HYPER-CONVERGED INFRASTRUCTURE

Inspur Announces MLPerf v2.0 Results for AI Servers

Inspur | July 04, 2022 | Read time : 3 min

Inspur
The open engineering consortium MLCommons released the latest MLPerf Training v2.0 results, with Inspur AI servers leading in closed division single-node performance.

MLPerf is the world’s most influential benchmark for AI performance. It is managed by MLCommons, with members from more than 50 global leading AI companies and top academic institutions, including Inspur Information, Google, Facebook, NVIDIA, Intel, Harvard University, Stanford University, and the University of California, Berkeley. MLPerf AI Training benchmarks are held twice a year to track improvements in computing performance and provide authoritative data guidance for users.

The latest MLPerf Training v2.0 attracted 21 global manufacturers and research institutions, including Inspur Information, Google, NVIDIA, Baidu, Intel-Habana, and Graphcore. There were 264 submissions, a 50% increase over the previous round. The eight AI benchmarks cover current mainstream usage AI scenarios, including image classification with ResNet, medical image segmentation with 3D U-Net, light-weight object detection with RetinaNet, heavy-weight object detection with Mask R-CNN, speech recognition with RNN-T, natural language processing with BERT, recommendation with DLRM, and reinforcement learning with MiniGo.

Among the closed division benchmarks for single-node systems, Inspur Information with its high-end AI servers was the top performer in natural language processing with BERT, recommendation with DLRM, and speech recognition with RNN-T. It won the most titles among single-node system submitters. For mainstream high-end AI servers equipped with eight NVIDIA A100 Tensor Core GPUs, Inspur Information AI servers were top ranked in five tasks (BERT, DLRM, RNN-T, ResNet and Mask R-CNN).

Continuing to lead in AI computing performance

Inspur AI servers continue to achieve AI performance breakthroughs through comprehensive software and hardware optimization. Compared to the MLPerf v0.5 results in 2018, Inspur AI servers showed significant performance improvements of up to 789% for typical 8-GPU server models.

The leading performance of Inspur AI servers in MLPerf is a result of its outstanding design innovation and full-stack optimization capabilities for AI. Focusing on the bottleneck of intensive I/O transmission in AI training, the PCIe retimer-free design of Inspur AI servers allows for high-speed interconnection between CPUs and GPUs for reduced communication delays. For high-load, multi-GPU collaborative task scheduling, data transmission between NUMA nodes and GPUs is optimized to ensure that data I/O in training tasks is at the highest performance state. In terms of heat dissipation, Inspur Information takes the lead in deploying eight 500W high-end NVIDIA Tensor Core A100 GPUs in a 4U space, and supports air cooling and liquid cooling. Meanwhile, Inspur AI servers continue to optimize pre-training data processing performance, and adopt combined optimization strategies such as hyperparameter and NCCL parameter, as well as the many enhancements provided by the NVIDIA AI software stack, to maximize AI model training performance.

Greatly improving Transformer training performance

Pre-trained massive models based on the Transformer neural network architecture have led to the development of a new generation of AI algorithms. The BERT model in the MLPerf benchmarks is based on the Transformer architecture. Transformer’s concise and stackable architecture makes the training of massive models with huge parameters possible. This has led to a huge improvement in large model algorithms, but necessitates higher requirements for processing performance, communication interconnection, I/O performance, parallel extensions, topology and heat dissipation for AI systems.

In the BERT benchmark, Inspur AI servers further improved BERT training performance by using methods including optimizing data preprocessing, improving dense parameter communication between NVIDIA GPUs and automatic optimization of hyperparameters, etc. Inspur Information AI servers can complete BERT model training of approximately 330 million parameters in just 15.869 minutes using 2,850,176 pieces of data from the Wikipedia data set, a performance improvement of 309% compared to the top performance of 49.01 minutes in Training v0.7. To this point, Inspur AI servers have won the MLPerf Training BERT benchmark for the third consecutive time.

Inspur Information’s two AI servers with top scores in MLPerf Training v2.0 are NF5488A5 and NF5688M6. The NF5488A5 is one of the first servers in the world to support eight NVIDIA A100 Tensor Core GPUs with NVIDIA NVLink technology and two AMD Milan CPUs in a 4U space. It supports both liquid cooling and air cooling. It has won a total of 40 MLPerf titles. NF5688M6 is a scalable AI server designed for large-scale data center optimization. It supports eight NVIDIA A100 Tensor Core GPUs and two Intel Ice Lake CPUs, up to 13 PCIe Gen4 IO, and has won a total of 25 MLPerf titles.

About Inspur Information
Inspur Information is a leading provider of data center infrastructure, cloud computing, and AI solutions. It is the world’s 2nd largest server manufacturer. Through engineering and innovation, Inspur Information delivers cutting-edge computing hardware design and extensive product offerings to address important technology sectors such as open computing, cloud data center, AI, and deep learning. Performance-optimized and purpose-built, our world-class solutions empower customers to tackle specific workloads and real-world challenges.

Spotlight

You know it; we know it. Evaluating the cost of software requires consideration beyond license fees. In a sea of backlogs and to-dos, making thoughtful decisions generates more work. Let’s strike one more thing off your to-do list by evaluating the critical decision points to select performance testing software. Consider this fi


Other News
APPLICATION STORAGE, DATA STORAGE, IT SYSTEMS MANAGEMENT

Fivetran and Oldcastle Infrastructure Win Ventana Research’s 2022 Digital Leadership Award

Fivetran | November 02, 2022

Fivetran, the global leader in modern data integration, has won Ventana Research’s 2022 Digital Leadership Award in the category of data for driving business transformation with Oldcastle Infrastructure, the leading building materials business in the world. This award recognizes the organization and technology that best exemplifies leadership in big data and related technologies for supporting data and information management-related needs. Oldcastle Infrastructure, a CRH company, is an industry leader in engineered building solutions offering more than 16,000 pipe, precast, stormwater, enclosure and building accessory products. To utilize data more effectively and generate comprehensive business insights, Oldcastle selected Fivetran as its data integration partner. ROI for the system in the last 12 months is estimated to be multiple millions of dollars in terms of cost savings, increased margins, and improved sales. “Oldcastle is thrilled to be recognized by Ventana for our work with Fivetran,” said Nick Heigerick, Head of Advanced Analytics at Oldcastle Infrastructure. “The use of Fivetran changed the way Oldcastle Infrastructure thinks about data – it’s an asset, not just a by-product of a process. By delivering cross-company metrics across the entire organization, we have been able to realize multimillion-dollar ROI year over year. We could not be happier with our partnership with Fivetran and the results, and are honored to be recognized for our efforts.” The Ventana Research Digital Leadership awards spotlight the individuals and organizations that have utilized and championed modernization and transformation across their people, processes, information, and technology to grow their business and industry market potential. As part of the grading process and methodology, the Ventana awards team examined case studies and submissions to evaluate the organization's use of people, processes, information, and technology to the impact and performance that resulted from the use of technology. The examination of best practices and methods used, the degree of team involvement, and the project's business impact and value. This year’s nominations crossed many industries and different-sized organizations. “Congratulations to Nick Heigerick and Oldcastle Infrastructure using Fivetran for receiving the 15th annual Ventana Research Digital Leadership Award in data,” said Matt Aslett, VP & Research Director at Ventana Research. “Turning data into business insight is a key element of any digital transformation initiative, helping to deliver efficiency improvements and competitive differentiation. Oldcastle’s use of Fivetran illustrates that investment in new data and analytics technology can have a profound impact on overall business strategy, overcoming entrenched organizational and cultural inefficiencies to deliver time and cost savings, as well as improved margins and sales.” About Fivetran Fivetran is the global leader in modern data integration. Our mission is to make access to data as simple and reliable as electricity. Built for the cloud, Fivetran enables data teams to effortlessly centralize and transform data from hundreds of SaaS and on-prem data sources into high-performance cloud destinations. Fast-moving startups to the world’s largest companies use Fivetran to accelerate modern analytics and operational efficiency, fueling data-driven business growth. Fivetran is headquartered in Oakland, California, with offices around the world. About Ventana Research Ventana Research provides insight and expert guidance on mainstream and disruptive technologies through a unique set of research-based offerings including benchmark research and technology evaluation assessments, education workshops, and our research and advisory services, Ventana On-Demand. Our unparalleled understanding of the role of technology in optimizing business processes and performance and our best practices guidance are rooted in our rigorous research-based benchmarking of people, processes, information, and technology across business and IT functions in every industry. This benchmark research, plus our market coverage and in-depth knowledge of hundreds of technology providers, means we can deliver education and expertise to our clients and increase the value they derive from technology investments while reducing time, cost, and risk.

Read More

HYPER-CONVERGED INFRASTRUCTURE,APPLICATION INFRASTRUCTURE,STORAGE MANAGEMENT

STACK Infrastructure Breaks Ground on 100MW Data Center Campus in Northern Virginia

STACK Infrastructure | November 30, 2022

STACK Infrastructure, the digital infrastructure partner to the world’s most innovative companies and leading global developer and operator of data centers, announced the groundbreaking of STACK’s latest hyperscale campus in the center of Prince William County, one of the most desirable locations in Northern Virginia. Delivery of the first building on the campus is targeted for Q1 2024. The latest among STACK’s portfolio of seven data center campuses in Northern Virginia, the 40-acre site will add nearly 100MW of committed and scalable power from Northern Virginia Electric Cooperative (NOVEC). Construction will begin with a 36MW facility, with plans to grow the campus to multiple data centers supported by a 300MW substation. The scalable campus offers a prime opportunity for clients interested in securing capacity within this critical land and power-constrained market. “Expanding our presence in the heart of Prince William County represents a strategic approach of continuing to deliver scalable capacity where it matters most. “Powered with 100% renewable energy, STACK’s new campus offers a sustainable solution and allows our clients the ability to grow quickly in the world’s largest data center market.” Matthew VanderZanden, Chief Operating Officer of STACK Americas STACK’s presence in Northern Virginia has markedly increased with continued growth announcements over the last four years. STACK has nearly 1GW of current and under-development capacity in one of the most constrained data center markets on the globe. Plans for this latest development were announced in April, on the heels of a 216MW Ashburn campus announcement shared earlier in 2022. Over the past three months, STACK has announced growth in the top data center markets across the globe, including a 230MW five-building campus in central Phoenix, an 80MW hyperscale campus in Frankfurt, Germany, a 48MW facility in Seoul, Korea, and multiple data centers in Australia. STACK’s presence within 23 markets distributed throughout the Americas, EMEA, and APAC regions makes it one of the largest private data center operators worldwide. ABOUT STACK INFRASTRUCTURE STACK provides digital infrastructure to scale the world’s most innovative companies. With a client-first approach, STACK delivers a comprehensive suite of campus, build-to-suit, colocation, and powered shell solutions in the Americas, EMEA and APAC regions. With robust existing and flexible expansion capacity in the leading availability zones, STACK offers the scale and geographic reach that rapidly growing hyperscale and enterprise companies need. The world runs on data. And data runs on STACK.

Read More

HYPER-CONVERGED INFRASTRUCTURE, APPLICATION INFRASTRUCTURE

Qii.AI and Skydio Enter Technology Partnership to Advance AI for Automated Infrastructure Inspections

Qii.AI | September 08, 2022

Qii.AI, the provider of digital inspection software for infrastructure, announced today that it is partnering with Skydio, the leading U.S. drone manufacturer and world leader in autonomous flight, to make drone-powered inspection more efficient and effective for customers across North America. Drones are powerful and cost-effective tools for inspection, and autonomy is the key to deploying these solutions for safe and reliable operation. Skydio's adaptive scanning software, Skydio 3D Scan™, extends the company's groundbreaking autonomous flight engine with advanced artificial intelligence (AI) skills that automate photographic data collection and mapping tasks ranging from infrastructure asset inspection to crime and accident scene reconstruction. Through this integration, customers using Skydio 3D Scan™ software will now be able to utilize Qii.AI's computer-assisted detection and quantification of corrosion-related defects. This integration will make inspection of large, complex structures significantly more efficient, while reducing the time required by inspectors to identify and classify defects. "Skydio's computer vision navigation and autonomous data capture capabilities, combined with the Qii system's automatic corrosion, crack, and defect detection algorithms, are a leap forward in remote digital inspection possibilities." Qii.AI CEO, Michael H. Cohen Earlier this year, Qii.AI and Skydio demonstrated the power of this integration, enabling the automatic detection and quantification of corrosion on naval ship hulls for the Canadian Department of National Defense in Halifax, Canada. During the day-long demonstration, Skydio's drones captured data from two naval ships, using Skydio 3D Scan to capture the data used to create digital twins of both ships before seamlessly importing the models and data through the Skydio Cloud API into the Qii system for auto-detection, classification, and quantification of visible corrosion. Commenting further on the agreement, Qii.AI's Cohen, said "Skydio is the clear leader in the small inspection drone market and a natural fit for Qii.AI's computer vision technology. We're proud and excited to be collaborating with such a great team and thrilled with the success of our joint-demonstration with the Canadian Department of National Defense." To learn more about Qii.AI's corrosion detection capability and to see the results of the Qii.AI integration with Skydio 3D Scan in person, visit booth #823 at Commercial UAV Expo on September 6-8, 2022 at Caesars Forum, Las Vegas, or for more information, visit Qii.AI online at www.qii.ai. About Qii.AI Qii.AI is a web-based platform that empowers remote, collaborative inspections of critical infrastructure assets such as bridges, dams, and wind turbines. Qii.AI uses computer vision and machine learning to improve the inspection process with computer-assisted detection and quantification of corrosion, cracking, delamination, and other problems in steel and concrete structures. Qii.AI is the world's first visualization software for infrastructure inspection data that merges below the waterline (sonar) data with above-the-waterline (visual, thermal, lidar) data, to provide a single, wholistic view of your asset. About Skydio Skydio is the leading U.S. drone manufacturer and world leader in autonomous flight. Skydio leverages breakthrough AI to create the world's most intelligent flying machines for use by consumer, enterprise, and government customers. Founded in 2014, Skydio is made up of leading experts in AI, robotics, cameras, and electric vehicles from top companies, research labs, and universities from around the world. Skydio designs, assembles, and supports its products in the U.S. from its headquarters in Redwood City, CA, to offer the highest standards of supply chain and manufacturing security. Skydio is trusted by leading enterprises across a wide range of industry sectors and is backed by top investors and strategic partners including Andreessen Horowitz, Levitate Capital, Next47, IVP, Playground, and NVIDIA.

Read More

HYPER-CONVERGED INFRASTRUCTURE,APPLICATION INFRASTRUCTURE,IT SYSTEMS MANAGEMENT

Effectual Achieves Inaugural HashiCorp Infrastructure Competency

Effectual | November 09, 2022

Effectual, a modern, cloud first managed and professional services company, has been awarded the HashiCorp Infrastructure Competency as part of the newly released Partner Technical Competency Program for Systems Integrators. Effectual is one of only six companies globally to be recognized with the HashiCorp Infrastructure Competency as part of the program's launch. As a leader in multi-cloud infrastructure automation software, HashiCorp's software suite enables organizations to adopt consistent workflows and create a system of record for automating the cloud: infrastructure provisioning, security, networking, and application deployment. The new competency program gives partners with the HashiCorp Systems Integrator designation the opportunity to be recognized for their ability to deliver and integrate HashiCorp products and solutions into end customers' initiatives. Competencies are only awarded after a successful audit that assesses a range of requirements, including the number of technical staff certified on HashiCorp products and proven customer success with HashiCorp products in deployment. HashiCorp's endorsement gives customers confidence in Effectual's ability to guide, advise, assist, and execute alongside them, bringing the required experience and expertise to succeed with the advantages of HashiCorp's portfolio of automation tools. "Effectual enables customers to accelerate automation and innovation utilizing HashiCorp Terraform, Vault, and the recently released HCP Boundary. "Together we help customers drive positive business outcomes and realize greater agility, security controls, and increased operational efficiency." Robb Allen, Effectual CEO As an elite pure-play AWS services provider, Effectual is a validated AWS MSP Partner holding six AWS competencies. Effectual prides itself on bringing deep cloud and modernization experience to all customer engagements. Complementing this deep partnership with AWS, the cloud first MSP has embraced and invested in its partnership with HashiCorp. This strategic combination of industry expertise ensures that customers have the confidence to migrate and establish a strong modern foundation on which to build and deploy next-generation applications, services, and operations. About Effectual An AWS Premier Consulting Partner, Effectual is a modern, cloud first managed and professional services company that works with commercial enterprises and the public sector to enable digital transformation and full stack IT modernization. Effectual's deeply experienced and passionate team of problem solvers apply proven methodologies to enable positive business outcomes with Amazon Web Services and VMware Cloud on AWS. Effectual is a member of the Cloud Security Alliance, and the PCI Security Standards Council.

Read More

Spotlight

You know it; we know it. Evaluating the cost of software requires consideration beyond license fees. In a sea of backlogs and to-dos, making thoughtful decisions generates more work. Let’s strike one more thing off your to-do list by evaluating the critical decision points to select performance testing software. Consider this fi

Resources