High-Performance Computing News Analysis | insideHPC

Today the University of Michigan announced their new $4.8 million Great Lakes computing cluster. A Dell EMC system, the new supercomputer will serve researchers across campus. "Great Lakes will also be the first cluster in the world to use the Mellanox HDR 200 gigabit per second InfiniBand networking solution, enabling faster data transfer speeds and increased application performance."The post Great Lakes Supercomputer from Dell EMC Coming to University of Michigan appeared first on insideHPC.

Today MIT announced a new $1 billion commitment to address the global opportunities and challenges presented by the prevalence of computing and the rise of artificial intelligence. The initiative marks the single largest investment in computing and AI by an American academic institution, and will help position the United States to lead the world in preparing for the rapid evolution of computing and AI. " As we look to the future, we must utilize these important technologies to shape our world for the better and harness their power as a force for social good."The post Video: MIT Makes Billion-Dollar Bet on Ai and Machine Learning appeared first on insideHPC.

After three years of working on key algorithmic challenges in CFD, the European ExaFLOW Project it touting a series of industry milestones. With three flagship runs, ExaFLOW has managed to work on different specific CFD use cases which highlight the importance of their outcomes for both industry and academia. "After three years of working on key algorithmic challenges in CFD, the European ExaFLOW Project it touting a series of industry milestones. With three flagship runs, ExaFLOW has managed to work on different specific CFD use cases which highlight the importance of their outcomes for both industry and academia."The post ExaFLOW Project takes on High-order CFD appeared first on insideHPC.

This guest article from Dr. Bhushan Desam, Lenovo’s Director, Global Artificial Intelligence Business covers how new HPC tools like Lenovo's LiCO (Lenovo Intelligent Computing Orchestration) are working to address the growing popularity of AI in HPC and to simplify the convergence of HPC and AI.The post The Rising AI Tide in HPC – Are You Ready? appeared first on insideHPC.

Elsevier publishers have launched their Call for Papers on their Special Issue on Advances on Parallel and High Performance Computing for AI Applications. "Clusters of computers and accelerators (e.g. GPUs) are routinely used to train and run models, both in research and production. On the other hand, ML and AI have also become a "killer application" for HPC and, consequently, have driven much of research in this area. For example, tailored computer architecture has been devised and new parallel programming frameworks developed to accelerate AI/ML models. The objective of this special issue is to bring together the HPC and AI/ML communities to present their applications and solutions to performance issues, and also to present how AI/ML can be used to solve HPC problems."The post Call for Papers: Special Issue on Advanced Parallel HPC for AI Applications appeared first on insideHPC.

Mark Dawson gave this talk at the Supercomputing Wales Swansea Symposium 2018. "Supercomputing Wales is funded to provide university research teams access to powerful computing facilities to undertake high-profile science and innovation projects within the consortium universities. Software Engineers across Wales to develop algorithms and customized software that harnesses the power of the facilities. The aim of the investment is to capture more external research funding, increase scientific partnerships, create highly-skilled research jobs and support collaborations with industrial and other partners. This will provide a step change in supercomputing-enabled scientific research in Wales."The post Video: Graphical User Interfaces for HPC appeared first on insideHPC.

Today ECMWF launched the ESCAPE-2 project on energy-efficient scalable algorithms for weather and climate prediction at exascale. "It brings together 12 partners, including national meteorological and hydrological services, HPC centers, hardware vendors and universities. The ESCAPE project aims to prepare NWP and climate models for new computing architectures towards exascale computing, with a focus on energy efficiency."The post ESCAPE-2 Project to develop algorithms for weather and climate prediction at exascale appeared first on insideHPC.

Today the Technion technological institute in Israel announced that Intel is collaborating with the institute on its new artificial intelligence (AI) research center. "AI is not a one-size-fits-all approach, and Intel has been working closely with a range of industry leaders to deploy AI capabilities and create new experiences. Our collaboration with Technion not only reinforces Intel Israel’s AI operations, but we are also seeing advancements to the field of AI from the joint research that is under way and in the pipeline.”The post Intel Powers New AI Research Center at Technion in Israel appeared first on insideHPC.

A team of theorists from Lawrence Livermore National Laboratory (LLNL) has solved a long-standing puzzle in the nucleation of a high-pressure phase of ice known as ice VII, which is believed to exist near the core of “ocean world” planets recently detected outside of the solar system, and has recently been discovered to exist within the Earth’s mantle. "By dissecting the thermodynamics and kinetics of interfaces, there are entirely new classes of problems that can be studied and, ultimately, controlled. A holy grail is to design self-regulating dynamic systems and machines that can utilize far-from-equilibrium dissipative dynamics to perform complex tasks, as in biological systems -- control of nucleation is a step on this path.”The post New Paper Looks at Solidification of High-pressure Ice in newly discovered Ocean Worlds appeared first on insideHPC.

Today Australia-based DownUnder GeoSolutions (DUG) announced a move to Skybox Datacenters in Houston for its global expansion of a revolutionary oil and gas exploration technology. "This was an exhaustive world-wide search for a data center location,” said Dr. Matthew Lamont, co-founder of DUG. “Houston was a natural choice given the low cost of power and the fact that Skybox had the available infrastructure ready to go. This facility will allow us to install the fastest supercomputer in the world at this time to meet the ever-increasing demand for energy. We are excited to expand our presence in Houston and expect to be operational by February 2019.”The post DownUnder GeoSolutions Moves to Skybox Datacenters for Oil & Gas Exploration appeared first on insideHPC.

As the trend toward exascale HPC systems continues, the complexities of optimizing parallel applications running on them increase too. Potential performance limitations can occur at the application level which relies on the MPI. While small-scale HPC systems are more forgiving of tiny MPI latencies, large systems running at scale prove much more sensitive. Small inefficiencies can snowball into significant lag.The post Video: The March to Exascale appeared first on insideHPC.

D.E. Shaw Research has multiple positions open for HPC System Administrators in our Job of the Week. "Ideal candidates should have strong fundamental knowledge of Linux concepts such as file systems, networking, and processes in addition to practical experience administering Linux systems. Relevant areas of expertise might include large-installation systems administration experience and strong programming and scripting ability, but specific knowledge of and level of experience in any of these areas is less critical than exceptional intellectual ability."The post Job of the Week: HPC System Administrators at D.E. Shaw Research appeared first on insideHPC.

Kristin Kolodge from J.D. Power and Associates gave this talk at the HPC User Forum in Detroit. "With the rapid emergence of new technology in the automotive industry, understanding driver behavior and preferences within the vehicle is more crucial than ever for automakers. For 50 years, J.D. Power has been providing rich automotive consumer insights. Now, J.D. Power will work with HARMAN to go beyond measuring consumer insights to help automakers around the world better measure and understand how their evolving products are satisfying customer needs."The post Video: Bridging the Automated Vehicle Gap: Consumer Trust, Technology and Liability appeared first on insideHPC.

Sharan Kalwani writes that that supercomputers a critical tool for predicting severe weather that could threaten entire regions with famine. "Models have been improved - although at too slow a pace for the general public at large – that now this HPC approach is taking root in more than just the monsoon predictions. Tropical storms are also important and cyclones are also quite rampant. Recently two Petaflop-capable machines have been installed in a special tropical weather department in the central part of the country."The post Supercomputing to Feed the World appeared first on insideHPC.

Today the HPC-AI Advisory Council Call for Team Entries for the 2019 Student Cluster Competition in Frankfurt. To be held at ISC 2019, the competition seeks student teams currently enrolled in four-year higher education and undergraduate-degree programs. “Students come away from competitions steeped in experience gained from deploying tools and technologies they’ve developed to launch, troubleshoot and run live applications, to solve complex problems and achieve very real-world results.”The post Enter Your Team Now in the ISC 2019 Student Cluster Competition appeared first on insideHPC.

Are you coming to SC18 in Dallas next month? Now is your last chance to enjoy Early-Bird registration discount rates. "Are you an ACM, IEEE, or SIGHPC member? If not, it is something that you should consider. With a membership in any of these societies you will receive a substantial discount on your SC conference registration. For example, members can save up to $400 over non-members by registering on or before October 14, 2018 for the Technical Program."The post Hurry! Register Now for SC18 to Enjoy Early-Bird Rates appeared first on insideHPC.

Today Atos announced a joint effort with E4, the Jülich Supecomputing Centre, Fraunhofer FOKUS, Huawei, Mellanox, and SUSE to help shorten time-to-market for HPC and Edge deployments. "Collaborative and grassroots initiatives focused on solving real world problems are the hallmark of a vibrant ecosystem," said Jean-Marc Denis, Head of Strategy, BigData and Security at Atos. "Addressing the challenge of supporting one trillion connected devices requires today’s infrastructure to evolve. Only the Arm ecosystem has the flexibility, scalability, and technical breadth to bring best-in-class solutions to market in areas as diverse as edge and high-performance compute.”The post Atos Launches Open Edge and HPC Initiative for the coming 5G World appeared first on insideHPC.

Today ARM server vendor Ampere introduced a new developer platform and program to facilitate the open development of next-generation cloud applications. Available through Lenovo, the new server platform with Ampere’s eMAG processors is available now for purchase and developers can order by visiting Ampere’s new developer web site. "By providing access to a development environment with open standards and full open documentation, we can expand the ecosystem and optimize the developer experience to innovate on cloud applications not yet imagined.”The post Ampere Augments ARM Servers with New Developer Platform appeared first on insideHPC.

Today Oracle announced support for the NVIDIA HGX-2 platform on Oracle Cloud Infrastructure. Designed to meet the needs of the next generation of analytics, machine learning, and AI, Oracle is offering GPU-accelerated deep learning and HPC containers from the NVIDIA GPU Cloud container registry. “Whether you are an engineer, data scientist, researcher or developer, we are bringing the power of compute and cloud to your fingertips.The post NVIDIA HGX-2 Ai Supercomputer Comes to Oracle Cloud Infrastructure appeared first on insideHPC.

Over at the University of Delaware, Julie Stewart writes that assistant professor Sunita Chandrasekaran has received an NSF grant to develop frameworks to adapt code for GPU supercomputers. She is working with complex patterns known as wavefronts, which are commonly found in scientific codes used in analyzing the flow of neutrons in a nuclear reactor, extracting patterns from biomedical data or predicting atmospheric patterns.The post Sunita Chandrasekaran Receives NSF Grant to Create Powerful Software Framework appeared first on insideHPC.

Today the University of Reims Champagne-Ardenne dedicated their new ROMEO supercomputer. A BullSequana X1000 model from Atos, ROMEO is ranked amongst the most-powerful supercomputers worldwide. "This installation confirms the strategic engagement of the Atos Group in R&D in scientific computing. The BullSequana supercomputer brings a level of power that is needed by the digital simulation projects of the University and it also offers the necessary functionalities for the convergence of Artificial Intelligence and scientific computing, which the URCA requested.”The post 1 Petaflop ROMEO Supercomputer Comes to Reims from Atos appeared first on insideHPC.

Today NVIDIA announced a GPU-acceleration platform for data science and machine learning, with broad adoption from industry leaders, that enables even the largest companies to analyze massive amounts of data and make accurate business predictions at unprecedented speed. "It integrates seamlessly into the world’s most popular data science libraries and workflows to speed up machine learning. We are turbocharging machine learning like we have done with deep learning,” he said.The post Open Source RAPIDS GPU Platform to Accelerate Predictive Data Analytics appeared first on insideHPC.

Today Hyperscale Storage Processing vendor Nyriad announced an exclusive agreement with Datacom to develop and deploy a next-generation computational storage solution for secure, collaborative processing of data in the Australian market. "Our next-generation data fabric harnesses the power of the GPU to deliver breakthrough storage performance. And that’s where things start to get interesting. The massive processing capacity of the GPU allows us to run artificial intelligence and machine learning models directly on the storage, as well as advanced cryptographic encryption with a blockchain audit ledger in real-time,” said Todd Papaioannou of Nyriad.The post Nyriad to Power Blockchain-Enabled, AI-Ready Storage at Datacom appeared first on insideHPC.

Thomas Schulthess from CSCS gave this talk at the Blue Waters Summit. "Application performance is given much emphasis in discussions of exascale computing. A 50-fold increase in sustained performance over today's applications running on multi-petaflops supercomputing platforms should be the expected target for exascale systems deployed early next decade. We will reflect on what this means in practice and how much these exascale systems will advance the state of the art."The post Video: Reflecting on the Goal and Baseline of Exascale Computing appeared first on insideHPC.

Today Verne Global announced that the Centro Epson Meteo meteorological forecasting organizations will start using the company's HPC optimized bare-metal infrastructure to accelerate and improve the accuracy of its forecasting and data analytics capabilities. "Verne Global provides us with the peace of mind that we have our own dedicated infrastructure that’s optimized to meet our specific needs and backed-up by a team of real HPC experts, together with the flexibility and scalability that’s usually only associated with cloud-based services."The post Centro Epson Meteo Moves HPC Weather Forecasting to Verne Global in Iceland appeared first on insideHPC.

Hyperion Research will host their annual Breakfast Briefing at SC18 in Dallas. The event takes place at the Adolphus Hotel at 7:15 am - 9:00 am on Tuesday, Nov. 13. Breakfast will be served at 6:45 am. "We will be presenting an update on HPC market results, key technology issues & trends, results of new research studies in AI & Big data, new ROI results, and the mapping of HPC centers across the US."The post Hyperion Research Breakfast Briefing Coming to Dallas for SC18 appeared first on insideHPC.

Tuning code has, for a long time, been an art. Knowing what to look for and how to correct inefficiencies in serious numerical computations has not been easy for most programmers. It’s often hard to even know which tool to start with. Which is why the Intel® VTune™ Amplifier Application Performance Snapshot could prove to be a great way to get an instant summary of an application’s performance characteristics and issues.The post Learn What to Do Next with Intel VTune Amplifier Application Performance Snapshot appeared first on insideHPC.

Sharan Kalwani writes that that one of the main reasons he got into computing long ago was the potential that he saw in using powerful news supercomputing tools towards addressing the needs of the society we live in. "The point here is that technically the challenges are very tractable, however society needs to also grow at a comparable, if not higher pace than the technology (which is evolving faster than us)."The post How HPC can Benefit Society appeared first on insideHPC.

Over at EPCC, Professor Peter Boyle describes the new Tesseract supercomputer, which is hosted at EPCC's Advanced Computing Facility. DiRAC is part of the UK’s e-infrastructure investment to improve computing, software development, data storage, networking and distributed computing networks. "Tesseract is an 844-node HPE SGI 8600 supercomputer. Each node has two Xeon scalable processor 4116 “Silver” nodes, and 96 GB of memory. In total, the new system has 20256 Intel Skylake computing cores, and is interconnected using 100 Gbit/s Intel Omnipath interconnect."The post HPE Powers New Tesseract Supercomputer at EPCC appeared first on insideHPC.

The TACC Frontera supercomputer, which will become the fastest academic supercomputer in the United States when it becomes operational in 2019, will rely on CPUs for visualization. "The ability to run and visualize anywhere with SDVis solutions, regardless of the scale of the visualization task and without requiring specialized hardware for interactive response, is the reason HPC centers no longer need to procure GPUs for visualization clusters."The post Cloud and Enterprise Adopting HPC-proven CPU-based Interactive, Photorealistic Rendering appeared first on insideHPC.

Today Sylabs announced that it has released a new version of its innovative container software: Singularity 3.0. With new enterprise-class features, Singularity is now the premier container runtime solution, enabling your company to seamlessly and efficiently tackle today’s most demanding AI, machine learning, and advanced analytic workloads. "Lenovo and Sylabs have collaborated for over two years around a shared vision of producing lightweight HPC containers that are flexible, secure, and reproducible, while also delivering performance that is equal to native OS performance."The post Sylabs Tunes Singularity 3.0 Containers for Machine Learning appeared first on insideHPC.

In this video, the SCinet team prepares the world's fastest show network for SC18. "With this improved infrastructure in place, SCinet will deliver an anticipated 4.12 terabits per second of network capacity to a million square feet of exhibit space for SC18 in November. When the conference ends, all that SCinet-facilitated fiber will remain in place for future use by Dallas and its millions of visitors."The post Video: SCinet Prepares World’s Fastest Network for SC18 appeared first on insideHPC.

Over at Argonne, Nils Heinonen writes that four new projects for the ALCF Data Science Program that will utilize machine learning, deep learning, and other artificial intelligence methods to enable data-driven discoveries across scientific disciplines. "Each project intends to implement novel machine learning techniques; some will integrate these methods with simulations and experiments, while others will pioneer uncertainty quantification and visualization to aid in the interpretation of deep neural networks.”The post Data Science Program at Argonne Looks to Machine Learning for New Breakthroughs appeared first on insideHPC.

"Using a climate dataset from Berkeley Lab on the Summit supercomputer at Oak Ridge, they trained a deep neural network to identify extreme weather patterns from high-resolution climate simulations. By tapping into the specialized NVIDIA Tensor Cores built into the GPUs at scale, the researchers achieved a peak performance of 1.13 exaops and a sustained performance of 0.999 – the fastest deep learning algorithm reported to date and an achievement that earned them a spot on this year’s list of finalists for the Gordon Bell Prize."The post Summit Supercomputer Breaks Exaop Barrier on Neural Net Trained to Recognize Extreme Weather Patterns appeared first on insideHPC.

The American University of Sharjah (AUS) is seeking an HPC System Administrator in our Job of the Week. "American University of Sharjah (AUS) invites applications for an HPC System Administrator position commencing in the Spring 2019 semester at the latest. AUS is seeking a highly motivated individual with proven experience in managing software and hardware HPC installations for research applications, with emphasis on utilizing open source tools and platform.The post Job of the Week: HPC System Administrator at AUS appeared first on insideHPC.

OARnet recently joined several other regional and federal network organizations on a three-year, $3.5 million grant to help scientists more efficiently work with massive datasets that have become essential to modern scientific discovery. "Throughout Ohio, and across the nation, we are seeing researchers transmitting an ever-increasing volume of larger and larger datasets over our networks,” said Paul Schopis, executive director of OARnet. “These vast stores of information are laying the foundations for improvements in health and prosperity. As research and education networks serving these educators, scientists and engineers, we have to develop the techniques, infrastructure and policies that will fully support their information-sharing needs.”The post OARnet in Ohio to Help Researchers Accelerate Big Data Sharing appeared first on insideHPC.

In this special guest feature, SC18 Conference Chair Ralph A. McEldowney writes that HPC is now more crucial than ever to empower science for the benefit of humanity. "From hurricane and earthquake predictions to solving global hunger challenges – the importance of HPC has never been greater. Join us in Dallas, Texas for the 30th Anniversary of the SC Conference Series, Nov 11-16, 2018, and be inspired to make a difference."The post Why SC18 is a Key Event for Tackling our Planet’s Toughest Challenges appeared first on insideHPC.

Today D-Wave Systems announced the immediate availability of free, real-time access to the D‑Wave Leap Quantum Application Environment. Leap is the first cloud-based QAE providing real-time access to a live quantum computer. "The next frontier of quantum computing is quantum application development. While we continue to advance our industry-leading quantum technology, our goal with Leap is to ignite a new generation of developers who will explore, experiment, and ultimately build our quantum application future."The post Video: D‑Wave Rolls Out Leap, A Quantum Application Development Environment appeared first on insideHPC.

Today AIC announced a world-record in data transfer: one petabyte in 29 hours encrypted data transfer, with data integrity checksum unconditionally enabled, over a distance of 5000 miles. The average transfer rate is 75Gbps, or 94% utilization of the available bandwidth of 80Gbps. "Even with massive amounts of data, this test confirmed once more that it's completely feasible to carry out long distance, fully encrypted and checksum-ed data transfer at nearly the line-rate, over a shared and production network."The post Big Data over Big Distance: Zettar Moves a Petabyte over 5000 Miles in 29 Hours appeared first on insideHPC.

Kim McMahon has joined the Cloud Foundry Foundation as the Senior Director of Digital Strategy. In this new role, she will collaborate with the open source community and their marketing teams to promote adoption and awareness of Cloud Foundry technologies.The post Kim McMahon joins Cloud Foundry as Senior Director of Digital Strategy appeared first on insideHPC.

Researchers from Stanford University have developed a deep learning based system that can automatically detect Alzheimer’s disease and its biomarkers from MRIs, with 94 percent accuracy. "Our method uses minimal preprocessing of MRIs (imposing minimum preprocessing artifacts) and utilizes a simple data augmentation strategy of downsampled MR images for training purposes,” the researchers stated in their paper.The post Using Ai to Automatically Diagnose Alzheimer’s Disease appeared first on insideHPC.

ISC 2019 has issued their Call for Submissions for tutorials and workshop proposals. If you have expertise in a particular area of high performance computing and are interested in sharing it with your like-minded peers, the conference folks would like to hear from you. "The tutorials attract over 300 attendees, while the workshops are attended by close to 700 people. Overall attendance for the 2019 conference is expected to exceed 3,500."The post Call for Submissions: ISC 2019 Tutorials and Workshops appeared first on insideHPC.

The Department of Energy today announced the first round of awardees for the new HPC4Materials (HPC4Mtls) Program. "We are excited about adding a materials portfolio to the HPC4EnergyInnovation umbrella program,” said acting HPC4Materials Director Robin Miles. “Understanding material properties at the macro level based on atomic scale effects is a sweet spot for HPC capabilities. New materials are essential for improved energy efficiency in electrical power plants and advanced vehicles.”The post HPC4Materials Program Funds Nine Projects to Fight Hot Corrosion Degradation appeared first on insideHPC.

David E. Martin gave this talk at the HPC User Forum. "In 2021, the Argonne Leadership Computing Facility (ALCF) will deploy Aurora, a new Intel-Cray system. Aurora, will be capable of over 1 exaflops. It is expected to have over 50,000 nodes and over 5 petabytes of total memory, including high bandwidth memory. The Aurora architecture will enable scientific discoveries using simulation, data and learning."The post ALCF – The March toward Exascale appeared first on insideHPC.

Today FPGA maker Xilinx unveiled Versal, "the industry's first adaptive compute acceleration platform (ACAP)". The company also announced new Alveo FPGA cards, which the company claims can deliver "4X the performance of GPUs, 90X the performance of CPUs, plus unprecedented adaptability across workloads." AMD, one of the Xilinx partners that is showcasing products based on the new Alveo boards, announced a server that will set a new world record for real-time AI inference processing, with a mind-boggling 30,000-images-per-second inference throughput.The post Xilinx Steps Up with Alveo FPGA boards and Versal Adaptive Compute Acceleration Platform appeared first on insideHPC.

Today SQream announced that its GPU-powered SQream DB software now includes optimized support for the IBM POWER9 multi-core architecture. "GPU-accelerated analytics are an increasingly important part of our industry,” said Sumit Gupta, VP of HPC and AI for IBM Cognitive Systems. “The announcement of SQream on the IBM POWER9 platform takes this concept to another level of performance, as the POWER9 CPU with embedded NVIDIA NVLink interface to NVIDIA’s GPUs allows SQream to enable even faster processing of data on POWER9 servers.”The post GPU-powered SQream DB Boosts Query Performance up to 150% for IBM POWER9 appeared first on insideHPC.

Today Lenovo announced that it is one of the first server vendors certified for SAP HANA for hyperconverged infrastructure. "With Lenovo’s ThinkAgile HX7820 four-socket system certified for SAP HANA, enterprises can now take advantage of SAP HANA workloads on a hyperconverged infrastructure that moves with agility for the business of your enterprise. With the speed of in-memory databases, you can gain faster access to integrated data analytics and machine learning, while scaling performance for storage, compute and networking."The post Lenovo ThinkAgile System Certified for SAP HANA on Hyperconverged Infrastructure appeared first on insideHPC.

Today STULZ announced a new strategic partnership with Cloud&Heat Technologies, a specialist in energy-efficient, scalable and reliable data centers. As part of this new technical and sales alliance, the hot water direct cooling system developed by Dresden-based Cloud&Heat is to be marketed in future as an equipment option for the STULZ MicroDC series. "In addition to IT cooling, efficient heat recovery is at the heart of Cloud&Heat’s solution. Its patented technology enables the company to bring the heat produced by the IT equipment to a constant temperature of 60°C degrees and exports it for heating buildings and water. This technique not only lowers the cost of cooling the data center, but also of heating and hot water. This concept therefore helps to massively reduce the CO2 footprint of IT systems."The post STULZ and Cloud&Heat team up with Watercooled Micro Data Center appeared first on insideHPC.

In this video from the HPC User Forum in Detroit, Earl Joseph from Hyperion Research moderates a panel discussion on Innovation in HPC. "Hyperion Research sponsors the annual HPC and ROI Innovation Awards to recognize noteworthy achievements made by users using High Performance Computing technologies. While there are multiple benchmarks to measure the performance of technical computers, there currently isn't an adequate methodology to evaluate the economic and scientific value HPC systems contribute. The HPC Innovation Excellence Award Program is designed to help close that gap."The post Video: Panel Discussion on Innovation in HPC appeared first on insideHPC.

NVIDIA has one-upped performance breakthroughs announced earlier this year, thanks to two new technologies. Related to data augmentation and tensor parameter interleaving, these technologies enabled a single Tesla V100 GPU to do a ResNet-50 training run in just under 24 hours (1,350 images/second). "We also discussed how these same technologies enabled a single DGX-1 server node to train ResNet in just over four hours (7,850 images/second). Since then, we put our 16-GPU DGX-2 server to the test, and using these same enabling technologies, DGX-2 finished the training run in 108 minutes (just under two hours)!"The post NVIDIA DGX-2 AI Supercomputer Sets Records for Training and Inference appeared first on insideHPC.