InfiniBand Strengthens Leadership as the Interconnect Of Choice By Providing Best Return on Investment

1. InfiniBand Strengthens Leadership as the Interconnect
Of Choice By Providing Best Return on Investment
TOP500 Supercomputers, Nov 2014

2. TOP500 Performance Trends
39% CAGR 79% CAGR
 Explosive high-performance computing market growth
 Clusters continue to dominate with 86% of the TOP500 list
Mellanox InfiniBand solutions provide the highest systems utilization in the TOP500
for both high-performance computing and clouds
© 2014 Mellanox Technologies 2

3. TOP500 Major Highlights
 InfiniBand is the de-facto Interconnect solution for High-Performance Computing
• Positioned to continue and expand in Cloud and Web 2.0
 InfiniBand is the most used interconnect on the TOP500 list, connecting 225 systems
• Increasing 8.7% year-over-year, from 207 system in Nov’13 to 225 in Nov’14
 FDR InfiniBand is the most used technology on the TOP500 list, connecting 141 systems
• Grew 1.8X year-over-year from 80 systems in Nov’13 to 141 in Nov’14
 InfiniBand enables the most efficient system on the list with 99.8% efficiency – record!
 InfiniBand enables 24 out of the 25 most efficient systems (and the top 17 most efficient systems)
 InfiniBand is the most used interconnect for Petascale-performance systems with 24 systems
© 2014 Mellanox Technologies 3

4. InfiniBand in the TOP500
 InfiniBand is the most used interconnect technology for high-performance computing
• InfiniBand accelerates 225 systems, 45% of the list
 FDR InfiniBand connects the fastest InfiniBand systems
• TACC (#7), Government (#10), NASA (#11), Eni (#12), LRZ (#14)
 InfiniBand connects the most powerful clusters
• 24 of the Petascale-performance systems
 The most used interconnect solution in the TOP100, TOP200, TOP300, TOP400
• Connects 48% (48 systems) of the TOP100 while Ethernet only 2% (2 system)
• Connects 51.5% (103 systems) of the TOP200 while Ethernet only 15.5% (31 systems)
• Connects 50.7% (152 systems) of the TOP300 while Ethernet only 24.3% (73 systems)
• Connects 49% (196 systems) of the TOP400 while Ethernet only 29.8% (119 systems)
• Connects 45% (225 systems) of the TOP500, Ethernet connects 37.4% (187 systems)
 InfiniBand is the interconnect of choice for accelerator-based systems
• 80% of the accelerator-based systems are connected with InfiniBand
 Diverse set of applications
• High-end HPC, commercial HPC, Cloud and enterprise data center
© 2014 Mellanox Technologies 4

5. Mellanox in the TOP500
 Mellanox FDR InfiniBand is the fastest interconnect solution on the TOP500
• More than 12GB/s throughput, less than 0.7usec latency
• Being used in 141 systems on the TOP500 list – 1.8X increase from the Nov’13 list
• Connects the fastest InfiniBand-based supercomputers
- TACC (#7), Government (#10), NASA (#11), Eni (#12), LRZ (#14)
 Mellanox InfiniBand is the most efficient interconnect technology on the list
• Enables the highest system utilization on the TOP500 – 99.8% system efficiency
• Enables the top 17 and 24 out of 25 highest utilized systems on the TOP500 list
 Mellanox InfiniBand is the only Petascale-proven, standard interconnect solution
• Connects 24 out of the 54 Petaflop-performance systems on the list
• Connects 1.5X the number of Cray based systems in the Top100, 5X in TOP500
 Mellanox’s end-to-end scalable solutions accelerate GPU-based systems
• GPUDirect RDMA technology enables faster communications and higher performance
© 2014 Mellanox Technologies 5

6. TOP500 Interconnect Trends
 InfiniBand is the de-facto interconnect solution for performance demanding applications
© 2014 Mellanox Technologies 6

7. TOP500 Petascale-Performance Systems
 Mellanox InfiniBand is the interconnect of choice for Petascale computing
• Overall 24 system of which 19 System use FDR InfiniBand
© 2014 Mellanox Technologies 7

8. TOP100: Interconnect Trends
 InfiniBand is the most used interconnect solution in the TOP100
 The natural choice for world-leading supercomputers: performance, efficiency, scalability
© 2014 Mellanox Technologies 8

9. TOP500 InfiniBand Accelerated Systems
 Number of Mellanox FDR InfiniBand systems grew 1.8X from Nov’13 to Nov’14
• Accelerates 63% of the InfiniBand-based systems (141 systems out of 225)
© 2014 Mellanox Technologies 9

10. InfiniBand versus Ethernet – TOP100, 200, 300, 400, 500
 InfiniBand is the most used interconnect of the TOP100, 200, 300, 400 supercomputers
 Due to superior performance, scalability, efficiency and return-on-investment
© 2014 Mellanox Technologies 10

11. TOP500 Interconnect Placement
 InfiniBand is the high performance interconnect of choice
• Connects the most powerful clusters, and provides the highest system utilization
 InfiniBand is the best price/performance interconnect for HPC systems
© 2014 Mellanox Technologies 11

12. InfiniBand’s Unsurpassed System Efficiency
Average Efficiency
• InfiniBand: 87%
• Cray: 79%
• 10GbE: 67%
• GigE: 40%
 TOP500 systems listed according to their efficiency
 InfiniBand is the key element responsible for the highest system efficiency; in average 30% higher than 10GbE
 Mellanox delivers efficiencies of up to 99.8% with InfiniBand
© 2014 Mellanox Technologies 12

13. Enabling The Most Efficient Systems
 Mellanox InfiniBand connects the most efficient system on the list
• 24 of the 25 most efficient systems
 Enabling a record system efficiency of 99.8%, only 0.2% less than the theoretical limit!
© 2014 Mellanox Technologies 13

14. TOP500 Interconnect Comparison
 InfiniBand systems account for 2.6X the performance of Ethernet systems
 The only scalable HPC interconnect solutions
© 2014 Mellanox Technologies 14

15. TOP500 Performance Trend
 InfiniBand connected systems’ performance demonstrate highest growth rate
 InfiniBand responsible for 2.6X the performance versus Ethernet on the TOP500 list
© 2014 Mellanox Technologies 15

16. InfiniBand Performance Trends
 InfiniBand-connected CPUs grew 28% from Nov‘13 to Nov‘14
 InfiniBand-based system performance grew 44% from Nov‘13 to Nov‘14
 Mellanox InfiniBand is the most efficient and scalable interconnect solution
 Driving factors: performance, efficiency, scalability, many-many cores
© 2014 Mellanox Technologies 16

17. Mellanox Accelerated World-Leading HPC Systems
Connecting Half of the World’s Petascale Systems (examples)
© 2014 Mellanox Technologies 17

18. Proud to Accelerate Future DOE Leadership Systems (“CORAL”)
“Summit” System “Sierra” System
5X – 10X Higher Application Performance versus Current Systems
Mellanox EDR 100Gb/s InfiniBand, IBM POWER CPUs, NVIDIA Tesla GPUs
Mellanox EDR 100G Solutions Selected by the DOE for 2017 Leadership Systems
Deliver Superior Performance and Scalability over Current / Future Competition
© 2014 Mellanox Technologies 18

19. Interconnect Solutions Leadership – Software and Hardware
Comprehensive End-to-End Interconnect Software Products
Comprehensive End-to-End InfiniBand and Ethernet Hardware Products
ICs Adapter Cards Switches/Gateways Metro / WAN Cables/Modules
© 2014 Mellanox Technologies 19

20. The Future is Here
Enter the Era of 100Gb/s
Copper (Passive, Active) Optical Cables (VCSEL) Silicon Photonics
36 EDR (100Gb/s) Ports, <90ns Latency
Throughput of 7.2Tb/s
100Gb/s Adapter, 0.7us latency
150 million messages per second
(10 / 25 / 40 / 50 / 56 / 100Gb/s)
© 2014 Mellanox Technologies 20

21. The Advantage of 100G Copper Cables (over Fiber)
Lower Cost, Lower Power Consumption, Higher Reliability
5000 Nodes Cluster
CAPEX Saving: $2.1M
Power Saving: 12.5KWh
1000 Nodes Cluster
CAPEX Saving: $420K
Power Saving: 2.5KWh
SC’14 Demonstration: 4m, 6m and 8m Copper Cables
© 2014 Mellanox Technologies 21

22. Enter the World of Scalable Performance – 100Gb/s Switch
Boundless Performance
7th Generation InfiniBand Switch
36 EDR (100Gb/s) Ports, <90ns Latency
Throughput of 7.2 Tb/s
InfiniBand Router
Adaptive Routing
Analyze Store
© 2014 Mellanox Technologies 22

23. Enter the World of Scalable Performance – 100Gb/s Adapter
ConnectX-4: Highest Performance Adapter in the Market
Connect. Accelerate. Outperform
InfiniBand: SDR / DDR / QDR / FDR / EDR
Ethernet: 10 / 25 / 40 / 50 / 56 / 100GbE
100Gb/s, <0.7us latency
150 million messages per second
OpenPOWER CAPI technology
CORE-Direct technology
GPUDirect RDMA
Dynamically Connected Transport (DCT)
Ethernet offloads (HDS, RSS, TSS, LRO, LSOv2)
© 2014 Mellanox Technologies 23

24. End-to-End Interconnect Solutions for All Platforms
Highest Performance and Scalability for
X86, Power, GPU, ARM and FPGA-based Compute and Storage Platforms
x86
Open
POWER
GPU ARM FPGA
Smart Interconnect to Unleash The Power of All Compute Architectures
© 2014 Mellanox Technologies 24

25. Mellanox Delivers Highest Application Performance
World Record Performance with Mellanox Connect-IB and HPC-X
Achieve Higher Performance with 67% less compute infrastructure versus Cray
25% 5%
© 2014 Mellanox Technologies 25

26. HPC Clouds – Performance Demands Mellanox Solutions
San Diego Supercomputing Center “Comet” System (2015) to
Leverage Mellanox Solutions and Technology to Build HPC Cloud
© 2014 Mellanox Technologies 26

27. Technology Roadmap – One-Generation Lead over the Competition
20Gbs 40Gbs 56Gbs 100Gbs
Mellanox
Terascale Petascale Exascale
3rd 1st
“Roadrunner”
Mellanox Connected
TOP500 2003
Virginia Tech (Apple)
200Gbs
2000 2005 2010 2015
2020
© 2014 Mellanox Technologies 27

28. Mellanox Interconnect Advantages
 Mellanox solutions provide a proven, scalable and high performance end-to-end connectivity
 Standards-based (InfiniBand , Ethernet), supported by large eco-system
 Flexible, support all compute architectures: x86, Power, ARM, GPU, FPGA etc.
 Backward and future compatible
 Proven, most used solution for Petascale systems and overall TOP500
Paving The Road to Exascale Computing
© 2014 Mellanox Technologies 28

29. Texas Advanced Computing Center/Univ. of Texas - #7
 “Stampede” system
 6,000+ nodes (Dell), 462462 cores, Intel Phi co-processors
 5.2 Petaflops
 Mellanox end-to-end FDR InfiniBand
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 29

30. NASA Ames Research Center - #11
 Pleiades system
 SGI Altix ICE
 20K InfiniBand nodes
 3.4 sustained Petaflop performance
 Mellanox end-to-end FDR and QDR InfiniBand
 Supports variety of scientific and engineering projects
• Coupled atmosphere-ocean models
• Future space vehicle design
• Large-scale dark matter halos and galaxy evolution
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 30

31. Exploration & Production ENI S.p.A. - #12
 “HPC2” system
 IBM iDataPlex DX360M4
 NVIDIA K20x GPUs
 3.2 Petaflops sustained Petaflop performance
 Mellanox end-to-end FDR InfiniBand
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 31

32. Leibniz Rechenzentrum SuperMUC - #14
 IBM iDataPlex and Intel Sandy Bridge
 147456 cores
 Mellanox end-to-end FDR InfiniBand solutions
 2.9 sustained Petaflop performance
 The fastest supercomputer in Europe
 91% efficiency
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 32

33. Tokyo Institute of Technology - #15
 TSUBAME 2.0, first Petaflop system in Japan
 2.8 PF performance
 HP ProLiant SL390s G7 1400 servers
 Mellanox 40Gb/s InfiniBand
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 33

34. DOE/SC/Pacific Northwest National Laboratory - #18
 “Cascade” system
 Atipa Visione IF442 Blade Server
 2.5 sustained Petaflop performance
 Mellanox end-to-end InfiniBand FDR
 Intel Xeon Phi 5110P accelerator
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 34

35. Total Exploration Production - #20
 “Pangea” system
 SGI Altix X system, 110400 cores
 Mellanox FDR InfiniBand
 2 sustained Petaflop performance
 91% efficiency
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 35

36. Grand Equipement National de Calcul Intensif - Centre Informatique
National de l'Enseignement Suprieur (GENCI-CINES) - #26
 Occigen system
 1.6 sustained Petaflop performance
 Bull bullx DLC
 Mellanox end-to-end FDR InfiniBand
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 36

37. Air Force Research Laboratory - #31
 “Spirit” system
 SGI Altix X system, 74584 cores
 Mellanox FDR InfiniBand
 1.4 sustained Petaflop performance
 92.5% efficiency
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 37

38. CEA/TGCC-GENCI - #33
 Bull Bullx B510, Intel Sandy Bridge
 77184 cores
 Mellanox end-to-end FDR InfiniBand solutions
 1.36 sustained Petaflop performance
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 38

39. Max-Planck-Gesellschaft MPI/IPP - # 34
 IBM iDataPlex DX360M4
 Mellanox end-to-end FDR InfiniBand solutions
 1.3 sustained Petaflop performance
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 39

40. National Supercomputing Centre in Shenzhen (NSCS) - #35
 Dawning TC3600 Blade Supercomputer
 5200 nodes, 120640 cores, NVIDIA GPUs
 Mellanox end-to-end 40Gb/s InfiniBand solutions
• ConnectX-2 and IS5000 switches
 1.27 sustained Petaflop performance
 The first Petaflop systems in China
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 40

41. NCAR (National Center for Atmospheric Research) - #36
 “Yellowstone” system
 72,288 processor cores, 4,518 nodes (IBM)
 Mellanox end-to-end FDR InfiniBand, full fat tree, single plane
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 41

42. International Fusion Energy Research Centre (IFERC), EU(F4E) -
Japan Broader Approach collaboration - #38
 Bull Bullx B510, Intel Sandy Bridge
 70560 cores
 1.24 sustained Petaflop performance
 Mellanox end-to-end InfiniBand solutions
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 42

43. SURFsara - #45
 The “Cartesius” system - the Dutch supercomputer
 Bull Bullx DLC B710/B720 system
 Mellanox end-to-end InfiniBand solutions
 1.05 sustained Petaflop performance
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 43

44. Commissariat a l'Energie Atomique (CEA) - #47
 Tera 100, first Petaflop system in Europe - 1.05 PF performance
 4,300 Bull S Series servers
 140,000 Intel® Xeon® 7500 processing cores
 300TB of central memory, 20PB of storage
 Mellanox 40Gb/s InfiniBand
Petaflop
Mellanox Connected
© 2014 Mellanox Technologies 44

45. Australian National University - #52
 Fujitsu PRIMERGY CX250 S1
 Mellanox FDR 56Gb/s InfiniBand end-to-end
 980 Tflops performance
© 2014 Mellanox Technologies 45

46. Purdue University - #53
 “Conte” system
 HP Cluster Platform SL250s Gen8
 Intel Xeon E5-2670 8C 2.6GHz
 Intel Xeon Phi 5110P accelerator
 Total of 77,520 cores
 Mellanox FDR 56Gb/s InfiniBand end-to-end
 Mellanox Connect-IB InfiniBand adapters
 Mellanox MetroX long Haul InfiniBand solution
 980 Tflops performance
© 2014 Mellanox Technologies 46

47. Barcelona Supercomputing Center - #57
 “MareNostrum 3” system
 1.1 Petaflops peak performance
 ~50K cores, 91% efficiency
 Mellanox FDR InfiniBand
© 2014 Mellanox Technologies 47

48. Thank You