Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference ›...
Transcript of Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference ›...
![Page 1: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/1.jpg)
Petascale to ExascaleExtending Intel’s HPC Commitment
Kirk SkaugenVice President, Intel Corporation
General Manager, Data Center Group
![Page 2: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/2.jpg)
Congratulations Prof. Dr. Meueron the 25th
Anniversary of ISC
Other names and brands may be claimed as the property of others
![Page 3: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/3.jpg)
25 Years Also = Intel Beginnings in HPC: “The Cosmic Cube”
• Scalable from 32 to 128 nodes– Intel 80286 microprocessor
– Intel 80287 math coprocessor
– 512K RAM local memory
– Ethernet-connected hypercube
• Peak performance of 3.2 MFLOPS
Cosmic Cube image © Marvel Comics
![Page 4: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/4.jpg)
A Rich History of Silicon and Software Innovation for HPC
OpenMP support
1995 20052000 20101990
Hyperthreading support
Multi-core support
Cilk, Co-Array Fortran support
UNIX Compilers
Linux Compilers
VTune™ Perf
Analyzer
MPI Library
Threading Building Blocks
Cluster Toolkit
Math Kernel Library
Other names and brands may be claimed as the property of others
![Page 5: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/5.jpg)
Intel Top 500 Market Adoption
20052000 20100
50
100
150
200
250
300
350
400
450
500Intel in Top 500 Supercomputers
Source: www.top500.org
Value Proposition:
Volume economics
IA programming model
Robust ecosystem
![Page 6: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/6.jpg)
Moore’s Law…the number of transistors
on a chip will double about every two years…
Performance for serial and parallel applications
More cores, threads and performance at similar to lower power levels
Transformed the Economics of HPC
![Page 7: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/7.jpg)
But can Moore’s Law
continue?
![Page 8: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/8.jpg)
Moore’s Law: Alive and Well at Intel
15nm
2013*11nm
2015*8nm
2017* 2019+MANUFACTURING DEVELOPMENT
45nm
200732nm
200922nm
2011*
RESEARCH
65nm
2005
Intel Innovation-Enabled Technology Pipeline is Full
![Page 9: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/9.jpg)
Still an Insatiable Need for Computing
10 PFlops
1 PFlops
100 TFlops
10 TFlops
1 TFlops
100 GFlops
10 GFlops
1 GFlops
100 MFlops
100 PFlops
10 EFlops
1 EFlops
100 EFlops
1993 20171999 2005 2011 2023
1 ZFlops
2029
Climate Simulation
Medical Imaging
Genomics Research
Source: www.top500.org
![Page 10: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/10.jpg)
High Performance Micro-Architecture for PetaScale Deployments
Tick Tock Tick Tock
32nm
Westmere Sandy Bridge
22nm
Tick Tock
Ivy Bridge Future
New instructions:
Tick Tock
65nm
Core™ Harpertown
45nm
Penryn Nehalem
AVX Future - FMA SSE4.2 AESSSE4.1SSSE3
![Page 11: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/11.jpg)
Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500
Intel® Xeon® 7500
4-socket Performance 20X
Source: Intel internally measured results 15 January 2010. Each bar represents the score or estimated score of best measured/estimated results on the geometric mean of internal benchmarks (server-side Java*, integer throughput, floating-point throughput, ERP, and OLTP). Results have been estimated based on internal Intel analysis and are provided for informational purposes only. Any difference in system hardware or software design or configuration may affect actual performance. Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, Go to: http://www.intel.com/performance/resources/benchmark_limitations.htm. Relative performance is calculated by assigning a baseline value of 1.0 to one benchmark result, and then dividing the actual benchmark result for the baseline platform into each of the specific benchmark results of each of the other platforms, and assigning them a relative performance number that correlates with the performance improvements reported.
![Page 12: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/12.jpg)
Jean Gonnord
Program Director for Numerical Simulation & Computer Sciences
CEA DAM
Other names and brands may be claimed as the property of others
![Page 13: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/13.jpg)
TERA 100
Jean Gonnord
Chef de ProjetSimulation numérique
CEA/DAM
May 27th 2010
First petaflop/s computer ever designed and built in Europe
Four weeks ahead of initial planning
A significant industrial success
Jean Philippe Nominé
Chargé d’affaire HPCMember of PRACE Technical Board
CEA/DAM
![Page 14: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/14.jpg)
TERA 100 a machine of world records
Beside the records,a production machine, with high level of reliabilty, for CEA strategic needs
1.25 Petaflop/s peak4300 nodes140 000 cores, Intel Xeon® 7500 series
300 TB memory20 PB disk storage
500 GB/s bandwidth to the global file system
QDR Infiniband interconnectOpen source software stack
![Page 15: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/15.jpg)
CEA : a major actor in the HPC field
Astrophysics BiologyNuclear Energy
Defense SecurityClimate
Numerical simulation is an essential tool
TERA 100 a step on the CEA roadmap
![Page 16: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/16.jpg)
CEA : a major actor in the HPC field
Astrophysics BiologyNuclear Energy
Defense SecurityClimate
Numerical simulation is an essential tool
TERA 100 a step on the CEA roadmap
1996 2001
2005
N°5 MondeN°1 EU4,8 TF
TERA 1
N°5 MondeN°1 EU63 TF
TERA 10
N° ? MondeN°? EU1Pf
TERA 100
x10
x20
x30
x30
N° ? MondeN°? EU30Pf
TERA 1000
N° ? MondeN°? EU 1 EfEXA 1
2001
2019
TGCC
TGCC
TGCC CEA open computing center
CEA classified computing center
![Page 17: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/17.jpg)
CEA : a major actor in the HPC field
Astrophysics BiologyNuclear Energy
Defense SecurityClimate
Numerical simulation is an essential tool
CEA/DAM has the operational responsibility of implementing this roadmap
TERA 100 a step on the CEA roadmap
Maintaining the capacity of designing and building very
large computing systems in Europe
An ambitiousroadmap
fora strategic goal:
1996 2001
2005
N°5 MondeN°1 EU4,8 TF
TERA 1
N°5 MondeN°1 EU63 TF
TERA 10
N° ? MondeN°? EU1Pf
TERA 100
x10
x20
x30
x30
N° ? MondeN°? EU30Pf
TERA 1000
N° ? MondeN°? EU 1 EfEXA 1
2001
2019
TGCC
TGCC
TGCC CEA open computing center
CEA classified computing center
![Page 18: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/18.jpg)
TERA 100 a great thank to INTEL
With a special mention to Richard Dracott
For delivering us on schedule the 18000 NEHALEM-EX chips
For giving us the opportunity of this presentation
Join us tomorrow at 1pm at BULL booth 320 for a drink
We will begin to prepare the future of European HPC with you
![Page 19: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/19.jpg)
The Next Generation Xeon ProcessorSandy Bridge “Tock”
• Significantly greater performancewith higher core-count & Intel® Hyper-threading Technology
• 2x Flops / clock peak using new AVX instructions
Making Petascale Widely Available for Leading Science
![Page 20: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/20.jpg)
Increasing number of cores & threadsVector instructions
Petascale Programming Challenges
?Irregular Patterns and Data Structures
Scale to Multi-Core → HardScale to Many-Core→ Harder
![Page 21: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/21.jpg)
IA Programming Flexibility
Programming choices and standards for range of parallel efficiency
Instruction Parallelism
Data Parallelism
Thread Parallelism
Cluster / Process
Parallelism
Serial Code Node LevelFast Scalar performance, Optimized C/C++,FORTRAN, Threading and
Performance Libraries, Debug / Analysis Tools
Parallel Node LevelMulti-core, Multi-Socket, SSE and AVX instructions, OpenMP, Threading Building
Blocks, Performance Libraries, Thread Checker , Ct , Cilk
Multi-Node / Cluster LevelCluster Tools, MPI Checker
![Page 22: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/22.jpg)
Simplifying Software Development:Intel® Software Development Tools
Parallel Studio HPC Tools Cluster Tools
Essential Parallelism Advanced Parallelism Distributed Parallelism
Tools to preserve your source code investments
![Page 23: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/23.jpg)
2K universities in 88 countries
4K faculty trained
320K students trained
Parallel Programming Education
Source: Evans Data Corporation, Intel
intel.com/thinkparallel
![Page 24: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/24.jpg)
Exascale: The Next Frontier
Intel committed to solving the challenges of Exascale
Energy Physics Biology Climate Astrophysics
Challenges• Power – energy / operation of computation, data transport, memory
• Threading software to millions/billions of threads
• Memory/Storage capacity and bandwidth
• Managing high-node count systems in the existence of failures (MTBF)
• Affordability
![Page 25: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/25.jpg)
Intel Co-Sponsored HPC Labs in Europe
Advancing Exascale Computing on Intel Architecture
ExaTec Lab, Paris
Performance and scalability of Exascale applications
ExaCluster Lab, Jülich
Exascale cluster scalability and reliability
Introducing Today
Other names and brands may be claimed as the property of others
![Page 26: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/26.jpg)
Dr. Pradeep Dubey
Senior Principal EngineerIEEE Fellow
Director of the ThroughputComputing Lab, Intel Labs
![Page 27: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/27.jpg)
Intel’s Many-Core Research Program
High BandwidthI/O & Communications
Stacked, Shared Memory
Scalable Architectures
Parallel ProgrammingTools & Techniques
VirtualEnvironments
FinancialModeling
Media Search& Manipulation
Web MiningBots
EducationalSimulation
Thread-AwareExecution Environment
![Page 28: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/28.jpg)
Application-Driven Architecture Research
Constantly Evaluating Options for All Workloads
0
10
20
30
40
50
60
70
0 8 16 24 32 40 48 56 64
Para
llel S
peed
up
Number of cores
Merge Sort (256M elements) Tree Search (64M keys)Graph Search (600 Regular Expressions) LU Foreground Estimation Text IndexingGame Cloth Home Video EditingSports Video Analysis Human Body TrackingProduction Fluid Marching CubesVideo Cast Indexing Portifolio MangementGame Rigid Body Production ClothVolume Rendering (0.5-1GB dataset) Crowd Sim (100K Agents)
![Page 29: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/29.jpg)
Intel LabsParallel Computing Research
Single Chip Cloud
ComputerDec 2009
Tera-scale Research Processor
Mar 2007
Research Processors from Intel Labs
![Page 30: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/30.jpg)
Intel LabsParallel Computing Research
Single Chip Cloud
ComputerDec 2009
Tera-scale Research Processor
Mar 2007
1999 - 2006Origination of Intel’s multi-core explorationsComing Challenges in Microarchitecture and Architecture“Era of Tera” Keynote at Intel Developer ForumRecognition, Mining , Synthesis Moves Computers to the Era of TeraHundreds of Cores: Scaling to Tera-scale ArchitectureFew Cores to Many: A Tera-scale Computing Research Overview
2007Demonstration of Intel experimental 80-core processor“Ct” language proposal for Tera-scale ArchitecturesIntel® C++ STM Compiler Prototype release Integration Challenges and Tradeoffs for Tera-scale Architectures Package Technology to Address the Memory Bandwidth Challenge for Tera-scale ComputingRuntime Environment for Tera-scale Platforms Architectural Support for Fine-Grained Parallelism on Multi-coreDatacenter-on-Chip Architectures: Tera-scale Opportunities and Challenges in Intel's Manufacturing EnvironmentMedia Mining—Emerging Tera-scale Computing Applications High-Performance Physical Simulations on Next-Generation
Research Processors from Intel Labs
![Page 31: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/31.jpg)
Intel LabsParallel Computing Research
Single Chip Cloud
ComputerDec 2009
Tera-scale Research Processor
Mar 2007
Architecture with Many CoresDemonstrated Intel McRT (“Manycore Runtime”) 11 Issue 03Carbon: Architectural Support for Fine-Grained Parallelism on Chip MultiprocessorsPhysical Simulation for Animation and Visual Effects: Parallelization and Characterization for Chip MultiprocessorsScaling performance of interior-point method on large-scale chip multiprocessor system
2008Second Life and the New Generation of Virtual WorldsLarrabee: A Many-Core x86 Architecture for Visual ComputingAtomic Vector Operations on Chip MultiprocessorsEfficient Implementation of Sorting on Multi-Core SIMD CPU ArchConvergence of Recognition, Mining, and Synthesis Workloads and Its Implications Accelerating Video-Mining Applications Using Many Small, General-Purpose Cores
2009Mapping High-Fidelity Volume Rendering for Medical Imaging to CPU, GPU and Many-Core ArchitecturesCl P h Hi hl P ll l C lli i A id M l i A Si l i
Research Processors from Intel Labs
![Page 32: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/32.jpg)
From Research to Realization. Announcing…
The Newest Addition to the Intel Server Family.Industry’s First General Purpose Many Core Architecture
Intel®Many
Integrated
Core
Architecture
![Page 33: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/33.jpg)
Intel® MIC Architecture:An Intel Co-Processor Architecture
VECTORIA CORE
INTERPROCESSOR NETWORK
INTERPROCESSOR NETWORK
FIX
ED F
UN
CTIO
N L
OGI
C
MEM
ORY
and
I/O
INTE
RFA
CESVECTOR
IA COREVECTORIA CORE
VECTORIA CORE
VECTORIA CORE
VECTORIA CORE
VECTORIA CORE
VECTORIA CORE
COHERENTCACHE
…
……
…
COHERENTCACHE
COHERENTCACHE
COHERENTCACHE
COHERENTCACHE
COHERENTCACHE
COHERENTCACHE
COHERENTCACHE
Many cores and many, many more threads
Standard IA programming and memory model
![Page 34: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/34.jpg)
Knights Ferry
• Software development platform
• Growing availability through 2010
• 32 cores, 1.2 GHz
• 128 threads at 4 threads / core
• 8MB shared coherent cache
• 1-2GB GDDR5
• Bundled with Intel HPC tools
Software development platform for Intel® MIC architecture
![Page 35: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/35.jpg)
Intel® MIC Architecture Programming
Intel® Xeon®
processor family
Intel® Xeon ®
processorIntel® MIC
architectureco-processor
Single Source
Compilersand Runtimes Common with Intel® Xeon®
• Languages
• C, C++, Fortran compilers
• Intel developer tools and libraries
• Coding and optimization techniques
• Ecosystem support
Eliminates Need for Dual Programming Architecture
![Page 36: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/36.jpg)
Knights Ferry Demo
![Page 37: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/37.jpg)
Summary
11/09: Leading performance SGEMM (>1 Teraflop)
11/09: Leading performance SpMVM
Today: Leading performance LU (>½ Teraflop)
Other names and brands may be claimed as the property of others
![Page 38: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/38.jpg)
The Knights FamilyFuture Knights
Products
Knights Corner1st Intel® MIC product
22nm process
>50 Intel Architecture cores
Knights Ferry
![Page 39: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/39.jpg)
Sverre Jarp
Chief Technical Officer
CERN Openlab
Other names and brands may be claimed as the property of others
![Page 40: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/40.jpg)
27 March 2006 Intel visit 40
• LHC is 27 km in circumference, 100 m underground, and operates at 1.9o Kelvin
• It has now been up and running since November 2009
• World record in beam energy• 3.5 T achieved as of 30 March• By now (May 2010): Over
1’000’000’000 events recorded
Four experiments, with detectors as ‘big as cathedrals’:ALICEATLASCMSLHCb
CERN’s Large Hadron Collider
![Page 41: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/41.jpg)
World-wide LHC Computing Grid Largest Grid service in the world !
41
• Almost 160 sites in 34 countries
• More than 200’000 IA processor cores (w/Linux)
• 20% at CERN
![Page 42: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/42.jpg)
Reconstruction
Online
Simulation (GEANT4)
Analysis (ROOT)
Batchphysicsanalysis
detector
Event Summary
Data
RAWdata
Eventreprocessing
Eventsimulation
analysis objects(extracted by physics topic)
Selection &reconstruction
processeddata
100% 10%
1%
Online triggerand filtering
Interactivephysicsanalysis
Data Handling and Computation for Physics Analysis
High Level Trigger
![Page 43: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/43.jpg)
Summary
• At Intel, Moore’s Law is alive and well
• Sandy Bridge & AVX drives Xeon family on a new FP trajectory
Broad new Intel Supercomputing investments:
• New: Exascale lab with FZ Jülich, Partec, Intel
• New: Intel® Many Integrated Core (MIC) architecture
• New: Heterogeneous IA HPC tools to simplify road to Exascale
• New: Knights family of co-processors – Knights Ferry software development platform
– Knights Corner product targeting 22nm and >50 Intel Architecture cores
![Page 44: Petascale to Exascale - Inteldownload.intel.com › pressroom › archive › reference › ISC... · Xeon 3.33 Xeon 7100 Xeon 7300 Xeon 7400 Xeon 7500 Intel® Xeon® 7500 4-socket](https://reader035.fdocuments.in/reader035/viewer/2022062414/5f03b8e27e708231d40a72f6/html5/thumbnails/44.jpg)