ew 2024: ADLINK Takes Edge AI Innovations to the Next Level

How are your customers embracing today’s edge AI capabilities?

Richard Pinnow: The edge AI is definitely a true mega trend. So for the computer and module part, I have to say it’s super horizontal. So there’s a very broad range of vertical we address typically with our computer and module, but we have some, how to say, we have some success stories, which are very, very interesting and which really leverage this AI technology to take the application and their systems to the next level. And it’s really inspiring, some of these applications, they really make the difference. And a very good example is a healthcare device. Everybody knows this. So whoever had a broken leg ended up maybe in an MRI system, right? And the MRI system is basically scanning a particular area of your body. And what is possible today, well, in the past it was always like, it was a post-production of video data and imaging you’ve captured during your examination. But nowadays, it’s a real life improving. So meaning you on the fly can process the data, you can improve the probes set around the area which is injured. You can modify the position of the patient, and this allows to improve significantly the image quality. And, you know, when it comes to our MRI, it’s not just limited for broken bones or whatever, when it comes to detection of tumor or cancer, right, it’s very beneficial to have a superior image quality. And this is, I think, a very good example that AI technology is really taking performance to the next level. And this is just possible because the number crunching is done in 20 seconds instead of hours, right? And this is a very, very good use case. How you can see that image quality has really an impact on people, on real life people, and it makes the world a better place.

What are the biggest edge AI challenges your customers face?

Richard Pinnow: One of the biggest challenge we see at our customer base is the technology is evolving super fast. So meaning you need to be super fast to the market because otherwise you’re dealing with outdated or old technology once you are ready for mass production. And for this reason, time to market is super important. And this is basically our value proposition. We are providing ready-to-use computer modules, they’re application ready. You can focus on your application layer while the underlying hardware is already ready to use and you’re simply focused on your differentiators. And this puts you really on the fast lane to overcome this problem of fast evolving technology.

How can customers address these challenges using the latest Intel technologies?

Richard Pinnow: We’ve seen from generation to generation, 11th generation, 12th generation, 13th generation, and today at the embedded world 14th generation, we see a lot of improvements when it comes to GPU performance, and GPUs or embedded GPUs and silicon is used for doing these AI workloads. And I think it’s not just limited to the CPUs. As well, what Intel provided is a discrete graphic card solution as well. So I have both with me here, so I have a discrete graphic cards based on the latest and greatest Intel^® Arc^™ technology and I have the Meteor Lake with me, the multi-tier architecture, and it can run standalone, it can be combined with each other. And this is what I mean with ecologies, right? And you are super flexible, and the computer and module approach, not just limited to the CPU, to the GPU as well, takes it to the next level while keeping the pace of being super fast to the market.

What is the importance of using toolkits like OpenVINO^™?

Richard Pinnow: Nowadays, one needs to think more in ecologies or ecosystem. So the whole design in is not just limited to take a piece of hardware which works, right? It’s about to enable customers to use this super fast and great technology and meaning you need to have very powerful and good tool chains in place, really, which allows to leverage all those number crunching abilities that you can really bring this down to, to do the calculation for your application, right? Is it image processing at the edge? Is it sensor processing at the edge? Is it detection of patterns and data at the edge? And a powerful toolkit like OpenVINO^™ is really putting the customer in the fast lane because it’s scalable across multiple CPUs. It’s scalable not just limited to CPUs. You can use the same tool chain on discrete GPUs as well. So it’s not just ICA performance, discrete GPUs, it’s the same for the Atom-level performance. You can use OpenVINO all across those targets and not just limited to one generation. You can easily migrate from one generation to another. And this makes OpenVINO a very, very powerful tool in the x86 interface.

How else is ADLINK meeting the needs of AI technology customers?

Richard Pinnow: ADLINK is providing a lot of different computer and modules. As you can see, this is all the latest and great Intel technology. We are talking here from the smallest compute solution, which is an OSM, which is directly soldered down to the carrier board, a highly integrated solution, up to a very high performance solution on COM Express. So we have full coverage, and as you know, every computer or module standard has its own benefits, upsides and downsides. What we do as a dealing, we consult the customer, right? The customer has a problem and we are checking the customer’s problems to put them on the fast lane. This means, for instance, we are providing reference data that they can build their carrier board in the fast lane. We want the customer to copy our predefined function blocks and to leverage them to their application to again reduce their total cost of ownership and their investment and time and energy to bring this down to their customer base. And ADLINK is very strong in not just providing engineering services. We are very strong in manufacturing as well. So as you see, we have an end-to-end solution, starting from prototyping on reference carrier. The next step is integration on carrier boards, doing full customized system level as a business. And then at the end of course, executing. We are doing the mass production. And with mass production, it’s super important to have the lifecycle management. And we are professionals in this area since nearly 30 years, so we know how to bring initial brilliant ideas into mass production over the entire lifecycle up to 10 or 15 years.

Why is embedded world 2024 important to your company?

Richard Pinnow: embedded world is the most leading affair. We are very proud to have to represent the latest and greatest technology. So Intel is a very strong partner of ADLINK since decades. And from generation to generation, we leverage all those technology to bring the best to our customer base.