We have written much about large-scale deep learning implementations over the last couple of years, but one question that is being posed with increasing frequency is how these workloads (training in particular) will scale to many nodes. While different companies, including Baidu and others, have managed to get their deep learning training clusters to scale across many GPU-laden nodes, for the non-hyperscale companies with their own development teams, this scalability is a sticking point.
The answer to deep learning framework scalability can be found in the world of supercomputing. For the many nodes required for large-scale jobs, the de facto …
Pushing MPI into the Deep Learning Training Stack was written by Nicole Hemsoth at The Next Platform.
The United States for years was the dominant player in the high-performance computing world, with more than half of the systems on the Top500 list of the world’s fastest supercomputers being housed in the country. At the same time, most HPC systems around the globe were powered by technologies from such major US tech companies as Intel, IBM, AMD, Cray and Nvidia.
That has changed rapidly over the last several years, as the Chinese government has invested tens of billions of dollars to expand the capabilities of the country’s own technology community and with a promise to spend even more …
Rise of China, Real-World Benchmarks Top Supercomputing Agenda was written by Nicole Hemsoth at The Next Platform.
Being at the bleeding edge of computing in the life sciences does not always mean operating at extreme scale. For some shops, advancements in new data-generating scientific tools requires forward thinking at the infrastructure level—even if it doesn’t require a massive cluster with exotic architectures. We tend to cover much of what happens at the extreme scale of computing here, but it’s worth stepping back and observing how dramatic problems in HPC are addressed in much smaller environments.
This “small shop, big problem” situation is familiar to the Van Andel Research Institute (VARI), which recently moved from a genomics and …
One Small Shop, One Extreme HPC Storage Challenge was written by Nicole Hemsoth at The Next Platform.
Containers continue to gain momentum as organizations look for greater efficiencies and lower costs to run distributed applications in their increasingly virtualized datacenters as well as for improving their application development environments. As we have noted before, containers are becoming more common in the enterprise, though they still have a way to go before being fully embraced in high performance computing circles.
There are myriad advantages to containers, from being able to spin them up much faster than virtual machine instances on hypervisors and to pack more containers than virtual machines on a host system to gaining efficiencies …
A New Twist On Adding Data Persistence To Containers was written by Timothy Prickett Morgan at The Next Platform.
In distributed computing, there are two choices: move the data to the computation or move the computation to the data. Public and off-site private clouds present a third option: move them both. In any case, something is moving somewhere. The “right” choice depends on a variety of factors – including performance, quantity, and cost – but data seems to have more intertia in many cases, especially when the volume approaches the scale of terabytes.
For the modern cloud, adding additional compute power is trivial. Moving the data to that compute power is less so. With a 10 gigabit connection, the …
Solving the Challenges of Scientific Clouds was written by Nicole Hemsoth at The Next Platform.
Data is quickly becoming the coin of the realm in most aspects of the business world, and analytics the best way for organizations to cash in on it. It’s easy to be taken in by the systems and devices – much of the discussion around the Internet of Things tends to be around the things themselves, whether small sensors, mobile devices, self-driving cars or huge manufacturing systems. But the real value is in the data generated by these machines, and the ability to extract that data, analyze it and make decisions based on it in as close to real-time as …
Data, Analytics, Probabilities and the Super Bowl was written by Nicole Hemsoth at The Next Platform.
Imagine how little fun online retailer Amazon would be having on its quarterly calls if it had not launched its Amazon Web Services cloud almost eleven years ago. The very premise of Amazon was to eliminate brick and mortar retailing, cutting out capital expenses as much as possible, to deliver books and then myriad other things to our doorsteps.
How ironic is it that Amazon pivoted to one of the most capital intensive businesses on earth – running datacenters – and has been able to extract predictable and sizable profits from it to prop up its other businesses and strengthen …
When Will AWS Move Up The Stack To Real Applications? was written by Timothy Prickett Morgan at The Next Platform.
China represents a big and growing market opportunity for IT vendors around the world. It’s huge population and market upside compared with the more mature regions across the globe is hugely attractive to system and component makers, and the Chinese government’s willingness to spend money to help build up the country’s compute capabilities only adds to the allure. In addition, it is home to such hyperscale players as Baidu, Alibaba and Tencent, which like US counterparts Google, Facebook and eBay are building out massive datacenters that are housing tens of thousands of servers.
However, those same Chinese government officials aren’t …
Chip Makers and the China Challenge was written by Nicole Hemsoth at The Next Platform.
Over the last two years, there has been a push for novel architectures to feed the needs of machine learning and more specifically, deep neural networks.
We have covered the many architectural options for both the training and inference sides of that workload here at The Next Platform, and in doing so, started to notice an interesting trend. Some companies with custom ASICs targeted at that market seemed to be developing along a common thread—using memory as the target for processing.
Processing in memory (PIM) architectures are certainly nothing new, but because the relatively simple logic units inside of …
Memory at the Core of New Deep Learning Research Chip was written by Nicole Hemsoth at The Next Platform.
The trend at the high end, from supercomputer simulations to large-scale genomics studies, is to push heterogeneity and software complexity while reducing the overhead on the infrastructure side. This might sound like a case of dueling forces, but there is progress in creating a unified framework to run multiple workloads simultaneously on one robust cluster.
To put this into context from a precision medicine angle, Dr. Michael McManus shared his insights about the years he spent designing infrastructure for life sciences companies and research. Those fields have changed dramatically in just the last five years alone in terms of data …
Many Life Sciences Workloads, One Single System was written by Nicole Hemsoth at The Next Platform.
Public clouds have proven useful to a growing number of organizations looking for ways to run their high-performance computing applications to scale without having to limit themselves to whatever computing capabilities they have in-house or to spending a lot of money to build up their infrastructure to meeting their growing needs.
The big three – Amazon Web Services, Microsoft Azure and Google Cloud – have rolled out a broad array of compute, networking and storage technologies that companies can leverage when their HPC workloads scale to the point that they can no longer be run on their in-house workstations or …
Orchestrating HPC Engineering in the Cloud was written by Nicole Hemsoth at The Next Platform.
The idea of ARM processors being used in datacenter servers has been kicking around more most of the decade. The low-power architecture dominates the mobile world of smartphones and tablets as well as embedded IoT devices, and with datacenters increasingly consuming more power and generating more heat, the idea of using highly efficient ARM chips in IT infrastructure systems gained steam.
That was furthered by the rise of cloud computing environments and hyperscale datacenters, which can be packed with tens of thousands of small servers running massive numbers of workloads. The thought of using ARM-based server chips that are more …
ARM Server Chips Challenge X86 in the Cloud was written by Nicole Hemsoth at The Next Platform.
There are individuals and companies that create whole new technologies for their own consumption and that sometimes open source them for others to help steer their development and fix their bugs. And then there are still other companies that polish these tools, giving them some enterprise fit and finish, and thereby make it possible for others to deploy a particular technology without having to have PhDs, who are not available anyway, on staff.
From the enterprise perspective, the Apache web server and related Tomcat application server needed its Big Blue, the Linux operating system needed its Red Hat, and the …
Riding The Coattails Of Google Kubernetes And AWS Lambda was written by Timothy Prickett Morgan at The Next Platform.
We are thrilled to announce the full-time addition of veteran IT journalist, Jeffrey Burt to The Next Platform ranks.
Jeffrey Burt has been a journalist for more than 30 years, with the last 16-plus year writing about the IT industry. During his long tenure with eWeek, he covered a broad range of subjects, from processors and IT infrastructure to collaboration, PCs, AI and autonomous vehicles.
He’s written about FPGAs, supercomputers, hyperconverged infrastructure and SDN, cloud computing, deep learning and exascale computing. Regular readers here will recognize that his expertise in these areas fits in directly with our coverage …
Veteran IT Journalist, Jeffrey Burt, Joins The Next Platform as Senior Editor was written by Nicole Hemsoth at The Next Platform.
Hardware and device makers are in a mad dash to create or acquire the perfect chip for performing deep learning training and inference. While we have yet to see anything that can handle both parts of the workload on a single chip with spectacular results (the Pascal general GPUs are the closest thing yet, with threats coming from Intel/Nervana in the future), there is promise for FPGAs to find inroads.
So far, most of the work we have focused on for FPGAs and deep learning has centered more on the acceleration of inference versus boosting training times and accuracy …
OpenCL Opens Doors to Deep Learning Training on FPGA was written by Nicole Hemsoth at The Next Platform.
The race is on to carve a path to efficient extreme-scale machines in the next five years but existing processing approaches fall far short of the efficiency and performance targets required. As we reported at the end of 2016, the Department of Energy in the U.S. is keeping its eye on non-standard processing approaches for one of its exascale-class systems by 2021, and other groups, including the IEEE are equally keeping pace with new architectures to explore as CMOS alternatives.
While there is no silver bullet technology yet that we expect will sweep current computing norms, superconducting circuits appear …
IARPA Spurs Race to Speed Cryogenic Computing Reality was written by Nicole Hemsoth at The Next Platform.
Breaking into the switch market is not an easy task, whether you are talking about providing whole switches or just the chips that drive them. But there is always room for innovation, which is why some of the upstarts have a pretty credible chance to shake up networking, which is the last bastion of proprietary within the datacenter.
Barefoot Networks is one of the up-and-coming switch chip makers, with its “Tofino” family of ASICs that, among other things, has circuits and software that allow for the data plane – that part of the device that controls how data moves …
Hyperscalers Ready To Run Barefoot In The Datacenter was written by Timothy Prickett Morgan at The Next Platform.
High performance computing (HPC) is traditionally considered the domain of large, purpose built machines running some *nix operating system (predominantly Linux in recent years). Windows is given little, if any, consideration. Indeed, it has never accounted for even a full percent of the Top500 list. Some of this may be due to technical considerations: Linux can be custom built for optimum performance, including recompiling the kernel. It is also historically more amenable to headless administration, which is a critical factor when maintaining thousands of nodes.
But at some point does the “Windows isn’t for high-performance computing” narrative become self-fulfilling? …
Looking Through the Windows at HPC OS Trends was written by Nicole Hemsoth at The Next Platform.
High performance computing in its various guises is not just determined by the kind and amount of computing that is made available at scale to applications. More and more, the choice of network adapters and switches as well as the software stack that links the network to applications plays an increasingly important role. And moreover, networks are comprising a larger and larger portion of the cluster budget, too.
So picking the network that lashes servers to each other and to their shared storage is important. And equally important is having a roadmap for the technology that is going to provide …
The Relentless Yet Predictable Pace Of InfiniBand Speed Bumps was written by Timothy Prickett Morgan at The Next Platform.
Every successive processor generation presents its own challenges to all chip makers, and the ramp of 14 nanometer processes that will be used in the future “Skylake” Xeon processors, due in the second half of this year, cut into the operating profits of its Data Center Group in the final quarter of 2016. Intel also apparently had an issue with one of its chip lines – it did not say if it was a Xeon or Xeon Phi, or detail what that issue was – that needed to be fixed and that hurt Data Center Group’s middle line, too.
Still, …
Skylake Xeon Ramp Cuts Into Intel’s Datacenter Profits was written by Timothy Prickett Morgan at The Next Platform.