Insight Tech

AI is the perfect example of a disruptive technology—and even established companies can take advantage of the digital transformation prospects that can bring.

For systems integrators (SIs) in particular, it is a time of unprecedented opportunity. In nearly every industry, SIs enjoy strong customer demand for AI-enabled digital transformation. But building an AI-enabled solution that’s market ready can be time- and labor-intensive.

“The AI product marketplace is extremely fragmented, and systems integrators lose a substantial amount of time sending their engineers out to look for all the different hardware and software elements to deliver a solution that meets spec,” says Allan Hia, South Asia Senior Director for WPG Holdings(WPG), an Intel^® IoT solution aggregator that serves OT and IT SIs.

This poses a real problem for today’s SIs, who are under pressure to bring innovative AI solutions to market quickly—or risk being left behind.

Scalable, Repeatable Edge AI Solutions Speed Time to Market

In this difficult environment, full-service aggregators like WPG make it possible for SIs to enter the AI marketplace. With large footprints that span borders and industry verticals, aggregators can provide a number of unique benefits to their partners:

• Edge AI platforms that are fully integrated and tested, which can be purchased as easily as a component part.
• Access to new markets through an international network of solution providers and established business relationships with hard-to-reach end customers.
• A range of value-added services, including financing, training, logistics, and help with import/export in international markets.

#IoT solution aggregators deliver value to all parties involved in #DigitalTransformation, across multiple industries. WPG Holdings via @insightdottech

In short, IoT solution aggregators deliver value to all parties involved in digital transformation, across multiple industries.

“For SIs, the biggest benefit to working with a solution aggregator is that it speeds time to market,” says Hia. “It gives them access to repeatable solutions that have already been developed to the proof-of-concept stage or beyond.”

For hardware, software, and cloud technology providers, aggregators offer a way to scale, and help open doors and smooth operations in new markets. And end customers benefit as well. They get best-of-breed, AI-powered solutions that deliver positive business outcomes—without the need to invest in technology development.

SIs Benefit from Partner Ecosystems

A good example of how aggregators can benefit multiple stakeholders at the same time is WPG’s partnership with V-Series, an IoT solution provider in Malaysia. The two companies collaborated to develop Kawanmu, an AI-enabled self-service delivery solution for multi-tenant residences and offices.

Kawanmu is an Intel^® processor-powered smart locker and vending machine system that integrates a number of hardware and software elements into a single system:

• Smart locker units for package and food deliveries
• Modular design flexibility for UV-C vending machine to combine with 3 locker modules
• UV-C disinfection system automatically sanitizes delivered goods and food items
• Barcode scanner for cashless payment and picking up parcel and food
• Interactive screen that enables customer recognition and intelligent display
• Back-end cloud management system for tracking, control, and inventory management
• Mobile app helps customers track parcels, receive alerts, and make cashless payments

It’s easy to imagine a solution like Kawanmu finding buyers in almost any major city in the world. There is a global demand for contactless experiences—and managing food and package delivery in busy lobbies and entryways is a universal problem.

But for a systems integrator, pulling these disparate technologies together into a unified, turnkey solution would have been a formidable undertaking requiring lengthy development work. Because of partnerships like the one between WPG and V-Series, they now have a better option. The result, remarks Hia, is a classic win-win scenario: “It’s good for WPG, good for V-Series, SIs—and great for the end customers.”

Edge AI Goes Mainstream

Solution aggregators create business partnerships where everyone benefits while establishing an AI-enabled product marketplace in which end customers can feel assured of quality. In this, WPG’s own technology partnership with Intel is a major factor.

“We view Intel as the key technology leader in terms of AI computation power—with their processors as well as development resources such as the Intel^® OpenVINO^™ Toolkit, which we leveraged to add smart display functionality in the Kawanmu solution,” says Hia.

He also cites Intel’s extensive stress and compatibility testing of particular value: “When you’re building on a computing platform with well-defined and well-documented capabilities, it speeds development, and it makes it easier to provide solutions that eliminate the unknown for SIs and end customers.”

The impact of solution aggregators on the adoption of edge AI is already significant. But their long-term effect may be even more important, as they are helping to develop a broader product ecosystem for solutions that use AI at the edge. Hia likens the evolution of AI to what happened in the world of ICT 20 years ago: “One day, perhaps as soon as five years, AI will be as seamless as ICT. People who want an AI-powered solution, will know exactly what they need, and they’ll just order it.”

And as Hia sees it, the potential of a mature ecosystem of edge AI-enabled products is enormous: “It will be huge. The usage model will be high. The deployment volume will be comparable to smartphones or laptops.”

At the moment, it may seem strange to think of a world in which fully realized edge AI solutions have become, of all things, commonplace. But that’s because the digital transformation of business is still just beginning. In the coming decade, solution aggregators will play a major role in driving that transformation—building bridges between enterprises, between markets, and to the future.

When it comes to IT, everyone knows that cybersecurity is crucial; you leave data unprotected at your peril. But what about OT? The boundary between IT networks and OT networks isn’t as distinct as it once was, and that means protecting your OT is now also incredibly important. Think of the environments where machines are linked to physical safely, like factory floors—or hospitals. But how does a plant manager go about preventing cybercrime?

Part of the solution may be in FPGAs. And if you’re not already familiar with the term, you will be soon. We’ll talk about FPGAs, among other things, with Louis Parks, CEO of Veridify Security, a developer of security IP and tools; and Mark Frost, FPGA Security, Communications, and Configuration Technical Marketing Manager. They’ll discuss the challenges of OT security, the role of FPGAs in addressing those challenges, and even some non-cost options for protecting your OT network right away. Because the vulnerabilities are everywhere, and the bad actors will find them.

What are some of the challenges you see in the security landscape today?

Mark Frost: It’s clear that there’s been a rapid expansion of connectivity between devices. We’re seeing the boundaries of OT and IT networks blurring, and I think many people have just not considered the security implications there, especially of older networks connected to newer ones. The prevailing thought seems to be, “This seems to be working, so everything is okay.” But we’re seeing more and more cyberattacks these days, so it’s something people need to be paying attention to now. 

What makes OT security especially challenging?

Louis Parks: The short answer is that OT networks have been around for decades, but they’ve typically been naturally air-gapped, or disconnected, from the outside world; they’re running buildings, industrial sites, etc. On the other side of the picture, IT networks have always been developed and defined in a very secure fashion—firewalls, VPNs, malware detection—because of the perceived value of the data in HR, accounting, patient records, etc.

The challenge is that we are all looking to better utilize platforms, buildings, industrial PLCs, etc., so these things are now being connected to IT and OT networks for better visibility. I want to look at my building in Chicago from San Francisco or from New York. Now you’re connecting a very secure platform, your IT network, to a very insecure platform, your OT network. You’re increasing the attack surface, and that’s all a hacker really needs.

The other challenge is that, unlike the very homogenous IT environment—a Windows, a Linux, an Apple environment—in the building world there are many different protocols from many different vendors. Also, you’re working on 32-bit or smaller devices that have little or no room for security, and yet are gateways into the system now.

Why do the solutions available today fall short when it comes to OT?

Louis Parks: What are the problems? First of all, at a very high level there could be two really different goals. In the IT space the goal of cybersecurity is to protect data and keep the devices on the system under control. In the OT world the goal might just be to keep things functioning—think of a hospital. In a utility operating in an OT world it could be safety. IT security is a pretty mature market space. So guess what? A lot of the OT cybersecurity solutions we see come from the IT market space, and there’s a divergence between what the IT products that are entering the OT space are made to do, versus what the OT world actually needs.

I want to add the important point that anything you’re doing to secure your network is a good thing. But the typical security tools we see are primarily network based; in the IT world it’s pretty important that an IT director or CESO knows if you brought a device from home and plugged it into the network. A big no-no in a lot of operations. In the OT world they’re not really thinking about people bringing thermostats from home and plugging them in.

The IT tools now being used typically do give you visibility, which is a good thing. They give you monitoring, detection, and alerts; but they’re not protecting the data—because nobody would think of transmitting open text on an IT network—and they’re not stopping attacks. Also, if you do learn of an attack, in the IT world you have network people, IT people, sitting there ready to respond. In the OT world you may be calling a plant or building manager and saying, “Hey, on the 23rd floor we see unidentified data traffic on your HVAC system.” Not really actionable by them.

“On the #IT side: patches, firmware updates—those are weekly events. In the #OT world in some cases they’re nonexistent. So the ability to move the processing to the edge with an #FPGA is huge” – Louis Parks, @Veridify via @insightdottech

Please explain what FPGAs are and the role they play here.

Mark Frost: FPGA stands for Field-Programmable Gate Array. It’s a bit of custom hardware that you can program and set up in certain ways. We see use cases in very high-speed applications where people have got super high-speed data they want to process, or in applications where you need very low latency or high determinism. Often industrial applications will have those particular requirements.

Also, FPGA is really good for custom IO. For example, if you want to interface to this MRI machine over here and this motor drive over there, you can’t buy something off the shelf to do that; that kind of interfacing application needs some custom hardware, and that’s where FPGAs really shine.

We see them across all the applications, but again, particularly in the industrial space. In my group we’ve tried to think about how FPGAs can be suitable in this application, so some of our solutions have an industrial bent to them. We’ll think about things like functional-safety applications, and also longevity; OT networks are often designed and installed to last for 20 years. And the nice thing about the FPGA is that you can update it in the field; should some security hole be found, you can update that in the FPGA as well.

Louis Parks: On the IT side: patches, firmware updates—those are weekly events. In the OT world in some cases they’re nonexistent. So the ability to move the processing to the edge with an FPGA is huge. And that’s one of the powers of an Intel^® FPGA—we don’t have to guess at everything today and just hope we’re good for the next 10, 15 years. We can address it.

What security strategies are you seeing out there?

Mark Frost: In the past, FPGA people have tended to rely on a security concept that is “security-through-obscurity.” FPGA was quite a niche product then, and there was no real published data about, say, how to properly configure the devices. However, there’s been massive growth recently, and people using FPGA devices now really need to start considering their security policies.

But it’s kind of mixed. Some people have gone to town, and they tend to be the ones who have really big security teams to work with. We have other customers with really small teams—maybe they just have one engineer to do everything. So the question is, how do we make it easier for them to start implementing some basic security features?

Louis Parks: Another issue we see is that the building or plant manager responsible for the system may not have the security background for protecting data that the ITs have. But the IT side may not see it as their purview to protect the HVAC system.

Another thing is that network segmentation—which is a common response from these network tools—works well in the IT world: I’ve got a bad data situation on a server here, and I can isolate it until I either replace the server or move the operations over. But if it’s operating a portion of a hospital, I may not be able to isolate it the same way.

So our focus at Veridify has been to take something like an Intel^® FPGA, and provide security at the edge to protect the devices. We are also trying to give a proactive solution that doesn’t require the replacement of previously installed technology. An Intel^® FPGA running our technology can be placed in front of a device as a security gateway, providing all of the authentication and data encryption you’d expect on an IT network, but running it almost like a VPN over an OT network.

How do Intel^® FPGAs help support IoT security efforts?

Mark Frost: Our main task here is as an enabler. We’re trying to be a jack-of-all-trades across all these different verticals, so our devices are designed for many markets. We try to think about basic device features that will support security—that foundational base—that then people like Louis and his team can plug into.

But we have to do the right thing in the foundational base; we have to have the right hooks into the device. We have to think about things like functional safety-data packages, for example; about specific silicon features, real-time processing, and all this other stuff that Veridify can then build upon.

Louis Parks: Intel has a focus on security, on securing firmware data—things running on the FPGA. We then extend that by looking at how a device interacts with the devices around it, which is our focus. That communication between devices creates, in essence, what we often think of as the IoT.

How can organizations be successful at addressing OT networks?

Louis Parks: There is a range of monitoring tools out there that will give you the ability to look at your network, and that network-level strategy will give you some capability. There are also protocols out there that do add security; but some of those can be difficult to implement, because implementation or management isn’t necessarily seen as a priority for them.

Ours is a device-level focus; we’ve basically packaged cybersecurity in a box. When you plug in one of our edge devices, it auto-onboards. Recognizing that there’s limited availability of IT and cyber skills—particularly in the field or at the edge—we’ve done a zero-touch process. With Intel as a partner, that has been our focus.

But there are some non-cost things you can do today. Think about: Where are my risks? What would the risk be if somebody entered this part of my network? Is that critical or not? So make an evaluation. Do you have a backup of your OT network—your building system, your plant/factory? Nobody would ever think of not having an IT-network backup.

You mentioned a partnership with Intel; what has been the value of that relationship?

Louis Parks: We have a lot of expertise—my partners are mathematician cryptographers—so we bring that, plus engineering to the lab. But our product, DOME, came into being because Intel came to us and said, “There is a challenge in how devices at the edge of a network are managed, and we think you have a platform that could solve it.” So they not only bring us to opportunities and sectors, but, more importantly, they expose us to the issues that we then have to address.

And there’s no substitution for the reach and the depth of the Intel team. Their support in developing protocols, in developing solutions, and in helping us bring those to the market space—it’s invaluable. Thank you, Mark!

Mark Frost: You’re welcome. And, from our point of view, our primary aim is to sell silicon. We rely on partners like Veridify, because we can’t do that without these really cool solutions that appeal to the marketplace. We do what we can to help extend the reach of Veridify into these global markets through our sales networks and our channel networks, but it’s the work of these guys and their solution that’s the exciting part, really.

Any final thoughts you’d like to leave us with?

Louis Parks: I think everybody should be thinking about security. Unfortunately, security threats are a potential in the world we live in. I think people should look for solutions, and understand not only what those solutions can do for them, but also their own ability to use and manage them. Also, no single solution is going to do it all. So even when you’re successful, please keep working, keep looking; this is an ongoing process.

Mark Frost: I agree—don’t ignore security stuff. Things as simple as authenticating and encrypting the configuration data for your FPGA are two very simple things to do that could make a huge difference to the security of your FPGA implementation. Many of the most high-profile security breaches we’ve seen in the past were quite innocuous at the start, and still many people are thinking, “It’s not going to happen to me.” We’re here to help.

Related Content

To learn more about OT security, listen to Demystifying OT and IoT Security and FPGAs: With Veridify. For the latest innovations from Veridify, follow them on Twitter at @Veridify and on LinkedIn.

This article was edited by Erin Noble, copy editor.

Data processing automation has seen numerous advancements in recent years—in applications as diverse as autonomous vehicles, eldercare nutrition, smart self-checkout, and factory automation. But certain data management workflows remain stubbornly manual.

A prime example is accounts payable (AP) invoice processing. The AP department is responsible for several critical, time-sensitive businesss functions such as receiving and reviewing incoming invoices, validating invoices and scheduling payments, and entering financial data into the enterprise resource planning (ERP) system. Yet in enterprises all over the world, AP teams still do much of this work by hand.

For businesses committed to digital transformation, it’s a frustrating situation—and a costly one.

“If you’re processing AP invoices manually, you’re not only spending a lot of time and effort on the work itself, but you’re also making mistakes,” says Anil Nair, Founder and CEO of GIBots, a provider of AI data processing automation solutions. “That gets expensive, in particular when you factor in penalties for past-due payments.”

In addition to the direct costs of processing invoices manually, there is a business intelligence cost as well.

“Without control over AP data, organizations don’t have a real-time picture of their finances. That makes it difficult for leadership to make fully informed, data-based decisions,” Nair explains.

Enterprises are aware of these problems, of course. But AP automation is easier said than done. Optical character recognition (OCR) technologies have historically been far too inaccurate for automated invoice processing. Modern GPU-based OCR systems offer one possible option, but they tend to be slow and expensive.

But advancements in AI and ML, along with next-generation CPUs built for computer vision tasks, enable a new breed of data processing automation solutions. These efficient, cost-effective systems promise to bring AP invoice processing into the 21^st century—and deliver a host of benefits to companies and their suppliers.

The Key to No-Code AP Automation

Effective AP automation is the result of combining AI/ML-based OCR and CPUs designed to handle visual processing workloads. GIBots’ DigiDoc AP automation solution, for example, uses proprietary OCR algorithms that run on Intel^® CPUs, which can optimize computer vision inferencing.

The resulting solution is faster, more accurate, and more cost-effective than older versions of OCR or current GPU-based systems—and is reliable enough to enable no-code deployments for end users, according to Nair.

Advancements in #AI and #ML, along with next-generation CPUs built for #ComputerVision tasks, enable a new breed of data processing automation solutions. @Gi_Bots via @insightdottech

In practice, the system is also very easy for AP teams to use. It captures a photo of an invoice or extracts invoice data from an email attachment, and then sends the data to a secure cloud environment for validation and processing. End users can access the results via a web portal or their organization’s own ERP system.

The seamlessness of the workflow is the result of extensive development and testing. Nair cites GIBots’ technology partnership with Intel as a major factor in their success.

“We spent months with Intel’s technical teams fine-tuning the algorithms and optimizing the platform’s performance. Intel has very high standards for efficiency, and set ambitious targets duing this process. That was the biggest factor in the success of our initial development work,” he explains.

The resulting solution has proven to be highly accurate in deployment—even when faced with complex data processing tasks.

Government Achieves 90% Straight-Through Processing

GIBots’ experience with a government in Europe is a case in point.

The government’s AP team needed to process a steady stream of invoices coming in from mulitple vendors in different countries. Due to oversight and compliance requirements, invoice validation was complicated. Nearly 60 business rules needed to be applied to each invoice entering the system. Adding to the difficulty was the international nature of the work: The invoices were arriving in different languages, currencies, exchange rates, and even data formats.

GIBots implemented its AP automation solution and integrated it with the government’s IT infrastructure and ERP system. The system was set up to identify invoices sent to specific email inboxes, extract information from the attached invoice, validate it using the government’s extensive business rules, and push the data to the ERP system.

The result was automated invoice processing that surpassed expectations for speed and accuracy. “As soon as a vendor’s email comes in with an invoice, the ERP system is ready for payment in a couple of minutes,” says Nair, adding, “Over 90% of new invoices are now handled via straight-through processing.”

The Future of AP Automation

The benefits of AP automation are quantifiable—and impressive. Switching from manual to automated invoice processing can result in cost savings of 78% and can shorten the average invoice processing time from 8.3 days to 2.9 days.

Vendors, of course, appreciate faster payment; and everyone benefits from improved trust, better business relationships, and an increased speed of doing business. More efficient invoice processing also reduces the risk of costly overdue payments, while automation ensures traceability, which helps to combat fraud.

But the most important benefits of automated invoice processing are yet to come, says Nair: “When AP teams are freed up from doing repetive, manual processing work, they can turn their energy to solving bigger problems. And when decision-makers have a clearer picture of their organization’s finances, they can make smarter choices and reduce risk around cash-flow—which is especially important during turbulent economic times.”

Perhaps most tantalizingly, Nair sees AP automation as something that will help businesses of all sizes. “Large enterprises have already moved toward data processing automation. But as affordable, no-code solutions come online, we’re going to see smaller, growing enterprises benefit from AP automation as well,” he says.

This article was edited by Christina Cardoza, Associate Editorial Director for insight.tech.

The cyberthreat landscape changes rapidly. With new threats emerging all the time, it’s hard for any organization to effectively defend itself. Often, companies must walk the line between fast and easy vs. robust and comprehensive in their efforts to secure the business without slowing it down in the process.

It doesn’t have to be that way, says Courtney Radke, Field CISO – Retail and Hospitality at Fortinet, a global cybersecurity leader. It is possible to have network performance and efficiency while still maintaining a strong security posture. The trick is making security a priority from the ground up and simplifying management as businesses expand their ecosystem with new branch locations, remote sites, and even home offices.

Too often, security becomes an afterthought in the march toward digital innovation. This leads to “bolt on” cybersecurity solutions as the network expands, which effectively limits a business’ ability to mitigate risk. In highly distributed environments, this also creates complexity, hinders visibility, increases the likelihood of vulnerabilities, and significantly reduces the ROI of technology investments.

But as C-suites and boardrooms become more attuned to cybersecurity needs—often a result of headline-grabbing breaches—organizations are taking a new approach. “They need something smarter, more consolidated,” says Radke. “They need a whole solution set; a platform to solve all the problems they can see today and the ones they can’t even anticipate. And they need it to scale.”

Fortinet addresses these needs with its Secure SD-Branch solution, which takes a platform approach to delivering enterprise-grade networking and security for the branch. By converging network and security through FortiOS, organizations gain access to benefits such as Next-Generation Firewall, Zero Trust Services, IoT protection, and integrated switching and wireless management via FortiLink—all in a single device.

Fortinet Secure SD-Branch also incorporates native software-defined wide area networking (SD-WAN), tightly coupled with next-generation firewall protection, to provide secure, “always-on” connectivity for today’s changing business needs. With templated configurations, zero-touch deployment, and centralized management and visibility, Fortinet Secure SD-Branch is also simple to manage at scale.

As Digital Transformation has made the branch more complex, Fortinet Secure SD-Branch provides comprehensive security at the network edge, and addresses the expanding attack surface, by consolidating network and security within a single platform that provides visibility and protection to the branch and all users and devices connecting to it.

Edge Security at Work

For Checkers Drive-In Restaurants, the SD-Branch approach was just what the company needed to manage its 265 locations across 38 states. Checkers was looking for a solution that could secure its many locations at the edge, scale as the company grows, and deliver centralized visibility. Its legacy SD-WAN platform couldn’t support the organization’s requirements.

The company decided to replace its outdated system with Fortinet SD-Branch’s security-driven networking approach. Specifically, Checkers deployed Fortigate Secure SD-WAN, which combines next-generation firewall with network orchestration and acceleration.

For Checkers, it meant simplified network management with fewer resources—lowering costs and improving productivity.

The proliferation of endpoints, including #PCs, smartphones, and #IoT devices, as well as the convergence of #IT with #OT, has resulted in an “explosion of edges.” @Fortinet via @insightdottech

Cybersecurity for the Expanding Edge

Like Checkers, just about any enterprise with a growing, distributed environment wants scalability and visibility. For many, the proliferation of endpoints, including PCs, smartphones, and IoT devices, as well as the convergence of IT with OT, has resulted in what Radke calls an “explosion of edges,” which strains network resources, lessens security effectiveness, and complicates management.

When organizations turn to Fortinet, Radke says, they are looking for simplified management and better cybersecurity solutions that scale to meet the business. Fortinet’s job is to understand business outcomes that customers seek and enable them while delivering consistent user experiences.

But most important, organizations are looking for stability and resilience. “It’s about maintaining business continuity if the worst were to happen,” says Radke. “How do I continue to meet customer demand? How do I keep my business moving forward? How do I keep pace with the competition in a hyper-competitive market?” Fortinet allows organizations to execute on their business objectives and digitally differentiate themselves securely and with confidence.

Consolidating Security Unifies System Management

Key to Fortinet’s approach is its partnership with Intel. The company uses Intel processors that support the scalability requirements of its solutions. “The beauty of the Intel and Fortinet partnership is scale, and that we’re in so many organizations together,” Radke says.

Together, Fortinet and Intel help customers secure the edges. At the same time, the SD-Branch technology helps lay the foundation for a new security approach often referred to as SASE (Secure Access Service Edge). The model calls for consolidation of cloud-based and WAN security to enable uniform management no matter where users or devices reside.

SD-Branch also sets the stage for moving toward a “Zero Trust” approach, a security framework that acknowledges that securing network egress and ingress is not enough. Lateral movement inside the network must be secured via microsegmentation, identity must be established, and access to resources must be based on least privilege—meaning users get access only to resources required for their roles within the organization.

Fortinet solutions, including the Fortigate NGFW, incorporate threat intelligence from FortiGuard Labs, Fortinet’s cybersecurity threat intelligence organization. FortiGuard Labs continuously monitors the worldwide attack surface using millions of network sensors and hundreds of intelligence-sharing partners to provide timely intelligence, product updates, and help customers better understand and defend their threat landscape.

Fortinet’s portfolio of more than 50 security technologies is designed with integration and automation in mind to share threat intelligence, correlate data, and mitigate threats across any network, endpoint, and cloud in real time. This allows customers to move beyond point products, gaining better ROI on technology investments, and accelerating business outcomes via consolidation and convergence of network and security.

Edited by Georganne Benesch, Associate Editorial Director for insight.tech.

The Internet of Things has advanced to the point that businesses across all kinds of sectors are realizing its value and looking to reap the benefits. But due to its wide-ranging use cases across industries, optimizing performance, cost, and efficiency has meant intelligent edge and network infrastructure needed to be developed and implemented from the ground up.

Thankfully, there are new and innovative edge-to-cloud platforms reaching the market that are designed to help unify IoT, such as the latest 4th Gen Intel^® Xeon^® Scalable processors, 13th Gen Intel^® Core^™ processors, and Intel Atom^® processors x7000E Series—all of which release in first quarter of 2023. These latest releases are designed to incorporate the power, efficiency, security, connectivity, AI abilities, and other features needed to successfully deploy edge solutions, while delivering step-change improvements over previous generations.

4th Gen Intel^® Xeon^® Scalable Processors Adapt to Workload Consolidation Demands

Originally code-named “Sapphire Rapids,” the 4th Gen Intel^® Xeon^® Scalable processors are some of the most versatile products available to IoT hardware engineers, AI software developers, and systems architects today. With up to 52 cores, and long-term availability, IoT variants of these processors deliver a 1.33x performance increase compared to their predecessor.

In addition, the latest release features AI inferencing improvements, such as a 3.01x performance increase for image classification and 4.25x increase for object detection, thanks to a built-in AI acceleration engine called Intel^® Advanced Matrix Extensions (Intel^®AMX). Part of the Intel^® Deep Learning Boost (Intel^® DL Boost) suite of technologies, AMX is a hardware block designed specifically for matrix multiplication tasks common in AI training and inferencing.

Intel partners are already taking advantage of these new features. For example, leading medical-technology company Siemens Healthineers leverages AMX with the OpenVINO^™ toolkit to improve the accuracy of oncology-radiation imaging diagnostics.

The 4th Gen Intel^® Xeon^® Scalable processors are some of the most versatile products available to #IoT hardware engineers, #AI software #developers, and systems architects today. @Inteliot via @insightdottech

Of course, one of the primary performance bottlenecks for AI workloads is memory. To ensure IoT developers can maximize the performance of features like AMX, the latest 4th Gen Intel^® Xeon^® Scalable processors support DDR5 memory, up to 50% more bandwidth vs DDR4. Intel^® Resource Director Technology (Intel^® RDT) directs traffic coming in from external memory devices, accelerators, and other peripherals to on-chip processing resources so these data flows remain streamlined, while Intel^® Scalable I/O Virtualization (Intel^® SIOV) enables efficient resource sharing across many virtual machines in workload-consolidation use cases.

All of this is protected by integrated security features—such as Intel^® Software Guard Extensions (Intel^® SGX), Intel^® Total Memory Encryption (Intel^® TME), and Intel^® Platform Firmware Resilience (Intel^® PFR)—that keep the chipsets and their data safe in connected environments.

Closer to Edge with Intel^® 13th Generation Core^™

But as you get closer to industrial IoT controllers and actuators, more tactical solutions may be in order.

For PCs and gateway-class systems, 13th Gen Intel^® Core^™ processors (code-named “Raptor Lake”) take a page from the hybrid 12th Gen Intel^® Core^™ processors’ microarchitecture. The 13th Gen Intel^® Core^™ processors package up to 24 Performance and Efficient cores and 32 threads flanked by Intel^® UHD Graphics 770 engine, while the 13th Gen Intel^® Core^™ Mobile processors feature up to 14 cores, 20 threads, and Intel^® Iris^® Xᵉ graphics, media, and display engine.

The 13th Gen Intel^® Core^™ processors also go a step further than their predecessor by delivering 1.04x faster single-thread and 1.34x faster multithread gains in a similar 35W to 65W power envelope, while the mobile processors deliver 1.08x faster singe-thread performance and 1.05x faster multithread performance.

Like the new 4^th Gen Intel^® Xeon^® processors, the 13th Gen Intel^® Core^™ processors make use of Intel^® DL Boost technology, add PCIe 5.0 connectivity, and include DDR5 RAM support to keep data flowing across edge environments. Resources are provisioned across the chip using Intel^® Thread Director.

You can find all these capabilities put to work in the real-time, zero-defect manufacturing system of machine-vision solutions company Eigen Innovations, which helps automation companies not only monitor and predict the maintenance needs of process control equipment but also deploy more cameras per processor.

An additional breakthrough for IoT engineers comes with new features like real-time capable 2.5GbE interfaces and Intel^® Time Coordinated Computing (Intel^® TCC) CPU features that allow devices to delivery deterministic performance in areas like machine control.

Intel Atom^® x7000E Processors Power IoT Efficiency

Even closer to (and, in some cases, in) IoT endpoints, Intel^® Core^™ processors and Intel Atom^® processors x7000E Series (code-named “Alder Lake-N”) have adopted Efficient cores with new instructions for better performance at low-power envelope. Integrating the same Efficient cores as the 12th Gen Intel^® Core^™ processors per device, the Alder Lake-N portfolio improves performance across the board, with 1.30x faster single-thread and 1.09x faster multithread performance, 1.68x faster graphics, and 6.85x faster AI inference performance.

With the help of up to 32 Intel^® UHD Graphics execution units, that’s enough performance to drive three independent 4K60 displays. Or, thanks to Intel^® DL Boost technology and support from Intel^® Advanced Vector Extensions 2 (Intel^® AVX2), IoT designers can execute deep-learning inference workloads at the edge.

Both 13th Gen Intel^® Core^™ processors and Intel Atom^® processors x7000E Series support a wide variety of operating systems; work with the Intel^® oneAPI Toolkit and the OpenVINO^™ toolkit; and include embedded SKUs and long-term support options. 13th Gen Intel^® Core^™ Mobile processors also include select SKUs that support industrial use conditions.

All of this helps lead the next big phase of edge productivity. In the past, foundational solutions were required to reach the performance, efficiency, connectivity, security, and ease of use of edge native applications. To be effective, those solutions couldn’t be adapted from existing ones.

With the introduction of 4th Gen Intel^® Xeon^® Scalable processors, 13^th Gen Intel^® Core^™ processors, and Intel Atom^® processors x7000E Series, it’s now possible.

To see these processors in action, look for them on display throughout the year from Intel and its ecosystem partners at upcoming tradeshows like CES, Mobile World Congress, embedded world, Hannover Messe, and others—and be sure to check back at insight.tech for the latest launch updates.

This article was edited by Christina Cardoza, Associate Editorial Director for insight.tech.