Supercomputing for the Exascale Era

Supercomputing for the Exascale Era

Cray, an HPE company, held a panel discussion webinar on October 18 to discuss Exascale (10^18, get it?) supercomputing. This is definitely not in my area of expertise, but it is certainly interesting.

Following is information I gleaned from links they sent to me. Basically, it is Why Supercomputing. And not only computers, but also networking to support them.

Today’s science, technology, and big data questions are bigger, more complex, and more urgent than ever. Answering them demands an entirely new approach to computing. Meet the next era of supercomputing. Code-named Shasta, this system is our most significant technology advancement in decades. With it, we’re introducing revolutionary capabilities for revolutionary questions. Shasta is the next era of supercomputing for your next era of science, discovery, and achievement.

WHY SUPERCOMPUTING IS CHANGING

The kinds of questions being asked today have created a sea-change in supercomputing. Increasingly, high-performance computing systems need to be able to handle massive converged modeling, simulation, AI, and analytics workloads.

With these needs driving science and technology, the next generation of supercomputing will be characterized by exascale performance, data-centric workloads and diversification of processor architectures.

SUPERCOMPUTING REDESIGNED

Shasta is that entirely new design. We’ve created it from the ground up to address today’s diversifying needs.

Built to be data-centric, it runs diverse workloads all at the same time. Hardware and software innovations tackle system bottlenecks, manageability, and job completion issues that emerge or grow when core counts increase, compute node architectures proliferate, and workflows expand to incorporate AI at scale.

It eliminates the distinction between clusters and supercomputers with a single new system architecture, enabling a choice of computational infrastructure without tradeoffs. And it allows for mixing and matching multiple processor and accelerator architectures with support for our
new Cray-designed and developed interconnect we call Slingshot.

EXASCALE-ERA NETWORKING

Slingshot is our new high-speed, purpose-built supercomputing interconnect. It’s our eighth generation of scalable HPC network. In earlier Cray designs, we pioneered the use of adaptive routing, pioneered the design of high-radix switch architectures, and invented a new low-diameter system topology, the dragonfly.

Slingshot breaks new ground again. It features Ethernet capability, advanced adaptive routing, first-of-a-kind congestion control, and sophisticated quality-of-service capabilities. Support for both IP-routed and remote memory operations broadens the range of applications beyond traditional modeling and simulation.

Quality-of-service and novel congestion management features limit the impact to critical workloads from other applications, system services, I/O traffic, or co-tenant workloads. Reduction in the network diameter from five hops (in the current Cray XCTM generation) to three reduces cost, latency, and power while improving sustained bandwidth and reliability.

FLEXIBILITY AND TCO

As your workloads rapidly evolve, the ability to choose your architecture becomes critical. With Shasta, you can incorporate any silicon processing choice — or a heterogenous mix — with a single management and application development infrastructure. Flex from single to multi-socket nodes, GPUs, FPGAs, and other processing options that may emerge, such as AI-specialized accelerators.

Designed for a decade or more of work, Shasta also eliminates the need for frequent, expensive upgrades, giving you exceptionally low total
cost of ownership. With its software architecture you can deploy a workflow and management environment in a single system, regardless of packaging.

Shasta packaging comes in two options: a 19” air- or liquid-cooled, standard datacenter rack and a high-density, liquid-cooled rack designed to take 64 compute blades with multiple processors per blade.

Additionally, Shasta supports processors well over 500 watts, eliminating the need to do forklift upgrades of system infrastructure to accommodate higher-power processors.

Supercomputing for the Exascale Era

How To Avoid Pilot Purgatory For Your Projects

This is still more followup from Emerson Global Users Exchange relative to sessions on Projects Pilot Purgatory. I thought I had already written this, but just discovered it languishing in my drafts folder. While in Nashville, I ran into Jonas Berge, senior director, applied technology for Plantweb at Emerson Automation. He has been a source for technology updates for years. We followed up a brief conversation with a flurry of emails where he updated me on some presentations.

One important topic centered on IoT projects—actually applicable to other types of projects as well. He told me the secret sauce is to start small. “A World Economic Forum white paper on the fourth industrial revolution in collaboration with McKinsey suggests that to avoid getting stuck in prolonged “pilot purgatory” plants shall start small with multiple projects – just like we spoke about at EGUE and just like Denka and Chevron Oronite and others have done,” he told me.

“I personally believe the problem is when plants get advice to take a ‘big bang’ approach starting by spending years and millions on an additional ‘single software platform’ or data lake and hiring a data science team even before the first use case is tackled,” said Berge. “My blog post explains this approach to avoiding pilot purgatory in greater detail.”

I recommend visiting Berge’s blog for more detail, but I’ll provide some teaser ideas here.

First he recommends

  • Think Big
  • Start Small
  • Scale Fast

Scale Fast

Plants must scale digital transformation across the entire site to fully enjoy the safety benefits like fewer incidents, faster incident response time, reduced instances of non-compliance, as well as reliability benefits such as greater availability, reduced maintenance cost, extend equipment life, greater integrity (fewer instances of loss of containment), shorter turnarounds, and longer between turnarounds. The same holds true for energy benefits like lower energy consumption, cost, and reduced emissions and carbon footprint, as well as production benefits like reduced off-spec product (higher quality/yield), greater throughput, greater flexibility (feedstock use, and products/grades), reduced operations cost, and shorter lead-time.

Start Small

The organization can only absorb so much change at any one time. If too many changes are introduced in one go, the digitalization will stall:

  • Too many technologies at once
  • Too many data aggregation layers
  • Too many custom applications
  • Too many new roles
  • Too many vendors

Multiple Phased Projects

McKinsey research shows plants successfully scaling digital transformation instead run smaller digitalization projects; multiple small projects across the functional areas. This matches what I have personally seen in projects I have worked on.

From what I can tell it is plants that attempt a big bang approach with many digital technologies at once that struggle to scale. There are forces that encourage companies to try to achieve sweeping changes to go digital, which can lead to counterproductive overreaching. 

The Boston Consulting Group (BCG) suggests a disciplined phased approach rather than attempting to boil the ocean. I have seen plants focus on a technology that can digitally transform and help multiple functional areas with common infrastructure. A good example is wireless sensor networks. Deploying wireless sensor networks in turn enables many small projects that help many departments digitally transform the way they work. The infrastructure for one technology can be deployed relatively quickly after which many small projects are executed in phases.

Small projects are low-risk. A small trial of a solution in one plant unit finishes fast. After a quick success, then scale it to the full plant area, and then scale to the entire plant. Then the team can move on to start the next pilot project. This way plants move from PoC to full-scale plant-wide implementation at speed. For large organization with multiple plants, innovations often emerge at an individual plant, then gets replicated at other sites, rolled out nation-wide and globally.

Use Existing Platform

I have also seen big bang approach where plant pours a lot of money and resources into an additional “single software platform” layer for data aggregation before the first use-case even gets started. This new data aggregation platform layer is meant to be added above the ERP with the intention to collect data from the ERP and plant historian before making it available to analytics through proprietary API requiring custom programming. 

Instead, successful plants start small projects using the existing data aggregation platform; the plant historian. The historian can be scaled with additional tags as needed. This way a project can be implemented within two weeks, with the pilot running an additional three months, at low-risk. 

Think Big
I personally like to add you must also think of the bigger vision. A plant cannot run multiple small projects in isolation resulting in siloed solutions. Plants successful with digital transformation early on establish a vision of what the end goal looks like. Based on this they can select the technologies and architecture to build the infrastructure that supports this end goal.
NAMUR Open Architecture (NOA)
The system architecture for the digital operational infrastructure (DOI) is important. The wrong architecture leads to delays and inability to scale. NAMUR (User Association of Automation Technology in Process Industries) has defined the NAMUR Open Architecture (NOA) to enable Industry 4.0. I have found that plants that have deployed digital operational infrastructure (DOI) modelled on the same principles as NOA are able to pilot and scale very fast. Flying StartThe I&C department in plants can accelerate digital transformation to achieve operational excellence and top quartile performance by remembering Think Big, Start Small, Scale Fast. These translate into a few simple design principles:

  • Phased approach
  • Architecture modeled on the NAMUR Open Architecture
  • Ready-made apps
  • East-to-use software
  • Digital ecosystem
Supercomputing for the Exascale Era

Collaborative Robots Just Keep Getting More Interesting

This entire area of collaborative robots (cobots) just keeps getting more interesting. The idea of humans and robots working collaboratively is intuitive but has been difficult to achieve. Cobots have ramifications far beyond industrial applications. But even here, they can lead the way to better productivity and effectiveness.

In this latest piece of news, Universal Robots (UR) announced the immediate availability of the UR16e which boasts an impressive 16 kg (35 lbs) payload capability.

This cobot combines the high payload with a reach of 900 mm and pose repeatability of +/- 0.05 mm making it ideal for automating tasks such as heavy-duty material handling, heavy-part handling, palletizing, and machine tending.

“In today’s uncertain economic climate manufacturers need to look at flexible solutions to stay competitive,” said Jürgen von Hollen, President of Universal Robots. “With UR16e, we meet the need for a collaborative robot that can tackle heavy-duty tasks reliably and efficiently. This launch significantly expands the versatility of our product portfolio and gives manufacturers even more ways to improve performance, overcome labor challenges, and grow their business.”

Developed on UR’s e-Series platform, the UR16e offers these benefits:

·        Fast and frictionless deployment with easy programming and a small footprint

UR16e makes accelerating automation easy and fast. Programming and integration is simple – regardless of the user’s experience or knowledge base. Like all UR’s cobots, UR16e can be unpacked, mounted and programmed to perform a task within less than an hour. With a small footprint and 900 mm reach, UR16e easily integrates into any production environment without disruption. 

·        Addresses ergonomic challenges while lowering cost

With its 16 kg payload, UR16e eliminates the ergonomic and productivity challenges associated with lifting and moving heavy parts and products, lowering costs, and reducing downtime.  

·        Ideal for heavy-duty material handling and machine tending

Rugged and reliable, UR16e is ideal for automating high-payload and CNC machine tending applications, including multi-part handling, without compromising on precision.

“At Universal Robots we continue to push the boundaries of what’s possible with collaborative automation,” continued von Hollen. “Today, we’re making it easier than ever for every manufacturer to capitalize on the power of automation by bringing a cobot to market that is built to do more, as it delivers more payload than our other cobots.”

Like with UR’s other e-Series cobots; UR3e, UR5e and UR10e, the UR16e includes built-in force sensing, 17 configurable safety functions, including customizable stopping time and stopping distance, and an intuitive programming flow. UR16e meets the most demanding compliance regulations and safety standards for unobstructed human-robot collaboration, including EN ISO 13849-1, PLd, Category 3, and full EN ISO 10218-1.

Supercomputing for the Exascale Era

Collaborative Robotics Offer Flexible Packaging Automation

That engineers would develop ways for humans and robots to co-exist, yes even collaborate, seemed inevitable. Why should we consign robots to cages as safety hazards when the future assuredly requires close collaboration. Therefore the burgeoning area of collaborative robotics or cobots.

I’m thinking not just about industrial applications. Robots surely will assist an aging population cope with everyday tasks in our (near) future of fewer people to populate those jobs.

Several of the “old guard” robotics companies have developed “co-bots” but I’ve watched the development of Universal Robots for some time. The company sponsored this blog for a while a few years ago. Here I’ve picked up on a couple of items. The UR marketing team was a bit surprised to discover that I have more than a passing interest in packaging. As a matter of fact, I noticed packaging as a likely growth area for automation about 18 years ago, and that feeling has been borne out.

One story concerns a packaging demonstration with a socially worthwhile goal mixed in. The other reports on a recent market study by ABI Research.

Universal Robots Solves Random Picking Challenge, Providing Food for At-Risk Youth

The challenge: Pick six differently sized food items randomly oriented on a moving conveyor and place each of these items into the same pouch. Then do this again 1,199 more times, ensuring each pouch has the same six items. This is the challenge Universal Robots and Allied Technology will address, quickly identifying and picking items – ranging widely from packs of Craisins to cans of beef ravioli – in Pack Expo’s Robotics Zone during the three-day show.

“Random picking is quickly becoming one of the most sought-after automation tasks from industries such as e-commerce, fulfillment centers and warehousing,” says Regional Sales Director of Universal Robots’ Americas division, Stuart Shepherd. “At Pack Expo, Universal Robots and Allied Technology will demonstrate how UR cobots can be quickly deployed in a compact, modular system, handling the entire process from box erecting, to vision-guided conveyor tracking, part picking, tote assembly, pouch filling and sealing, kitting and palletizing,” he says, adding how the packaging line is also a testament to the capabilities of Universal Robots’ growing number of Certified System Integrators (CSIs). “Allied Technology was able to quickly create this fully-automated solution. We are delighted to see our cobots competently integrated in so many new packaging applications now.”

Allied Technology and Universal Robots’ packaging line features four UR cobots equipped with products from the UR+ platform that certifies grippers, vision cameras, software, and other peripherals to work seamlessly with UR’s collaborative robot arms. The latest flexible grasping technology will be showcased by a UR5e with Piab’s new Kenos® KCS vacuum gripper  guided by a vision camera from UR+ partner Cognex.

Once completed, the 1,200 bags of food will be delivered to “Blessings in a Backpack” a leader in the movement to end childhood hunger, ensuring that children receiving free or subsidized school lunches during the week do not go hungry over the weekends. “We look forward to showcasing this demo that is meaningful in so many ways,” says Shepherd. “We are excited to partner with Blessings in a Backpack while also addressing the needs of the packaging industry with solutions that will simplify and fast-track cobot deployment on their lines.”

Unlike traditional robots caged away from show attendees, visitors to the UR booth are able to walk right up to the UR cobots and interact with them. The booth “playpen area” will feature several cobot arms including a U53e with Robotiq’s new UR+ certified E-Pick Vacuum Gripper, allowing attendees to explore on-the-spot programming. The gripper is one of the recent additions in a rapidly expanding UR+ product portfolio that now includes no less than 195 UR+ certified products with 400+ companies participating in the UR+ developer program.

Market Leadership

Meanwhile, Universal Robots maintains top spot in ABI Research’s Ranking of Cobot Companies in Industrial Applications; Doosan, Techman Robot, and Precise Automation are closing in.

This news originates with ABI Research. There are well over 50 manufacturers of collaborative robots (cobots) worldwide, but only a handful of these companies have so far deployed cobots on any meaningful level of scale. Tens of thousands of cobots have been sold as of 2019 and earned US$500 million in annual revenue for world markets. In its new Industrial Collaborative Robots Competitive Assessment, global tech market advisory firm, ABI Research finds Universal Robots (UR) to be the clear forerunner, particularly in implementation.

The Industrial Collaborative Robots Competitive Assessment analyzed and ranked 12 collaborative robot vendors in the industry – ABB, Aubo Robotics, Automata, Doosan Robotics, FANUC, Franka Emika, Kuka AG, Precise Automation, Productive Robotics, Techman Robot, Universal Robots, and Yaskawa Motoman – using ABI Research’s proven, unbiased innovation/implementation criteria framework. For this competitive assessment, innovation criteria included payload, software, Ergonomics and human-machine interaction, experimentation and safety; implementation criteria focused on units and revenue, cost and ROI, partnerships, value-added services, and the number of employees.

“Market leaders in cobots generally have well-developed cobot rosters, in many cases backed up by an ecosystem platform that integrates applications, accessories, and end-of-arm-tooling (EOAT) solutions in with the base hardware,” said Rian Whitton, Senior Analyst at ABI Research. With 37,000 cobots sold so far, UR leads, followed by Taiwanese provider Techman with 10,000, and Korea-based Doosan with over 2,000. Precise Automation, which uses an advanced direct drive solution to develop faster collaborative robots, was cited as the most innovative of the 12 providers, just edging out Universal Robots, who claimed the overall top spot due to their significant lead in implementation.

There are several companies that are too young to be challenging the dominant parties in the cobot market but are developing new and disruptive technologies that will allow them rise to prominence in the years to come. Productive Robotics is a case-in-point. The California-based developer has an arm with inbuilt vision, 7 axes for superior flexibility, long reach, and a very affordable price point, but has yet to deploy at scale. Automata, a British company that develops a ‘desk-top’ cobot costing less than US$7,000, is significantly lowering the barriers to entry for smaller actors and is championing the use of open-source middleware like ROS to program cobots for industrial use-cases. Germany-based Franka Emika and Chinese-American provider Aubo Robotics also represent relatively new entrants to the market who are building on the success of Universal Robots and are beginning to compete with them.

Perhaps surprisingly, while the major industrial robotics providers have developed cobot lines, they have generally been less successful in marketing them or gaining market traction relative to the pure-cobot developers. In part, this is down to focus. While collaborative robots are valuable, they generally suit deployments and use-cases with smaller shipments and a wider variety of small and large end-users. For industrial players like ABB, FANUC, KUKA AG and Yaskawa Motoman, their client-base tends to be large industrial players who buy fixed automation solution through bulk orders. Aside from this, all four of these companies are competing extensively for greater shipment figures in China, where the cobot oppurtunity relative to the market for traditional industrial systems is much less apparent than in Europe or North America.

“Though many of the cobots deployed by these companies are impressive, and they have a lot of software services, the high-cost and lack of easy use among their systems largely defeat the current value proposition of cobots, making them the laggards in this competitive assessment.” says Whitton.

Looking forward, the larger industrial players are likely to improve their relative position, as future growth in cobots rests on scaling up and large deployments. “Universal Robots, though likely to remain the market leader for the foreseeable future, will be increasingly competing on an even footing with near-peer cobot developers, who are already developing second-generation cobots with significant hardware improvements. Meanwhile, some more innovative companies will be able to accelerate adoption through price decreases, improved flexibility, and common platforms to retrofit collaborative capability on industrial robots,” Whitton concluded.

These findings are from ABI Research’s Industrial Collaborative Robots Competitive Assessment report. This report is part of the company’s Industrial Solution, which includes research, data, and analyst insights. Competitive Assessment reports offer comprehensive analysis of implementation strategies and innovation, coupled with market share analysis, to offer unparalleled insight into a company’s performance and standing in comparison to its competitors.

Looking For Skilled Labor–Try Hiring a Welding Bot

Looking For Skilled Labor–Try Hiring a Welding Bot

I think even mainstream media has caught up to the current hiring situation—it is hard to find qualified people to fill positions. Heck, even last night stopping at a Cracker Barrel on the drive home from Tennessee we saw a harried staff and only about half the tables filled. They didn’t have enough people to staff the place.

Now try to find skilled welders. How about hiring a robotic welder? Easy to program and install. Always shows up for work. No drug test required. A very interesting idea.

The new for-hire BotX Welder—developed by Hirebotics and utilizing Universal Robots’ UR10e collaborative robot arm—lets manufacturers automate arc welding with no capital investment, handling even small batch runs not feasible for traditional automation.

The press release tells us, “Nowhere in manufacturing is the shortage of labor felt as urgently as in the welding sector, which is now facing an acute shortage of welders nationwide. The industry’s hiring challenge, combined with the struggle metal fabrication companies experience in producing quality parts quickly and in small runs, prompted Hirebotics to develop the BotX Welder.

“Many people didn’t believe that collaborative robots could perform such heavy-duty tasks as welding,” says Rob Goldiez, co-founder of Hirebotics. “We realized the need of a solution for small and medium sized metal fabricators trying to find welders.” Hirebotics’ hire-a-robot business model built on the Universal Robot, set the foundation for the BotX. It is a welding solution powered by the UR10e cobot that is easy to teach, producing automation quality with small batch part runs.

BotX Welder1

The BotX is now available to early access customers and will officially launch at FABTECH in Chicago, November 11-14.

In developing BotX, Hirebotics addressed two major hurdles of robotic welding: the ease of programming and the ease in which a customer can obtain the system without assuming the risk of ownership. There are no installation costs with BotX and with cloud monitoring, manufacturers pay only for the hours the system actually welds. “You can hire and fire BotX as your business needs dictate,” explains Goldiez.

The complete product offering comes with the UR10e cobot arm, cloud connector, welder, wire feeder, MIG welding gun, weld table, and configurable user-input touch buttons. The customer simply provides wire, gas, and parts. Customers can teach BotX the required welds simply via an intuitive app on any smartphone or tablet utilizing welding libraries created in world-class welding labs. A cloud connection enables 24/7 support by Hirebotics.

“We chose Universal Robots’ e-Series line for several reasons,” says Goldiez. “With Universal Robots’ open architecture, we were able to control, not only wire feed speed and voltage, but torch angle as well, which ensures a quality weld every time,” he says. “UR’s open platform also enabled us to develop a cloud-based software solution that allows us to ensure a customer is always running with the latest features at no charge,” explains the Hirebotics co-founder. “We can respond to a customer’s request for additional features within weeks and push those features out to the customer with no onsite visits,” says Goldiez, emphasizing the collaborative safety features of the UR cobots. “The fact that they’re collaborative and don’t require safety fencing like traditional industrial robots means a smaller foot print for the equivalent working space, or put another way; less floor space to produce same size part. In many cases less than half the floor space of traditional automation,” he says. “The collaborative nature of the solution enables an operator to move between multiple cells without interrupting production, greatly increasing the productivity of an employee.”

PMI LLC in Wisconsin, was one of the first customers of the BotX. “A large order would mean, we need to hire 10-15 welders to fulfill it – and they’re just not out there,” says VP of Operations at PMI, Erik Larson. “Therefore, we would No Bid contracts on a regular basis. With the BotX solution, we now quote that work and have been awarded contracts, so it has really helped grow our business,” says Larson. “The BotX Welder doesn’t require expensive, dedicated fixturing and robot experts on the scene.” Now PMI’s existing operators can handle the day-to-day control of the BotX, which welds a variety of smaller product runs.

The Wisconsin job shop has now stored weld programs for more than 50 different parts in their BotX app. “We are now able to deliver quality equivalent to what we could accomplish manufacturing with very expensive tooling typically used with higher-volume part runs,” says PMI’s VP of Operations, mentioning the ease of accessing the solution. “Being able to simply hire the BotX Welder, and quickly switch between welds by using our smart phone—and only pay for the hours it works—is huge for us. It took our area lead, who had no prior robotics experience, half an hour to teach it how to weld the first part.”

Another significant benefit was PMI’s ability to get the BotX welds certified for customers who require this. “This now means we do not need to use certified welders to oversee the operation. As long as the cobot welder’s program is certified, any operator can tend the cobot welder. This really unlocks a lot of resources for us,” says Larson. “Hirebotics and Universal Robots really hit the mark with this, we’re looking forward to a long partnership with them.”

Follow this blog

Get every new post delivered right to your inbox.