Facebook’s Expanding Machine Learning Infrastructure

Here at The Next Platform, we tend to keep a close eye on how the major hyperscalers evolve their infrastructure to support massive scale and evermore complex workloads.

Not so long ago the core services were relatively standard transactions and operations, but with the addition of training and inferencing against complex deep learning models—something that requires a two-handed approach to hardware—the hyperscale hardware stack has had to quicken its step to keep pace with the new performance and efficiency demands of machine learning at scale.

While not innovating on the custom hardware side quite the same way as Google, …

Facebook’s Expanding Machine Learning Infrastructure was written by Nicole Hemsoth at The Next Platform.

HPC Optimizes Energy Exploration for Oil and Gas Startups

In its quest to meet the world’s ever-increasing demand for energy, the oil and gas industry has become one of the largest users—and leading innovators—of high-performance computing (HPC). As natural resources deplete, and the cost of accessing them increases, highly sophisticated computational modeling becomes an essential tool in energy exploration and development.

Advanced computational techniques provide a high-fidelity model of the subsurface, which gives oil and gas companies a greater understanding of the geophysics of the region they propose to explore. A clearer picture of the earth enables targeted drilling, reduced acquisition costs, and minimal environmental impact. In an industry …

HPC Optimizes Energy Exploration for Oil and Gas Startups was written by Nicole Hemsoth at The Next Platform.

GENCI: Advancing HPC in France and Across Europe

One of the more significant efforts in Europe to address the challenges of the convergence of high performance computing (HPC), high performance data analytics (HPDA) and soon artificial intelligence (AI), and ensure that researchers are equipped and familiar with the latest technology, is happening in France at GENCI (Grand équipement national de calcul intensif).

Grand équipement national de calcul intensif (GENCI) is a “civil company” (société civile) under French law and 49% owned by the State, represented by the French Ministry of Higher Education, Research and Innovation (MESRI), 20% by the Commissariat à l’Energie Atomique et aux énergies alternatives ( …

GENCI: Advancing HPC in France and Across Europe was written by Nicole Hemsoth at The Next Platform.

Creating Chaos to Save the Datacenter

Downtime has been plaguing companies for decades, and the problems have only been exacerbated during the internet era and with the rise of ecommerce and the cloud.

Systems crash, money is lost because no one is buying anything, more money is spent on the engineers and the time they need to fix the problem and get things back online. In the meantime, enterprises have to deal with frustrated customers and risk losing many of them, who lose trust the in the company and opt to move their business elsewhere. For much of that time, the response to system failures has …

Creating Chaos to Save the Datacenter was written by Nicole Hemsoth at The Next Platform.

No Slowdown in Sight for Kubernetes

Kubernetes has quickly become a key technology in the emerging containerized application environment since it was first announced by Google engineers just more than three years ago, catching hold as the primary container orchestration tool used by hyperscalers, HPC organizations and enterprises and overshadowing similar tools like Docker Swarm, Mesos and OpenStack.

Born from earlier internal Google projects Borg and Omega, the open-source Kubernetes has been embraced by top cloud providers and growing numbers of enterprises, and support is growing among datacenter infrastructure software vendors.

Red Hat has built out its OpenShift cloud application platform based on both …

No Slowdown in Sight for Kubernetes was written by Nicole Hemsoth at The Next Platform.

FICO CIO on the Costs, Concerns of Cloud Transition

Moving large-scale enterprise operations into the cloud is not a decision to be made lightly. There are engineering and financial considerations, and the process of determining the costs pros and cons of such a move is significantly more complex than simply comparing the expense of running a workload on-premises or in a public cloud.

Still, the trend is toward businesses making the move to one degree or another, driven by the easy ability to scale up or down depending on the workload and paying only for the infrastructure resources they use, not having to put up the capital expense to …

FICO CIO on the Costs, Concerns of Cloud Transition was written by Nicole Hemsoth at The Next Platform.

Faster Machine Learning in a World with Limited Memory

Striking acceptable training times for GPU accelerated machine learning on very large datasets has long-since been a challenge, in part because there are limited options with constrained on-board GPU memory.

For those who are working on training against massive volumes (in the many millions to billions of examples) using cloud infrastructure, the impetus is greater than ever to pare down training time given the per-hour instance costs and for cloud-based GPU acceleration on hardware with more memory (the more expensive Nvidia P100 with 16 GB memory over a more standard 8 GB memory GPU instance). Since hardware limitations are not …

Faster Machine Learning in a World with Limited Memory was written by Nicole Hemsoth at The Next Platform.

When POSIX I/O Meets Exascale, Do the Old Rules Apply?

We’ve all grown up in a world of digital filing cabinets. POSIX I/O has enabled code portability and extraordinary advances in computation, but it is limited by its design and the way it mirrors the paper offices that it has replaced.

The POSIX API and its implementation assumes that we know roughly where our data is, that accessing it is reasonably quick and that all versions of the data are the same. As we move to exascale, we need to let go of this model and embrace a sea of data and a very different way of handling it.

In …

When POSIX I/O Meets Exascale, Do the Old Rules Apply? was written by Nicole Hemsoth at The Next Platform.

The Systems of the Future Will Be Conversational

It’d be difficult to downplay the impact Amazon Web Services has had on the computing industry over the past decade. Since launching in 2006, Amazon’s cloud computing division has become the set the pace in the public cloud market, rapidly growing out its capabilities from the first service – Simple Storage Service (S3) – it rolled out to now offering thousands of services that touch on everything from compute instances to databases, storage, application development and emerging technologies like machine learning and data analytics.

The company has become dominant by offering organizations of all sizes a way of simply accessing …

The Systems of the Future Will Be Conversational was written by Nicole Hemsoth at The Next Platform.

Reinventing the FPGA Programming Wheel

For several years, GPU acceleration matched with Intel Xeon processors were the dominating news items in hardware at the annual Supercomputing Conference. However, this year that trend shifted in earnest, with a major coming-out party for ARM servers in HPC and more attention than ever paid to FPGAs as potential accelerators for future exascale systems.

The SC series held two days of lightening-round presentations on the state of FPGAs for future supercomputers, with insight from both academia, vendors, and end users at scale, including Microsoft. To say Microsoft is an FPGA user is a bit of an understatement, however, since …

Reinventing the FPGA Programming Wheel was written by Nicole Hemsoth at The Next Platform.

Hospitals Untangling Infrastructure from Deep Learning Projects

Medical imaging is one areas where hospitals have invested significantly in on-premises infrastructure to support diagnostic analysis.

These investments have been stepped up in recent years with ever-more complex frameworks for analyzing scans, but as cloud continues to mature, the build versus buy hardware question gets more complicated. This is especially true with the addition to deep learning for medical images into more hospital settings—something that adds more hardware and software heft to an already top-heavy stack.

Earlier this week, we talked about the medical imaging revolution that is being driven forward by GPU accelerated deep learning, but as it …

Hospitals Untangling Infrastructure from Deep Learning Projects was written by Nicole Hemsoth at The Next Platform.

AWS Flexes Cloud Muscles with Host of New Additions

Tech vendors often like to boast about being first movers in a particular market, saying that leading the charge puts them at a great advantage over their competitors. It doesn’t always work that way, but sometimes it does.

A case in point is Amazon Web Services (AWS), which officially launched in 2006 with the release of the Simple Storage Service (S3) after several years of development and with it kicked off what is now the fast-growing and increasingly crowded public cloud space. Eleven years later, AWS owns just over 44 percent of the market, according to CEO Andy Jassy, pointing …

AWS Flexes Cloud Muscles with Host of New Additions was written by Nicole Hemsoth at The Next Platform.

Julia Language Delivers Petascale HPC Performance

Written in the productivity language Julia, the Celeste project—which aims to catalogue all of the telescope data for the stars and galaxies in in the visible universe—demonstrated the first Julia application to exceed 1 PF/s of double-precision floating-point performance (specifically 1.54 PF/s).

The project took advantage of all 9300 Intel Xeon Phi Phase II nodes on the NERSC (National Energy Research Scientific Computing Center) Cori supercomputer.

Even in HPC terms, the Celeste project is big, as it created the first comprehensive catalog of visible objects in our universe by processing 178 terabytes of SDSS (Sloan Digital …

Julia Language Delivers Petascale HPC Performance was written by Nicole Hemsoth at The Next Platform.

Medical Imaging Drives GPU Accelerated Deep Learning Developments

Although most recognize GE as a leading name in energy, the company has steadily built a healthcare empire over the course of decades, beginning in the 1950s in particular with its leadership in medical X-ray machines and later CT systems in the 1970s and today, with devices that touch a broad range of uses.

Much of GE Healthcare’s current medical device business is rooted in imaging hardware and software systems, including CT imaging machines and other diagnostic equipment. The company has also invested significantly in the drug discovery and production arena in recent years—something the new CEO of GE, John …

Medical Imaging Drives GPU Accelerated Deep Learning Developments was written by Nicole Hemsoth at The Next Platform.

IO-500 Goes Where No HPC Storage Metric Has Gone Before

The landscape of HPC storage performance measurement is littered with unrealistic expectations. While there are seemingly endless strings of benchmarks aimed at providing balanced metrics, these far too often come from the vendors themselves.

What is needed is an independent set of measurements for supercomputing storage environments that takes into account all of the many nuances of HPC (versus enterprise) setups. Of course, building such a benchmark suite is no simple task—and ranking the results is not an easy exercise either because there are a great many dependencies; differences between individual machines, networks, memory and I/O tricks in software, and …

IO-500 Goes Where No HPC Storage Metric Has Gone Before was written by Nicole Hemsoth at The Next Platform.

Oak Ridge Lab’s Quantum Simulator Pulls HPC Future Closer

Oak Ridge National Laboratory has been investing heavily in quantum computing across the board. From testing new devices, programming models, and figuring out workflows that combine classical bits with qubits, this is where the DoE quantum investment seems to be centered.

Teams at Oak Ridge have access to the range of available quantum hardware devices—something that is now possible without having to own the difficult-to-manage quantum computer on sight. IBM’s Q processor is available through a web interface, as is D-Wave’s technology, which means researchers at ORNL can test their quantum applications on actual hardware. As we just …

Oak Ridge Lab’s Quantum Simulator Pulls HPC Future Closer was written by Nicole Hemsoth at The Next Platform.

Looking Ahead to Intel’s Secret Exascale Architecture

There has been a lot of talk this week about what architectural direction Intel will be taking for its forthcoming exascale efforts. As we learned when the Aurora system (expected to be the first U.S. exascale system) at Argonne National Lab shifted from the planned Knights Hill course, Intel was seeking a replacement architecture—one that we understand will not be part of the Knights family at all but something entirely different.

Just how different that will be is up for debate. Some have posited that the exascale architecture will feature fully integrated hardware acceleration (no offload model needed for …

Looking Ahead to Intel’s Secret Exascale Architecture was written by Nicole Hemsoth at The Next Platform.

D-Wave Makes Quantum Leap with Reverse Annealing

The art and science of quantum annealing to arrive at a best of all worlds answer to difficult questions has been well understood for years (even if implementing it as a computational device took time). But that area is now being turned on its head—all for the sake of achieving more nuanced results that balance the best of quantum and classical algorithms.

This new approach to quantum computing is called reverse annealing, something that has been on the research wish-list at Google and elsewhere, but is now a reality on the newest D-Wave 2000Q (2048 qubit) hardware. The company described …

D-Wave Makes Quantum Leap with Reverse Annealing was written by Nicole Hemsoth at The Next Platform.

Samsung Invests in Cray Supercomputer for Deep Learning Initiatives

One of the reasons this year’s Supercomputing Conference (SC) is nearing attendance records has far less to do with traditional scientific HPC and much more to do with growing interest in deep learning and machine learning.

Since the supercomputing set has pioneered many of the hardware advances required for AI (and some software and programming techniques as well), it is no surprise new interest from outside HPC is filtering in.

On the subject of pioneering HPC efforts, one of the industry’s longest-standing companies, supercomputer maker Cray, is slowly but surely beginning to reap the benefits of the need for this …

Samsung Invests in Cray Supercomputer for Deep Learning Initiatives was written by Nicole Hemsoth at The Next Platform.

ARM Benchmarks Show HPC Ripe for Processor Shakeup

Every year at the Supercomputing Conference (SC) an unofficial theme emerges. For the last two years, machine learning and deep learning were focal points; before that it was all about data-intensive computing and stretching even farther back, the potential of cloud to reshape supercomputing.

What all of these themes have in common is that they did not focus on the processor. In fact, they centered around a generalized X86 hardware environment with well-known improvement and ecosystem cadences. Come to think of it, the closest we have come to seeing the device at the center of a theme in recent years …

ARM Benchmarks Show HPC Ripe for Processor Shakeup was written by Nicole Hemsoth at The Next Platform.