In-Network Computing and Next Generation HDR 200G InfiniBand

October 23, 2017

Three HPC practitioners stand at the dawn of the fourth industrial revolution. What do they call “Big Data”?  The humorous answer is of course, just “Data,” but the reality is that with the exponential growth of data that needs to be analyzed and the data resulting from ever-more complex workflows, the need for data movement has never been more challenging and critical to the worlds of High Performance Computing (HPC) and machine learning.  Mellanox Technologies, the leading global supplier of end-to-end InfiniBand interconnect solutions and services for servers, storage, and hyper-converged infrastructure, is once again moving the bar forward with the introduction of and end-to-end HDR 200G InfiniBand product portfolio.

When Fast Isn’t Fast Enough

In an age of digital transformation and big data analytics, the need for HPC and deep learning platforms to move and analyze data both in real-time and at faster speeds is ever increasing. Machine learning enables enterprises to leverage the vast amounts of data being generated today to make faster and more accurate decisions. Whether for brain mapping or for homeland security, the most demanding supercomputers and data center applications need to produce astounding achievements, often in real time!

Until relatively recently, state-of-the-art applications analyzing automotive construction or weather simulations enjoyed the data performance and throughput speeds offered by 100G interconnect. Today’s HPC, machine learning, storage and hyperscale now require both faster interconnect solutions and more intelligent networks to analyze data and run complex simulations with greater speed and efficiency. With the expected volume of network data availability doubling by the end of 2019, it only makes sense to prepare for the eventuality of demand for HDR 200G interconnectivity capabilities.

Exponential Data Growth – Everywhere

From the CPU to the Data

The need to support growing data speeds, throughput and simulation complexity accompanies a widespread recognition that the CPU has reached the limits of its scalability. Many have joined the ranks of those believing that it is no longer feasible to move data all of the way to the compute elements; rather computational operations should be performed on the data wherever the data is. Thus the start of a “Data-Centric” trend toward offloading network functions from the CPU to the network. By lightening the load on the server’s processors, the CPUs can devote all their cycles to the application. This approach increases system efficiency by allowing users to run algorithms on the data in-transit rather than waiting for the data to reach the CPU.

Next-Generation Machine Learning Interconnect Solutions

Mellanox’s forthcoming HDR 200G InfiniBand solutions represents the industry’s most advanced interconnect solution for HPC and deep learning performance and scalability.  Mellanox is the first company to enable 200G data speeds, doubling the previous data rate and expanding In-Network Computing capabilities to accommodate the larger message sizes typically found in deep learning application workloads.  Utilizing Mellanox technology, the world’s most data-intensive applications and popular frameworks are leveraging Mellanox to accelerate the performance of  their applications and frameworks:  Yahoo has demonstrated 18X speedup for image recognition; Tencent has been able to achieve world record performance for data sorting; and NVIDIA has incorporated Mellanox solutions inside their deep learning DGX-1 appliance in order to provide 400Gb/s data throughput, and to build one of the most power-efficient machine learning supercomputers.

In-Network Computing and Security Offloads

Machine learning applications are based on training deep neural networks, which require complex computations and fast and efficient data delivery. Besides doubling the speed and providing the higher radix switch, Mellanox’s new HDR 200G switch and adapter hardware supports in-networking computing (application offload capability) and in-network memory.  Mellanox’s HDR InfiniBand solution offers offloads beyond that of RDMA and GPUDirect to computation for higher level communication framework collectives. This dramatically improves neural network training performance and overall machine learning applications, while saving on CPU cycles and increasing the efficiency of the network.

New data-centric paradigm enables in-network computing

Optimized Switching – from 100G to 200G

To show the improved performance in bandwidth, from 100G to 200G, we can compare Mellanox’s edge and chassis InfiniBand switch offerings:

Delivers more than twice the performance in the same QSFP package and 1RU edge switch

High Performance Adapters

To complete its end-to-end HDR solution, Mellanox ConnectX-6 delivers HDR 200G throughput with 200 million messages per second at under 600 nanoseconds of latency for both InfiniBand and Ethernet. Backward compatible, ConnectX-6 also supports HDR100, EDR, FDR, QDR, DDR and SDR InfiniBand as well as 200, 100, 50, 40, 25, and 10G Ethernet speeds.

ConnectX-6 also offers improvements in Mellanox’s Multi-Host® technology, allowing for up to eight hosts to be connected to a single adapter by segmenting the PCIe interface into multiple and independent interfaces. This leads to a variety of new rack design alternatives, lowering the total cost of ownership in the data center by reducing CAPEX (cables, NICs, and switch port expenses), and OPEX (cutting down on switch port management and overall power usage). ConnectX-6 200G InfiniBand and Ethernet (VPI) Network Adapter Storage customers will benefit from ConnectX-6’s embedded 16-lane PCIe switch, which allows them to create standalone appliances in which the adapter is directly connected to the SSDs. By leveraging ConnectX-6 PCIe Gen3/Gen4 capability, customers can build large, efficient high speed storage appliances with NVMe devices.

Rounding out Mellanox’s HDR 200G InfiniBand portfolio is its line of LinkX cables. Mellanox offers direct-attach 200G copper cables reaching up to 3 meters and 2 x 100G splitter breakout cables to enable HDR100 links, as well as 200G active optical cables that reach up to 100 meters. All LinkX cables in the 200G line come in standard QSFP packages. Furthermore, the optical cables provide the world’s first Silicon Photonics engine to support 50Gb/s lanes, with clean optical eyes even at such high speeds.

To see how Mellanox fares against the competition, we can compare the benefits of Mellanox’s HDR100  versus  Omni-Path .

The higher port radix enables Mellanox to build a similar cluster size as Intel, using almost half the amount of switches and cables. Lowering TCO and increasing ROI, the Mellanox HDR solution reduces power consumption, thus lowering OPEX as well as reducing rack space and thus, CAPEX. Moreover, the higher density solution delivers performance improvements with a 50% reduction in latency, as well as a simpler interconnect solution that requires fewer cable connections and has HDR fat uplinks which are more robust for congestion and microbursts.

HDR100 Savings vs. Competition:
3X Real Estate Saving, 4X Cable Saving, 2X Power & Latency Saving
HDR100  Savings of Mellanox vs. the Competition:
1.6X Switch Saving, 2X Cable Saving, Wider Uplink Pipes

Empowering Next-Generation Data Centers

As the requirement for intensive data analytics increases, there is a corresponding demand for higher bandwidth. Even 100Gb/s is insufficient for some of today’s most demanding data centers and clusters. Moreover, the traditional CPU-centric approach to networking has proven to be too inefficient for such complex applications. The Mellanox HDR 200G solution address these issues by providing the world’s first 200Gb switches, adapters, and cables and software, and by enabling In-Network Computing to handle data throughout the network instead of exclusively in the CPU. With its 200G solution, Mellanox continues to push the industry toward Exascale computing and remains a generation ahead of the competition.

The advantages of leveraging an all-Mellanox ecosystem range from inter-product compatibility of switches, adapters and cables to simpler integration and implementation with the customer’s pre-existing systems, and streamlined training sessions for IT staff.

About Mellanox

With over 18 years of experience designing high-speed communication fabrics, Mellanox is a leading supplier of end-to-end intelligent interconnect solutions and services for a wide range of markets including high performance computing, machine learning, enterprise data centers, Web 2.0, cloud, storage, network security, telecom and financial services. Mellanox adapters, switches, cables and software implement the world’s fastest and most robust InfiniBand and Ethernet networking solutions for a complete, high-performance machine learning infrastructure. These capabilities ensure optimum application performance.

For more information, please visit: http://www.mellanox.com/solutions/hpc/.

 

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire