Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability

Similar documents
VPI / InfiniBand. Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability

VPI / InfiniBand. Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability

Ethernet. High-Performance Ethernet Adapter Cards

OCP3. 0. ConnectX Ethernet Adapter Cards for OCP Spec 3.0

PERFORMANCE ACCELERATED Mellanox InfiniBand Adapters Provide Advanced Levels of Data Center IT Performance, Productivity and Efficiency

Highest Levels of Scalability Simplified Network Manageability Maximum System Productivity

InfiniBand Switch System Family. Highest Levels of Scalability, Simplified Network Manageability, Maximum System Productivity

MELLANOX EDR UPDATE & GPUDIRECT MELLANOX SR. SE 정연구

Interconnect Your Future

In-Network Computing. Paving the Road to Exascale. 5th Annual MVAPICH User Group (MUG) Meeting, August 2017

Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA

Choosing the Best Network Interface Card for Cloud Mellanox ConnectX -3 Pro EN vs. Intel XL710

The Future of High Performance Interconnects

RoCE vs. iwarp Competitive Analysis

Mellanox Technologies Maximize Cluster Performance and Productivity. Gilad Shainer, October, 2007

Introduction to High-Speed InfiniBand Interconnect

Solutions for Scalable HPC

Building the Most Efficient Machine Learning System

In-Network Computing. Sebastian Kalcher, Senior System Engineer HPC. May 2017

Storage Protocol Offload for Virtualized Environments Session 301-F

High Performance Computing

Interconnect Your Future

Interconnect Your Future Enabling the Best Datacenter Return on Investment. TOP500 Supercomputers, November 2017

SUSE Linux Enterprise Server (SLES) 12 SP4 Inbox Driver Release Notes SLES 12 SP4

Introduction to Infiniband

Paving the Road to Exascale Computing. Yossi Avni

Mellanox Virtual Modular Switch

Building the Most Efficient Machine Learning System

ARISTA: Improving Application Performance While Reducing Complexity

Learn Your Alphabet - SRIOV, NPIV, RoCE, iwarp to Pump Up Virtual Infrastructure Performance

Scaling to Petaflop. Ola Torudbakken Distinguished Engineer. Sun Microsystems, Inc

Mellanox CloudX, Mirantis Fuel 5.1/ 5.1.1/6.0 Solution Guide

Mellanox OFED for FreeBSD for ConnectX-4/ConnectX-4 Lx/ ConnectX-5 Release Note. Rev 3.5.0

Application Acceleration Beyond Flash Storage

Accelerating Hadoop Applications with the MapR Distribution Using Flash Storage and High-Speed Ethernet

The Future of Interconnect Technology

N V M e o v e r F a b r i c s -

Birds of a Feather Presentation

Paving the Road to Exascale

InfiniBand Networked Flash Storage

Interconnect Your Future

Chelsio Communications. Meeting Today s Datacenter Challenges. Produced by Tabor Custom Publishing in conjunction with: CUSTOM PUBLISHING

Future Routing Schemes in Petascale clusters

InfiniBand OFED Driver for. VMware Infrastructure 3. Installation Guide

2008 International ANSYS Conference

How to Network Flash Storage Efficiently at Hyperscale. Flash Memory Summit 2017 Santa Clara, CA 1

Mellanox InfiniBand Solutions Accelerate Oracle s Data Center and Cloud Solutions

Benefits of Offloading I/O Processing to the Adapter

Red Hat Enterprise Linux (RHEL) 7.5-ALT Driver Release Notes

NVMe over Universal RDMA Fabrics

Flex System IB port FDR InfiniBand Adapter Lenovo Press Product Guide

SR-IOV Support for Virtualization on InfiniBand Clusters: Early Experience

In-Network Computing. Paving the Road to Exascale. June 2017

Interconnect Your Future

NFS/RDMA over 40Gbps iwarp Wael Noureddine Chelsio Communications

Hardened Security in the Cloud Bob Doud, Sr. Director Marketing March, 2018

InfiniBand OFED Driver for. VMware Virtual Infrastructure (VI) 3.5. Installation Guide

The Road to ExaScale. Advances in High-Performance Interconnect Infrastructure. September 2011

IBM WebSphere MQ Low Latency Messaging Software Tested With Arista 10 Gigabit Ethernet Switch and Mellanox ConnectX

STAR-CCM+ Performance Benchmark and Profiling. July 2014

Broadberry. Artificial Intelligence Server for Fraud. Date: Q Application: Artificial Intelligence

Creating an agile infrastructure with Virtualized I/O

2-Port 40 Gb InfiniBand Expansion Card (CFFh) for IBM BladeCenter IBM BladeCenter at-a-glance guide

Red Hat Enterprise Linux (RHEL) 7.4-ALT Driver Release Notes

Interconnect Your Future Paving the Road to Exascale

LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance

Intel PRO/1000 PT and PF Quad Port Bypass Server Adapters for In-line Server Appliances

Interconnect Your Future

Cavium FastLinQ 25GbE Intelligent Ethernet Adapters vs. Mellanox Adapters

SUSE Linux Enterprise Server (SLES) 12 SP3 Driver SLES 12 SP3

SNIA Developers Conference - Growth of the iscsi RDMA (iser) Ecosystem

iscsi Technology: A Convergence of Networking and Storage

Uncompromising Performance. Elastic Network Manageability. Maximum System Productivity.

SwitchX Virtual Protocol Interconnect (VPI) Switch Architecture

iser as accelerator for Software Defined Storage Rahul Fiske, Subhojit Roy IBM (India)

Ron Emerick, Oracle Corporation

The Mellanox ConnectX-2 Dual Port QSFP QDR IB network adapter for IBM System x delivers industryleading performance and low-latency data transfer

Mellanox ConnectX-4 NATIVE ESX Driver for VMware vsphere 5.5/6.0 Release Notes

Uncompromising Performance Elastic Network Manageability Maximum System Productivity

MPI Optimizations via MXM and FCA for Maximum Performance on LS-DYNA

INCREASE IT EFFICIENCY, REDUCE OPERATING COSTS AND DEPLOY ANYWHERE

PCI Express x8 Single Port SFP+ 10 Gigabit Server Adapter (Intel 82599ES Based) Single-Port 10 Gigabit SFP+ Ethernet Server Adapters Provide Ultimate

Optimizing LS-DYNA Productivity in Cluster Environments

Cisco UCS Virtual Interface Card 1225

SUSE Linux Enterprise Server (SLES) 15 Inbox Driver Release Notes SLES 15

Mellanox OFED for FreeBSD for ConnectX-4/ConnectX-5 Release Note. Rev 3.4.1

Performance Analysis and Evaluation of Mellanox ConnectX InfiniBand Architecture with Multi-Core Platforms

Cisco UCS Virtual Interface Card 1227

AcuSolve Performance Benchmark and Profiling. October 2011

QuickSpecs. Overview. HPE Ethernet 10Gb 2-port 535 Adapter. HPE Ethernet 10Gb 2-port 535 Adapter. 1. Product description. 2.

Mellanox ConnectX-4 NATIVE ESX Driver for VMware vsphere 5.5/6.0 Release Notes

2017 Storage Developer Conference. Mellanox Technologies. All Rights Reserved.

DataON and Intel Select Hyper-Converged Infrastructure (HCI) Maximizes IOPS Performance for Windows Server Software-Defined Storage

QLogic TrueScale InfiniBand and Teraflop Simulations

NVMe Direct. Next-Generation Offload Technology. White Paper

Altair OptiStruct 13.0 Performance Benchmark and Profiling. May 2015

40Gb/s InfiniBand Switch Module (HSSM) for IBM BladeCenter

At the heart of a new generation of data center infrastructures and appliances. Sept 2017

QuickSpecs. HP InfiniBand Options for HP BladeSystems c-class. Overview

Sharing High-Performance Devices Across Multiple Virtual Machines

Transcription:

Performance Accelerated Mellanox InfiniBand Adapters Provide Advanced Data Center Performance, Efficiency and Scalability

Mellanox InfiniBand Host Channel Adapters (HCA) enable the highest data center performance through the delivery of state-of-the-art solutions for High-Performance Computing (HPC), Machine Learning, Data Analytics, Database, Cloud and Storage Platforms. Today s exponential data growth is driving the need for intelligent and faster interconnect. Leveraging faster speeds and innovative In- Network Computing technologies, Mellanox InfiniBand adapters enable clustered databases, parallelized applications, transactional services, and high-performance embedded I/O applications to achieve significant performance improvements and scale, lowering cost per operation and increasing overall ROI. Mellanox delivers the most technologically advanced HCAs. Providing best-in-class performance and efficiency, they are the ideal solution for HPC clusters that demand high bandwidth, high message rate and low latency to achieve the highest server efficiency and application productivity. With RDMA traffic consolidation and hardware acceleration for virtualization, Mellanox HCAs provide optimal I/O services such as high bandwidth and server utilization to achieve the maximum return on investment (ROI) for data centers, high scale storage systems and cloud computing. By providing Virtual Protocol Interconnect (VPI), Mellanox HCAs offer the flexibility of connectivity for InfiniBand and Ethernet protocols within the same adapter.

World-Class Performance and Scale Mellanox InfiniBand adapters deliver industry-leading bandwidth with ultra lowlatency and efficient computing for performance-driven server and storage clustering applications. Network protocol processing and data movement such as RDMA and Send/Receive semantics are completed in the adapter without CPU intervention. Application acceleration, participation in the Scalable Hierarchical Aggregation and Reduction Protocol (SHARP TM ), and GPU communication acceleration bring further levels of performance improvement. The innovative acceleration technology in Mellanox InfiniBand adapters enables higher cluster efficiency and large scalability to hundreds of thousands of nodes. Complete End-to-End HDR InfiniBand Networking ConnectX adapters are part of Mellanox s full HDR 200Gb/s InfiniBand end-to-end portfolio for data centers and high-performance computing systems, which includes switches, application acceleration packages, and cables. Mellanox s Quantum family of HDR InfiniBand switches and Unified Fabric Management software incorporate advanced tools that simplify networking management and installation, and provide the needed capabilities for the highest scalability and future growth. Mellanox s HPC-X collectives, messaging, and storage acceleration packages deliver additional capabilities for the ultimate server performance, and the line of HDR copper and fiber cables ensure the highest interconnect performance. With Mellanox end to end, IT managers can be assured of the highest performance and most efficient network fabric. BENEFITS World-class cluster performance Networking and storage access Efficient use of compute resources Guaranteed bandwidth and low-latency services Smart interconnect for x86, Power, Arm, and GPU-based compute and storage platforms Increased VM per server ratio Virtualization acceleration Scalability to hundreds-of-thousands of nodes TARGET APPLICATIONS High-performance parallel computing Machine Learning and data analysis platforms Clustered database applications and high-throughput data warehousing Latency-sensitive applications such as financial analysis and high frequency trading Embedded systems leveraging high performance internal performance Performance storage applications such as backup, restore, mirroring, etc.

Virtual Protocol Interconnect VPI flexibility enables any standard networking, clustering, storage, and management protocol to seamlessly operate over any converged network leveraging a consolidated software stack. Each port can operate on InfiniBand or Ethernet fabrics, and supports IP over InfiniBand (IPoIB) and RDMA over Converged Ethernet (RoCE). VPI simplifies I/O system design and makes it easier for IT managers to deploy infrastructure that meets the challenges of a dynamic data center. I/O Virtualization Mellanox adapters provide comprehensive support for virtualized data centers with Single- Root I/O Virtualization (SR-IOV) allowing dedicated adapter resources and guaranteed isolation and protection for virtual machines (VM) within the server. I/O virtualization on InfiniBand gives data center managers better server utilization and LAN and SAN unification while reducing cost, power, and cable complexity. Multi-Host Solution Mellanox s Multi-Host technology provides high flexibility and major savings in building next generation, scalable high-performance data centers. Multi-Host connects multiple compute or storage hosts into a single interconnect adapter, separating the adapter PCIe interface into multiple and independent PCIe interfaces with no performance degradation. The technology enables designing and building new scale-out heterogeneous compute and storage racks with direct connectivity between compute elements, storage elements and the network, better power and performance management, while achieving maximum data processing and data transfer at minimum capital and operational expenses. Various Form Factors Mellanox adapter cards are available in a variety of form factors to meet every data center s specific needs. Open Compute Project (OCP) cards integrate into the most cost-efficient, energyefficient and scalable enterprise and hyperscale data centers, delivering leading connectivity for performance-driven server and storage applications. The OCP Mezzanine adapter form factor is designed to mate into OCP servers. Socket DirectTM cards split the PCIe bus into 2 buses, such that each CPU socket gets direct connectivity to the network. With this direct connectivity, traffic can bypass the inter-processors interface, optimizing performance and reducing latency for dual socket servers. Also, each CPU handles only its own traffic, improving CPU utilization. GPUDirect RDMA is also enabled for all CPU/GPU pairs, ensuring that all GPUs are linked to those CPUs that are closest to the adapter card. Socket Direct cards enable HDR 200Gb/s transmission rates for PCIe Gen3 servers, leveraging two PCIe Gen3 x16 slots. Other flavors of Socket Direct cards split 16-lane PCIe into two 8-lane buses. Storage Accelerated A consolidated compute and storage network provides significant cost-performance advantages over multi-fabric networks. Standard block and file access protocols leveraging InfiniBand RDMA result in high-performance storage access. Adapters support SRP, iser, NFS RDMA, SMB Direct, SCSI and iscsi, as well as NVMe over Fabrics storage protocols. ConnectX adapters also offer a flexible Signature Handover mechanism based on the advanced T-10/DIF implementation, and Erasure Coding offloading capability enabling distributed RAID (Redundant Array of Inexpensive Disks).

Enabling High Performance Computing (HPC) Applications Mellanox InfiniBand/VPI adapters are the perfect solution for the evolving Data-centric paradigm. Technologies within this model include the innovative In-Network Computing offloads that transform the data center interconnect into a distributed CPU, and distributed memory, overcoming performance bottlenecks and enabling faster and more scalable data analysis. Mellanox s advanced In-Network Computing accelerations and RDMA offload capabilities optimize the performance of a wide variety of HPC and machine learning systems in bioscience, media, automotive design, CFD and manufacturing, weather research, oil and gas, and other markets. As a core In-Networking Computing technology, Mellanox Scalable Hierarchical Aggregation and Reduction Protocol (SHARP) optimizes MPI operations performance, decreasing data load on the network and dramatically reducing MPI operations time, while freeing up CPU resources needed for other tasks. Software Support All Mellanox adapters are supported by a full suite of drivers for Microsoft Windows, Linux and FreeBSD major distributions. The adapters support OpenFabrics-based RDMA protocols and software and are compatible with configuration and management tools from various OEMs and operating system vendors. ConnectX -6 ConnectX-6 is the world s first 200Gb/s HDR InfiniBand and Ethernet network adapter card, offering world-leading performance, smart offloads and In-Network Computing. ConnectX-6 with VPI provides two ports of 200Gb/s supporting HDR, HDR100, EDR, FDR, QDR, DDR and SDR InfiniBand speeds, as well as 200, 100, 50, 40, 25, and 10Gb/s Ethernet speeds. ConnectX-6 supports sub-600 nanosecond latency, up to 215 million messages/sec, an embedded PCIe switch, and NVMe over Fabric. In addition to all the features included in earlier versions of ConnectX, ConnectX-6 offers Multi-Host support for up to 8 hosts and block-level encryption as a crucial innovation to network security, altogether delivering the highest performance, most secure and extremely flexible solution for today s demanding applications and markets. ConnectX -5 Intelligent ConnectX-5 adapter cards support Co-Design and In-Network computing, while introducing acceleration engines for maximizing HPC, data analytics and storage platforms. Supporting two ports of EDR 100Gb/s InfiniBand and 100Gb/s Ethernet connectivity, sub-600 nanosecond latency, a very high message rate, plus PCIe switch and NVMe over Fabric offloads. Includes new Message Passing Interface (MPI) offloads, e.g., MPI Tag Matching and MPI AlltoAll operations, advanced dynamic routing, and new data algorithms capabilities. Offers advanced application offloads supporting 100Gb/s for servers without x16 PCIe slots. ConnectX -4 Mellanox ConnectX-4 adapter cards with VPI combine the flexibility of InfiniBand and Ethernet protocol connectivity within the same adapter, to support EDR 100Gb/s InfiniBand and 100Gb/s Ethernet connectivity. Enabling extremely high throughput and low latency, ConnectX-4 is a high performance and flexible solution for data analytics, Web access and storage platforms. Enabling efficient I/O consolidation, the ConnectX-4 adapter card significantly reduces data center costs and complexity. ConnectX -3 Pro Mellanox s ConnectX-3 Pro Virtual Protocol Interconnect (VPI) adapter delivers high throughput across the PCI Express 3.0 host bus, by providing a FDR 56Gb/s InfiniBand and 40Gb Ethernet interconnect solution (up to 56GbE when connected to a Mellanox switch). Enabling fast transaction latency (less than 1usec), and delivery of more than 90M MPI messages/second, makes ConnectX -3 Pro a highly-scalable, suitable solution for transaction-demanding applications.

General Specs Ports Single, Dual Single, Dual Single, Dual Single, Dual Port Speed (Gb/s) PCIe IB: SDR, DDR, QDR, FDR10, FDR Eth: 10, 40, 56 Gen3 x8 IB: SDR, DDR, QDR, FDR, EDR Eth: 10, 25, 40, 50, 56, 100 Gen3 x8 Gen3 x16 IB: SDR, DDR, QDR, FDR, EDR Eth: 10, 25, 40, 50, 100 Gen3 x16 Gen4 x16 IB: SDR, DDR, QDR, FDR, EDR, HDR100, HDR Eth: 10, 25, 40, 50, 100, 200 Gen3/4 x16 32 lanes as 2x Gen3 x 16-lane PCIe Connectors QSFP+ QSFP28 QSFP28 QSFP56 RDMA Message Rate (million msgs/sec) 36 150 200 (ConnectX-5 Ex, Gen4 server) 165 (ConnectX-5, Gen3 server) Latency (us) 0.64 0.6 0.6 0.6 Typical Power (2 ports, max. speed) 6.2W 16.3W 19.3W (ConnectX-5 Ex, Gen4 server) 16.2W (ConnectX-5, Gen3 server) 215 Contact Mellanox Support RDMA OOO RDMA (Adaptive Routing) Dynamically Connected Transport Multi-Host 4 hosts 4 hosts 8 hosts Storage PRO * * NVMe-oF Target Offload Erasure Coding (RAID Offload) T-10 Dif/Signature Handover *ConnectX-5 and ConnectX-6 offer richer feature sets that are recommended for the latest market applications.

* * * PRO Virtualization SR-IOV 127 VFs 16 PFs per port, 254 VFs 16 PFs per port, 1000 VFs per port 16 PFs per port, 1K VFs per port Congestion Control (QCN, ECN) MPI Tag Matching Offload OVS Offload VM Isolation and Protection Security Block-level XTS-AES hardware encryption FIPS Capable Management Hairpin (Host Chaining) Host Management Multi-Host Isolation and Protection QoS Packet Pacing Form Factors OCP Socket Direct *ConnectX-5 and ConnectX-6 offer richer feature sets that are recommended for the latest market applications.

World-leading HPC centers are making the smart decision and choosing infiniband. This is why: Julich Supercomputer Centre The Julich Supercomputer Centre chose HPC Testimonia for a balanced, Co-Design approach to its interconnect, providing low latency, high throughput, and future scalability to its cluster, which contributes to projects in the areas of energy, environment, and brain research. CHPC South Africa The Centre for High Performance Computing in South Africa, the largest HPC facility in Africa, chose HPC Testimonia to enhance and unlock the vast potential of its system, which provides high end computational resources to a broad range of users in fields such as bioinformatics, climate research, material sciences, and astronomy. We chose a co-design approach, the appropriate hardware, and designed the system. This system was of course targeted at supporting in the best possible manner our key applications. The only interconnect that really could deliver that was HPC Testimonia. The heartbeat of the cluster is the interconnect. Everything is about how all these processes shake hands and do their work. InfiniBand and the interconnect is, in my opinion, what defines HPC.

For detailed information on features, compliance, and compatibility, please see each product s specific product brief. 350 Oakmead Parkway, Suite 100, Sunnyvale, CA 94085 Tel: 408-970-3400 Fax: 408-970-3403 www.mellanox.com This brochure describes hardware features and capabilities. Please refer to the driver release notes on mellanox.com for feature availability. Product images may not include heat sync assembly; actual product may differ. Copyright 2019. Mellanox Technologies. All rights reserved. Mellanox, Mellanox logo, ConnectX, GPUDirect, Mellanox Multi-Host, UFM, and Virtual Protocol Interconnect are registered trademarks of Mellanox Technologies, Ltd. HPC-X, Socket Direct, and Quantum are trademarks of Mellanox Technologies, Ltd. All other trademarks are property of their respective owners. 3525BR Rev 11.5