The Arm Blueprint team is attending Arm DevSummit, a three-day virtual event that provides a place for the industry to learn, connect and develop. Over the next three days, and we’ll be bringing you key takeaways from the keynotes, innovation tech talks and ‘fireside chat’ group sessions, alongside our thoughts on some of the technical sessions. Missed a session that sounds interesting? Sessions will be available on-demand 2 hours after they air for the next 30 days.
Tuesday October 6
7 am PT: We’re gearing up for the opening keynote from Arm CEO Simon Segars at 8 am PT.
8am PT: Arm DevSummit 2020 is go! In his opening keynote (Unleashing the World’s Technology Potential on Arm), Simon announces that 180bn Arm-based chips have now been shipped by our partners. He shares some amazing case studies, such as how Arm-powered AI is being used to spot COVID-19 symptoms based on images of lungs, which demonstrate how the Fifth Wave of Computing (the way Simon describes the confluence of Artificial Intelligence (AI), 5G and the Internet of things (IoT) is maturing at pace, creating a global cyber-infrastructure.
Simon then introduces three pillars that define the Arm ecosystem in 2020. Watch the keynote for the detail, or read Simon’s own blog: Arm DevSummit: Unleashing the World’s Technology Potential.
As part of the second pillar, Platform, Simon launches Arm SystemReady, a new program bringing a level of consistency across a broad range of Arm-based devices in the cloud, in the network and in high-performance IoT (HPIoT) endpoints. It will tackle the common software stack as well as system architecture.
Finally, Simon announces that Arm is expanding its partnership with Microsoft to accelerate the deployment of AI-enabled IoT devices that just work. We’ll hear more about this collaboration from Microsoft later this week.
8:18am PT: Simon is now joined by Rene Haas, President of Arm’s Intellectual Property Group, and Jensen Huang, CEO of Nvidia (A Fireside Chat with Arm CEO, Simon Segars and NVIDIA CEO, Jensen Huang).
This explores some of the big questions that many in our ecosystem have around the future of Arm+Nvidia.
Jensen shares his vision for enabling Nvidia’s own IP to become part of the Arm business model, and how Arm’s own GPUs and NPUs fit alongside Nvidia’s own designs.
“The most important thing that software developers want is a platform that is vibrant and rich and growing” says Jensen. “We’re going to combine two of the most exciting ecosystems in the world.”
8:40am PT: In his session The Software Side of Arm, Mark Hambleton—who leads the open-source software team at Arm—explores how today, we’re moving closer to the goal of easier software development that maximizes code optimization and compatibility across platforms and across a diverse ecosystem of software.
In this spirit, he describes the PSA Certified program, founded by Arm and embraced by industry, that outlines a set of design considerations developers need to be mindful of to ensure security. Complementary to this is Arm SystemReady, which helps ease device design and compatibility.
Both of these, he says, will reduce the amount of low-level bespoke code that needs to be written, which, over time, should reduce total cost of ownership should you adopt them.
Hambleton calls on Arm partners to engage with standards, consider their lifetime technology commitments (not just the device itself but how the device will be managed) and look at what technologies add value to your products and which can be handled by common frameworks.
8:55am PT: In the session Inside the Cloud, Arm’s Chris Bergey discusses the opportunities and challenges for running cloud-native applications on Arm with key industry executives.
Two engineers + two weeks = 20% better performance and 20% lower cost. That, roughly, is the equation prompting a growing number of customers to run workloads on Amazon Web Services M6g instances, i.e. the services powered by Graviton 2, the Arm-based server processor designed by AWS.
By investing a little time and a few resources on the sly, companies are finding that they can quickly shift workloads to Graviton 2 and increase their price/performance by 40% with half of the gains coming from lower costs and half coming from more throughput. (To further help the process, AWS’ Dave Brown also announces that they’re giving customers 750 free hours on a new Graviton 2 instance for test runs).
Liz Fong-Jones, principal developer advocate at Honeycomb.io, equated the gains to an extra month of runway for startups. Are we at a tipping point for cloud architectures?
9:15am PT: Content streaming used to be fairly straightforward, explains Dianne Marsh, Director of Device and Content Security at Netflix, in her innovation tech talk titled Challenges of Addressing Multi-tenancy in the Device Ecosystem. Shows used to be delivered to fixed-function boxes produced and ultimately managed by, respectively, a small number of OEMs and TV operators.
Now, consumers want to be able to download, stream and/or tune into live broadcasting from a broad plethora of devices. Security and credential standards vary across the globe and so do the practices of manufacturers and integrators.
But, as she explains, “consumers really aren’t interested in hearing about our complexities. They just want it to just work.”
9:30am PT: We’re now into the technical sessions. There are multiple concurrent sessions for each track—all can be watched later on demand. Here are some of our highlights…
- Arm’s Rod Watt and David Fritz, senior director for autonomous and ADAS SoCs at Mentor, A Siemens Business, offered a fascinating look at evolving autonomous vehicles in A Digital Twin for the Development of Autonomous Ready Systems.
- In the Teens in AI Versus COVID Hack Winners: All-In session, we meet Andreea Giurgiu, Hritika Kaila and Suhaas Mairal from All-In. “If you want to leave some mark on this world as a legacy, the best way to do that is through technology” says Hritika.
- When it comes to Building Apps for Windows 10 on Arm, it’s nowhere nearly as hard as you might think, says Marc Sweetgall, sensor program manager at Microsoft. Microsoft Edge, Firefox and Visual Studio Code run natively on Arm64 and, while 32-bit x86 apps now run through emulation, recompiling for Arm64 and updating dependencies takes only a small amount of work.
- Arduino founder Massimo Banzi presents A Low Code Approach to Secure and Productive AIoT Application Development, building an end-to-end TinyML IoT application on an Arduino Portenta board (running on mbedOS and CMSIS) communicating with the Arduino IoT Cloud. Banzi recently blogged on Arm Blueprint about his work—it’s a great read.
- In Reinventing Live Communications and Collaboration Around AI Speech, well-known entrepreneur Chris Rowen explores the ultimate human interface: speech. It’s an interface that’s used by 7.5bn people (yes, the population of Earth..) for hundreds of thousands of years, yet it remains a complex electronics problem—especially in noisy environments. How are neural networks helping to improve what the computer hears?
Wednesday October 8
8am PT: We’re back for the day 2 keynotes of Arm DevSummit 2020. Right now, Arm IP president Rene Haas is taking a moment to thank our silicon partners who have shipped 180bn Arm based silicon chips in the past 30 years.
In his keynote The Future is Being Developed on Arm, Rene describes how a new, holistic approach to design is enabling Arm partners to develop the future together, now. With the goal of deploying Arm everywhere data happens, Haas describes Arm Total Compute—a solution-based approach for delivering system-level performance, end-to-end security and complete developer access.
Total Compute is applied across the entire Arm IP portfolio and includes software frameworks across all applications. For example, developers can use a single tool suite for performance analysis, to maximize software performance. The Total Compute approach also includes Project Cassini, the foundation of cloud-native development.
Haas said Arm IP represents the lowest-risk, easiest-to-deploy technology that “just works.” It’s a seamless experience for hardware and software developers.
Rene also announces that more than 60 partners have signed up for Arm Flexible Access, the majority of those being startups.
Paul talks about the evolution of smartphone apps, leading to a revelation: Arm Cortex-A “big” cores will only support 64-bit code from 2022. From AI to immersive mobile gaming, this is a major enabler for the Android ecosystem. What does that mean for developers? Read the blog What 64-Bit Android Apps Mean for the Future of Mobile here on Arm Blueprint.
Make no mistake, says Paul: mobile is where personal computing happens. New technologies such as AR and VR to are gaining their foothold in the mainstream through mobile and, as Evans Data Group notes, 13 million of the world’s 24 million professional developers are working in mobile.
Success, however, brings its own challenges. Namely: how you take advantage of everything that’s there? A big part of the Total Compute strategy revolves around making access to the underlying power of the silicon easier for developers, notes Paul.
And Memory Tagging Extensions, or MTE, a feature coming to new silicon, will identify potential memory safety issues before products get to market. Ralph Hauwert, SVP of R&D at Unity, also provides insight on what they’re doing to accelerate creating 3D content for mobile.
8:30am PT: We now join Arun Kishan, Technical Fellow, Microsoft, for his keynote Microsoft: Building for an Arm Ecosystem.
It’s an Arm Conference: So Why Is Everyone Discussing PCs?
Processors from the Arm ecosystem are pervasive in IoT and mobile, running critical services at some of the largest cloud providers and now, with increasing regularity, showing up in Windows on Arm laptops from leading vendors like Acer, Samsung and HP. Chalk it up to customer demand (for extended battery battery life), technical advances (at the silicon level) and a growing body of software.
Arun Kishan, a Microsoft technical fellow who oversees the development of base operating systems at Microsoft, walks through what Microsoft is doing to enhance its offerings in the cloud, at the edge, and on the notebook.
Microsoft is investing in a range in tools to make it easy for developers to write software for Windows on Arm devices. “The vast majority of existing 64 bit applications can easily be ported to Arm architecture”, he says.
9am PT: Over to Greg Bowman, Associate Professor of Biochemistry & Molecular Biophysics at Washington University and Director at Folding@home for his Innovation Tech Talk, Citizen Scientists Create an Exascale Computer to Combat COVID-19.
20 years ago, the Folding@Home distributed computing project was founded to harness compute from computers around the world. At the beginning of this year there were around 30,000 computers lending their processing power to solving complex problems.
Then Covid-19 arrived, and since then there’s been an explosion of interest in using Folding@Home to target the virus. So much so that this distributed computer has just cleared the exascale boundary.
The fight against Covid-19 might ultimately depend on the whether or not scientists can take advantage of the above jittery molecular motion depicted at around the twelve minute mark of Dr Bowman’s talk at Arm DevSummit.
The simulation shows the crucial moment when the Covid-19 virus exposes its ACE2 receptor used to latch onto a host, according to Bowman, an associate professor in the Department of Biochemistry and Molecular Biophysics at Washington University. Blocking the receptor with small molecule therapeutics could mean blocking a pathway for the virus.
The simulation in part was built through data collected through Folding@home, the distributed computing project. Over 1 million people are donating spare smartphones and PC computing cycles to help model Covid-19, Ebola and other viruses. With an exaflop capacity, Folding@home has arguably become the world’s biggest supercomputer.
And now, Folding@Home is actively working with Arm to launch an Arm version of its software, extending its invitation to join the world’s biggest supercomputer to every Arm device.
9:30am PT: We’re now into the technical sessions for day 2. There are multiple concurrent sessions for each track—these can be watched later on demand by clicking the session name below or logging into the DevSummit platform.
- In an era of ever more complex systems, it’s crucial that elements of any system “just work,” especially in the infrastructure. That’s the key takeaway from the appropriately-titled Making Arm Devices “Just Work” Panel session. Arm Fellow Dong Wei led a panel of experts that discussed the newly-launched Arm SystemReady, a foundational compliance-certification program whose goal is to ensure that software works seamlessly across the vibrant diverse Arm ecosystem.
- In How Arm Technology Will Help The World to Build Back Smarter Post COVID-19, Simprints’ Christine Kim explained how the company has used facial and fingerprint recognition to develop an application for creating electronic records for individuals that otherwise might otherwise slip through the grid. The results speak for themselves: In Bangladesh, maternal health visits during pregnancy increased by 38% while in Malawi health workers were able find the pre-existing health records for HIV patients 81 percent of the time, something that only occurred 50 percent of the time before. (Simprints also developed on cash voucher system for refugees from Boko Haram.) An insight into tech at its best.
Thursday October 9
8am PT: Welcome back to Arm DevSummit 2020 and the Day 3 Keynotes. We join Dipti Vachani, Senior Vice President & General Manager, Automotive & IoT, who announces that together with Microsoft, Arm is optimizing the developer experience with streamlined, scalable tools that put creativity first.
Dipti now talks about The Journey To Autonomous and how Arm’s Project Cassini, alongside new Cortex-R technology, will enable that journey for developers. Arm is focused on proving tools and experiences in both software and hardware that just work.
Autonomous decision making can make a real difference in manufacturing, and Arm is focusing efforts on this sector, says Dipti. You can read more about how Arm is addressing developer needs in deploying next-gen autonomous vehicles in this blog: Arm Enables Next-Generation Autonomous Systems.
Read Dipti Vachani’s new blog on Arm Blueprint
Since then, says the Arm Neoverse roadmap has delivered cost savings, power efficiencies and compute-performance gains that have changed the way data center architects think about compute resources.
Read Chris Bergey’s new blog on Arm Blueprint
Digital infrastructure is one of the biggest opportunities in technology. Annual demand for public cloud services is expected to surpass $364 billion by 2022. The investment in underlying hardware to run those services, meanwhile, is slated to pass $100 billion by 2024.
But, like roads, grids and other traditional infrastructures, it’s not going to be easy.
Technological breakthroughs, agreement on standards and cooperation between companies that may not have ever met before will be required. Chris walks viewers through what Arm partners are doing to lay a foundation for the future and what comes next.
Things like Arm SystemReady, which Simon Segars announced on Tuesday. Building for SystemReady doesn’t add to code size or development cost, says Chris. It means you can write an application once, and have it work on devices from a wide range of manufacturers.
8:35am PT: We’re now into our panel session, How Software Influences the Arm Architecture.
“A feature is for life, not just for Christmas,” says Richard Grisenthwaite, Arm’s chief architect. “The last thing you want to do is put it in and then two years later say, ‘Oh, my god. That was a mistake.’”
Richard, alongside VP of software Mark Hambleton and Jem Davies, general manager of machine learning, provide a behind-the-scenes view of how new concepts get integrated into the roadmap, the role partners play and some of the upcoming features.
The panel discusses the Arm Memory Tagging Extension (MTE) and its vital role in protecting Arm devices from software vulnerabilities, and how ArmNN provides a seamless software layer for running ML workloads on Arm hardware.
“How does someone who’s spent a career as technical leader in racing teams end up working for a subsidiary of VW?” asks Chet.
“I wanted to do something that touches more lives and has more impact on society”, says Alexander. He moved from Porsche to Apple—and that’s where he came into contact with autonomy.
Artemis is a company accelerating the transformation of the Volkswagen group from a car company to a technology company, adding expertise in autonomy, software and robotics.
Asked how pervasive electrification will become, Alexander says it’s a done deal. From a carbon footprint perspective, it is the only option for our automotive future. The question is how quickly it ramps up, as the global automotive industry is enormous.
Efficiency is the name of the game, says Alexander. Much of the R&D around electric vehicles right now is in increasing the range while reducing battery size. Not an easy feat!
Alexander points out how important software is—the ‘software defined car’ is already here, as almost every feature of the modern car involves software. The software stacks in future cars will also require AI and ML as their requirements evolve, and that’s a challenge that Volkswagen is rising to with its Artemis accelerator.
“How can the Arm ecosystem help?” asks Chet. “We need choice” says Alexander. “We need the flexibility to change the mix between hardware and software and for that we need transparency and a stable platform with certain standards”.
9:20am PT: We’re now into the technical sessions for day 3. There are multiple concurrent sessions for each track—these can be watched later on demand by clicking the session name below or logging into the DevSummit platform.
- Pavel Macenauer, software engineer from NXP, and David Steele, Director of Innovation, from Arcturis networks, took the edge conversation deeper in the panel Using Arm NN to Develop Edge AI in the Smart City on using Arm NN to develop edge applications in the smart city. Arm NN is an open-source middleware inference engine for ML at the edge that connects high-level frameworks to compute engines and serves as a single API integrating high-level frameworks, such as TensorFlow, TF Lite, Caffee and ONNX.
- The Machine Learning at the Edge session explored how deploying and scaling software on the edge hasn’t been a problem, but it’s with machine learning models where the challenges come into play, according to a panel hosted by Qeexo on Thursday. It’s important because, according to Vijay Janapa Reddi, associate professor at Harvard, less than 1% of the data that’s available to us is actually being processed and leveraged today.
- In Riding the Next HPC Wave With Arm Processors, Ashwin Matta, director of IP product management reveals how high performance computing is in the initial stages of a massive transformation in how systems are built, deployed and used. Cloud providers are expanding their HPC-as-a-service offerings, allowing more customers to run ornate simulations and other supercomputer-class tasks. National security concerns are prompting research agencies to redouble resources for developing their own processors and systems. And a growing number of designers are discovering the performance capabilities of Arm’s Neoverse platform, which combines Arm’s traditional strengths in performance-per-watt and a robust software ecosystem with HPC capabilities like vector processing through Single Vector Extensions (SVE) and SVE2. The results? 60 percent less processing time and 70 percent lower costs on benchmarks for computational fluid dynamics.
That’s it for Arm DevSummit 2020, but it’s far from over!
Throughout October you can access everything you missed first time around through the Arm DevSummit platform. A huge thank you from all of us for attending this year, and we hope to see you next year, whatever the circumstances may be!