing and Deploying End to End HPC and AI SolutionsPenguin Computing Designing and Deploying End to End HPC and AI Solutions
Penguin Computing Designing and Deploying End to End HPC and AI Solutions
 
High Performance Computing and Big Data: The coming wave
High Performance Computing and Big Data: The coming waveHigh Performance Computing and Big Data: The coming wave
High Performance Computing and Big Data: The coming wave
 
FC Brochure & Insert
FC Brochure & InsertFC Brochure & Insert
FC Brochure & Insert
 

Mehr von TigerGraph

Mehr von TigerGraph (19)

MAXIMIZING THE VALUE OF SCIENTIFIC INFORMATION TO ACCELERATE INNOVATION
MAXIMIZING THE VALUE OF SCIENTIFIC INFORMATION TO ACCELERATE INNOVATIONMAXIMIZING THE VALUE OF SCIENTIFIC INFORMATION TO ACCELERATE INNOVATION
MAXIMIZING THE VALUE OF SCIENTIFIC INFORMATION TO ACCELERATE INNOVATION
 
Building an accurate understanding of consumers based on real-world signals
Building an accurate understanding of consumers based on real-world signalsBuilding an accurate understanding of consumers based on real-world signals
Building an accurate understanding of consumers based on real-world signals
 
Care Intervention Assistant - Omaha Clinical Data Information System
Care Intervention Assistant - Omaha Clinical Data Information SystemCare Intervention Assistant - Omaha Clinical Data Information System
Care Intervention Assistant - Omaha Clinical Data Information System
 
Correspondent Banking Networks
Correspondent Banking NetworksCorrespondent Banking Networks
Correspondent Banking Networks
 
Delivering Large Scale Real-time Graph Analytics with Dell Infrastructure and...
Delivering Large Scale Real-time Graph Analytics with Dell Infrastructure and...Delivering Large Scale Real-time Graph Analytics with Dell Infrastructure and...
Delivering Large Scale Real-time Graph Analytics with Dell Infrastructure and...
 
Deploying an End-to-End TigerGraph Enterprise Architecture using Kafka, Maria...
Deploying an End-to-End TigerGraph Enterprise Architecture using Kafka, Maria...Deploying an End-to-End TigerGraph Enterprise Architecture using Kafka, Maria...
Deploying an End-to-End TigerGraph Enterprise Architecture using Kafka, Maria...
 
Fraud Detection and Compliance with Graph Learning
Fraud Detection and Compliance with Graph LearningFraud Detection and Compliance with Graph Learning
Fraud Detection and Compliance with Graph Learning
 
Fraudulent credit card cash-out detection On Graphs
Fraudulent credit card cash-out detection On GraphsFraudulent credit card cash-out detection On Graphs
Fraudulent credit card cash-out detection On Graphs
 
FROM DATAFRAMES TO GRAPH Data Science with pyTigerGraph
FROM DATAFRAMES TO GRAPH Data Science with pyTigerGraphFROM DATAFRAMES TO GRAPH Data Science with pyTigerGraph
FROM DATAFRAMES TO GRAPH Data Science with pyTigerGraph
 
Customer Experience Management
Customer Experience ManagementCustomer Experience Management
Customer Experience Management
 
Davraz - A graph visualization and exploration software.
Davraz - A graph visualization and exploration software.Davraz - A graph visualization and exploration software.
Davraz - A graph visualization and exploration software.
 
Plume - A Code Property Graph Extraction and Analysis Library
Plume - A Code Property Graph Extraction and Analysis LibraryPlume - A Code Property Graph Extraction and Analysis Library
Plume - A Code Property Graph Extraction and Analysis Library
 
TigerGraph.js
TigerGraph.jsTigerGraph.js
TigerGraph.js
 
GRAPHS FOR THE FUTURE ENERGY SYSTEMS
GRAPHS FOR THE FUTURE ENERGY SYSTEMSGRAPHS FOR THE FUTURE ENERGY SYSTEMS
GRAPHS FOR THE FUTURE ENERGY SYSTEMS
 
Recommendation Engine with In-Database Machine Learning
Recommendation Engine with In-Database Machine LearningRecommendation Engine with In-Database Machine Learning
Recommendation Engine with In-Database Machine Learning
 
The key to creating a Golden Thread: the power of Graph Databases for Entity ...
The key to creating a Golden Thread: the power of Graph Databases for Entity ...The key to creating a Golden Thread: the power of Graph Databases for Entity ...
The key to creating a Golden Thread: the power of Graph Databases for Entity ...
 
Training Graph Convolutional Neural Networks in Graph Database
Training Graph Convolutional Neural Networks in Graph DatabaseTraining Graph Convolutional Neural Networks in Graph Database
Training Graph Convolutional Neural Networks in Graph Database
 
Deep Link Analytics Empowered by AI + Graph + Verticals
Deep Link Analytics Empowered by AI + Graph + VerticalsDeep Link Analytics Empowered by AI + Graph + Verticals
Deep Link Analytics Empowered by AI + Graph + Verticals
 
Jaguar Land Rover: A supply chain case study
Jaguar Land Rover: A supply chain case studyJaguar Land Rover: A supply chain case study
Jaguar Land Rover: A supply chain case study
 

Kürzlich hochgeladen

Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
amitlee9823
 
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
amitlee9823
 
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
amitlee9823
 
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men 🔝mahisagar🔝 Esc...
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men  🔝mahisagar🔝   Esc...➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men  🔝mahisagar🔝   Esc...
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men 🔝mahisagar🔝 Esc...
amitlee9823
 
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
only4webmaster01
 
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICECHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
amitlee9823
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Riyadh +966572737505 get cytotec
 
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night StandCall Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night StandCall Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
amitlee9823
 
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
gajnagarg
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
amitlee9823
 
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night StandCall Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
amitlee9823
 
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men 🔝malwa🔝 Escorts Ser...
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men  🔝malwa🔝   Escorts Ser...➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men  🔝malwa🔝   Escorts Ser...
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men 🔝malwa🔝 Escorts Ser...
amitlee9823
 

Kürzlich hochgeladen (20)

Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
 
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
Vip Mumbai Call Girls Marol Naka Call On 9920725232 With Body to body massage...
 
Aspirational Block Program Block Syaldey District - Almora
Aspirational Block Program Block Syaldey District - AlmoraAspirational Block Program Block Syaldey District - Almora
Aspirational Block Program Block Syaldey District - Almora
 
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
Chintamani Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore ...
 
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men 🔝mahisagar🔝 Esc...
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men  🔝mahisagar🔝   Esc...➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men  🔝mahisagar🔝   Esc...
➥🔝 7737669865 🔝▻ mahisagar Call-girls in Women Seeking Men 🔝mahisagar🔝 Esc...
 
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 9155563397 👗 Top Class Call Girl Service B...
 
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICECHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
CHEAP Call Girls in Saket (-DELHI )🔝 9953056974🔝(=)/CALL GIRLS SERVICE
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
 
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
 
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night StandCall Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Attibele ☎ 7737669865 🥵 Book Your One night Stand
 
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night StandCall Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Doddaballapur Road ☎ 7737669865 🥵 Book Your One night Stand
 
DATA SUMMIT 24 Building Real-Time Pipelines With FLaNK
DATA SUMMIT 24  Building Real-Time Pipelines With FLaNKDATA SUMMIT 24  Building Real-Time Pipelines With FLaNK
DATA SUMMIT 24 Building Real-Time Pipelines With FLaNK
 
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
 
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
VIP Model Call Girls Hinjewadi ( Pune ) Call ON 8005736733 Starting From 5K t...
 
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
Just Call Vip call girls Palakkad Escorts ☎️9352988975 Two shot with one girl...
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
 
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night StandCall Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
Call Girls In Nandini Layout ☎ 7737669865 🥵 Book Your One night Stand
 
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men 🔝malwa🔝 Escorts Ser...
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men  🔝malwa🔝   Escorts Ser...➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men  🔝malwa🔝   Escorts Ser...
➥🔝 7737669865 🔝▻ malwa Call-girls in Women Seeking Men 🔝malwa🔝 Escorts Ser...
 
Discover Why Less is More in B2B Research
Discover Why Less is More in B2B ResearchDiscover Why Less is More in B2B Research
Discover Why Less is More in B2B Research
 

Graph Hardware Architecture - Enterprise graphs deserve great hardware!

  • 1. 1 Graph Hardware Architecture Enterprise graphs deserve great hardware! Dan McCreary, Distinguished Engineer, Optum Nikhil Deshpande, Director, AI & HPC, Intel Corp. Graph + AI World 2020 September 29, 2020
  • 2. About Dan 2 • Distinguished Engineer at Optum Healthcare (330K employees and 32K technical staff) • Focused on AI enterprise knowledge graphs • Help create the worlds largest healthcare graph • Coauthor of book "Making Sense of NoSQL” • Worked at Bell Labs as a VLSI circuit designer • Worked for Steve Jobs at NeXT
  • 3. Clinical Data Fits Well With Graph Database Models 3 Clinical Data Graph • There is a natural fit between clinical data and graph technologies • Clinical data is both complex and shows high-variability • Graphs are ideal for storing complex and highly-variable data
  • 4. Business Impact of Moving from RDBMS to Graph 4 Relational Graph Business Impact of Using Graph for Clinical Data Relationships calculated at query time Relationships calculated at load time Relationship-intensive clinical data is queried faster Fixed data model Flexible data model It is easier to add new clinical data to a graph at any time Limited analytical algorithms Rich library of analytical algorithms such as pathfinding, PageRank, clustering, random walk and similarity It is easy to execute patient clustering, similarity and care path recommendation engines Difficult to build custom hardware to optimize algorithms Easy to build custom hardware to optimize algorithms Hardware for optimizing algorithms is low cost
  • 5. The Real Time Clinical Decision Support Challenge 5 • A new patient walks into the ER, urgent care or our clinic • We gather patient data into the Electronic Medical Record (EMR) • How quickly can we compare this patient with the best outcomes of 250 million other patients? Goal: 200 msec 250M Patients Patient Data Recommended Care Path
  • 6. How We Got Here: Pointer Hopping • Several years ago we started working on graph scaling problems • Graph traversal is they key function that must scale in parallel • Graph traversal is all about pointer hopping • Having a VLSI circuit design background help me visualize how narrow the instruction sets were needed to do this quickly • I realized that both the instruction set and the memory access patterns were not optimized with modern general purpose CISC hardware • We needed a new distributed graph architecture with many cores and faster memory access patterns 6
  • 7. Example: Learning Embeddings Via Random Walk • NLP technologies have shown us that we can create low-dimensional embeddings for every word by looking at the sequence of words in text. • Using Random Walk algorithms we can create a similar sequence of items to create embeddings for primary Vertices (Customer, Product etc.) • We can use these embeddings for 50 msec comparisons over 100M items if we do the calculations in parallel • Random Walks are very CPU intensive and require traversal of hundreds of millions of edges 7 Embedding: 200 32-bit integer
  • 8. 8Intel Confidential 8 RoleofHardware... q Graph based analytics is the next frontier q Architectures need to evolve to support next-gen graph usages in healthcare q Intel® Programmable Integrated Unified Memory Architecture (PIUMA) q ~1,000x Speedup on graphs! Copyright © 2020, Intel Corporation
  • 9. 9Intel Confidential 9 Regular versus GraphApplications Branch Prediction Locality Data Access Regular Graph Branches have regular pattern Same or neighboring data likely used Same operation on neighboring data Branch outcome is data dependent (“Pointer hopping”) Data is randomly scattered Operations on scattered data Copyright © 2020, Intel Corporation
  • 10. 10Intel Confidential 10 TraditionalComputeBottlenecksforGraph AdvancedNIC AdvancedNIC Core level: Cache Socket level: DRAM limitations Network Level: high latency/SW layers AdvancedNIC AdvancedNIC Rack Network Infrastructure Level: designed for big messages Copyright © 2020, Intel Corporation
  • 11. 11Intel Confidential 11 Predictive Analytics / Graph Characteristics Data Size Communications Intensity Search, Hadoop Web services, IT, CRM, E- commerce, Gaming, Security, SaaS, IaaS Predictive Analytics in Graphs Scalability Wall Dense Matrix Multiplication Random walk Sub-graph search Page Rank The scalability wall is the limit to performance scaling Copyright © 2020, Intel Corporation
  • 12. 12Intel Confidential 12 ImpactofHitting Scalability Wall See backup for configuration details. For more complete information about performance and benchmark results, visit www.intel.com/benchmarks. GTEPS: Giga Traversed Edges Per Second GTEPS(measureaffectedbycomms.intensity) Random Walks – OpenMP Input Data Size (GB) 0 20 40 60 80 100 120 140 160 16 80 125 226 318 429 459 744 GFLOPS Input Data Size (MB) Page Rank / SpMV 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 4 8 12 24 28 Copyright © 2020, Intel Corporation
  • 13. 13Intel Confidential 13 Intel®PIUMA Design Principles • New architecture to address scalability wall • Balanced I/O, memory and compute • General programmable accelerator • Real time query on large graph datasets constantly being updated Intel® PIUMA Compute Memory I/O Traditional Optim ized for TOPS Optim ized for TOPS and TTEPS TOPS: Tera-Operations Per Second TTEPS: Tera-Traversed Edges Per Second Copyright © 2020, Intel Corporation
  • 14. 14Intel Confidential 14 Intel®PIUMA core Cores are designed for throughput or single-thread performance Novel instruction set with native 64-bit physical addresses and graph operation optimizations Distributed global address space (GAS) for all memory visible by all cores Memory and fabric subsystems optimized for both 8B and 64B transfers Copyright © 2020, Intel Corporation
  • 15. 15Intel Confidential 15 Intel® PIUMA System Intel® PIUMA HW Runtime/Tools C/C++/Python Graph Scalable SW and Hardware • Extremely high bandwidth, low energy and latency direct fabric (TB/s Vs GB/s) • Optimized for Random Accesses to PB of memory • Balanced Compute to I/O Ratio C M M M C M C M M C M C M M C Intel® PIUMA M Multi-threaded compute Memory optimized for TEPs C CC C Compute Nodes Today’s Architecture • Limited I/O BW & High Latency Networks • Low bandwidth to Memory • High Compute to I/O Ratio (Unbalanced) C M C M M M C M C M C M C M C M C M C C Traditional Compute Memory optimized for FLOPs Compute Nodes Traditional Switch Copyright © 2020, Intel Corporation
  • 16. 16Intel Confidential 16 Intel®PIUMA:SpeedUpGraphTraversalby>1,000x Performance Graph Search (Random Walks) >1,000x No. of Communicating Systems See backup for configuration details. For more complete information about performance and benchmark results, visit www.intel.com/benchmarks. Intel® PIUMA System and Fabric Traditional Fabric Linear Scaling Copyright © 2020, Intel Corporation Accelerated Enterprise Knowledge Graph Queries
  • 17. Conclusion: Think BIG about your EKG! • We encourage everyone to continue to “think big” when doing strategic planning around your Enterprise Knowledge Graphs • Many organizations besides Google, Amazon, Facebook etc. will be building 100B to 1TB vertex graphs in the next few years to better serve their customers • New hardware will soon be arriving that will dramatically accelerate today’s graph queries • Real time product recommendation systems that take complex content into account will be cost effective a accelerate data-driven decision making in healthcare and other industries 17