Driving Purpose-Built Compute and Specialized Processing

Geof: Good morning. Good afternoon and Good evening. Welcome back to the Arm Viewpoints podcast. And our topic today is Total Compute Solutions: Driving Purpose-Built Compute and Specialized Processing. Our guest today is Paul Williamson, senior vice president and general manager for the client line of business at Arm. The client team defines the compute platform that shapes user experiences in the smartphone, AR, VR, Digital TV, gaming, and laptop tablet markets. Welcome, Paul, can you give us a bit of background on Arm’s success in consumer computing, including mobile and other consumer devices?

Paul: Yeah, so Arm has been in the business of producing processor technology for over 30 years now. But it really kicked off in the consumer space with the advent of mobile phones and their uptake in the late 90s. Nokia in particular, was a big pioneer in using arm in early handsets, little digital screens and keypads. But over the next decade, with the introduction of Arm’s later processors, the Arm11, and the first Arm Cortex A’s and Mali GPUs, early smartphones became enabled with much richer user experiences using Arm technology. And that progressed further in 2011, with new technologies like big.LITTLEe where we brought together big and small processor cores to optimize workloads. And it’s spread even further into adjacent technology spaces as well. But a lot of it was formed by that progression in mobile phones and mobile technology.

Geof: Do you want to talk also a little bit about some of the other technologies like AR and VR and wearables that you’ve been involved in?

Paul: Sure, yeah. So you know, coming from that background in mobile, and over the software and experiences that were developed first, in that segment, have become the industry standard platform to take that technology across a range of other sectors. So we’ve seen our technology from mobile expand out particularly into things like digital television, and set top box where launching apps and interacting with rich content is a similar experience. And you know, the video and graphics are equally important. More recently, we’ve seen that expand into, as you said, AR and VR, were a headset and rich screen is initially a little like a smartphone experience but mounted directly into your field of vision and it is fusing cameras and sort of presenting images at the same time. So common technology platforms have been really useful there. And then wearables is the other end, so going from these richer experiences down to power constrained and body worn devices. Wearables have sort of grown and as you feel seen, you know, the smartphone on your wrist is effectively what you’re seeing when you look at the latest watches that we’re seeing developed now.

Geof: So you talked about a lot of different consumer devices there. How do you see the current state of the consumer compute market at the moment?

Paul: Yeah, so it’s been an interesting recent 12 months, I guess, there’s a lot happening in the industry. And I think what’s really stood out for me has been, I guess, the reliance we have on technology is becoming ever increasing. You know, while our physical worlds have become somewhat more constrained through some of the recent challenges, you know, our, you know, technical world, and our technical horizons is somewhat expanded. So people are growing in their expectations of what to expect from these consumer devices. And they’re expecting to see that experience in all of these devices and see it improve. So yeah, lots of change driven by the recent sort of pandemic and so on in the industry.

Geof: Yeah. And now off the top, we said we were going to talk a little bit about purpose-built compute. And I understand this is a significant focus of the Arm total compute strategy. Maybe you can tell me a bit more about that?

Paul: Yeah, so purpose-built compute for me, it feeds into that trend. As we’ve seen something like the smartphone evolve, it’s brought in a range of different shapes of compute under the hood that serve different purposes in the in the solution for the consumer. So, you know, being able to process camera images in real time and do overlays and and so on. It requires a much more sort of focused SoC that brings together the various capabilities of compute, to deliver a really good user experience. And I think what we’ve seen, particularly in the last 12 months is just an expectation that you’ll, you’ll get that experience on all the devices you use. So rather than having a sort of large box in the compute, in the corner, where you go to for compute, and compute being something you think about, you just expect these rich user experiences, and therefore the system has to adjust to be able to deliver that immersive experience in any device. And so in response to that, we’ve been talking about a new strategy we have called total compute, that brings together you know, three key pillars to try and ensure that that rich SoC can be built and be made appropriate for the different use cases that people want, whether it be a little wearable or sort of future, or augmented reality headset. And that total compute strategy is got three really simple key pillars. One is performance: so driving the ultimate performance, so you get a really responsive product when you’re interacting with it. The second is, you don’t get that performance unless it’s accessible to developers: so you know, if you look at something like a smartphone, the richness of the number of different apps and the different applications we expect to get from that device can only be unlocked if the developer who’s coding for it can access that underlying performance. And then the final one is security: and this is becoming ever more important as we rely on these devices, you know, being able to trust that our data is kept secure, and trust these more important services to these devices across our lives, is ever more important. So those three things come together and allow us to create, you know, purpose built compute solutions with our partners for the different markets they want to play. And so just a few months ago, we launched Armv9, which is the latest generation architecture from Arm and something that comes along only once in a decade. And this defines the technology that will be in compute products for that next decade, it’s going to drive great performance, but also real steps improvement in machine learning and other new technologies as well as security. Now the industry is really recognizing the need and the value of purpose built compute, which involves pulling together a chip that incorporates the right technologies for the market in which it’s being deployed. And allows us to create these well suited mixes of software and hardware to serve the needs of different markets and different use cases within those markets.

Geof: And all of that, I think brings us to today’s announcement of the arm total compute solutions. Maybe you can tell me a bit more about that?

Paul: Yeah, so actually, today’s launch is our biggest launch we have ever done at Arm in our business full stop. But specifically, it’s all targeted for these consumer products that are shaping our lives. And it’s bringing together a new architecture, every decade or so, Arm releases a new architecture and the Armv9 architecture is our latest release that’s really going to define the next decade of compute, and really build on those pillars I talked about of performance and security. And so today’s launch was about bringing the first full suite of our IP in these new compute solutions. That means you can take that Armv9 architecture and deploy it everywhere in the industry. So the solutions target the richest highest performance smartphones all the way up into laptops, and even desktop computing solutions. And all the way down into the smallest, wearable, long battery life devices. So it’s a full suite of our technology to support all of those with the latest Armv9 architecture features.

Geof: Sounds like a major launch Paul, maybe you could dive in a bit more on the underlying technologies?

Paul: Yeah, so for each of the different markets, there’s a range of common technologies that are needed to make the user experience the very best it can be. And perhaps the one that is most significant well known is the CPU itself, the processor at the heart of these devices. So three new processors were launched, which all work together to give the ultimate performance at the highest end. Our Arm Cortex-X2 is our sort of flagship CPU, it really drives the highest performance in your, your product when you’re playing games or launching your applications. We then have our Cortex A-710, which is our first big core for Armv9. And it’s all coupled with a new small core, our first small core in four years, which is driving the efficiency performance of keeping your battery life going and, and taking on the ongoing tasks for a long period of time. And the underlying interconnect for those cores in the cluster, the DSU-110, allows that to be scaled down to something like four small cores in a wearable all the way up to, you know, eight, big Cortex-X2 cores that can be used in a laptop or desktop computer. So really scalable CPUs in that launch today. And it’s coupled with a really strong line-up of GPUs, so graphics processing units, that are driving our performance, direct trajectory as well in the gaming segment. So at the top end of that the Arm Mali-G710, is the highest performing GPU we’ve ever designed. That gives you really the highest performance, you know, a 20%, performance uplift from previous generation. So it’s going to make the latest games and you know, high frequency displays really sing and stand out. And what we’ve also done at the low end with Arm Mali-G310, is bring the features of our latest architecture, to lower cost devices for the first time. And this is going to be a real step change in raising the overall visual user experience on devices. So it’s a six times improvement in some of the core sort of display rendering and texturing capabilities of those lower end devices. And then we have connecting all of these pieces it of IP together are calling IP, which is the system IP that’s tailored to bring it together to drive the ultimate system level performance. And it lowers the latency of the communications to memory and between the systems and allows that scalability as well across the different markets. So it drives a real vision for compute, that allows us to give the best user experience across the broadest range of applications for the first time.

Geof: So Paul, that that’s a tremendous amount of new technology, maybe you could give us an example of how that might pull together in a use case, where somebody would be enjoying the fruits of this this fantastic work.

Paul: Yeah, so maybe maybe to give you an idea of what next year’s smartphone would look like at the high end. So it’s likely to have our latest Arm Cortex-X2 as a large processor, coupled with some Cortex-A710s and some Cortex-A510s. As well as a Mali-G710, the leading performance GPU and bring that together, that’s going to give you faster launch times for your apps, it’s going to allow you to do you know, higher frame rate gaming. And it’s going to allow you to do things like with its improved machine learning performance across the system, it’s going to allow you to do things like these, you know, continuous video editing, or filters applied in real time to content as you’re recording it. So really upping the performance of your device that’s in your hands.

Geof: And it kind of brings me to my next question about where you see consumer compute heading in the next few years. You mentioned that like immersive and invisible consumer technologies of the future previously, maybe you can expand on that.

Paul: I think I’ve sort of talked about this in the past as immersive and invisible. And this is around the sort of increasing way in which it becomes seamless into our lives. So you know, the ability to deliver a service across any of the platforms that you use in your life is something that we’re going to see increasingly so. You take something like the ability to watch your preferred, you know, streaming video channel or listen to your music, you want to be able to just take that service from, you know, laptop, to smartphone to tablet to car to in home display or sort of smart speaker and not have to think about that and know that that service will be uniformly available across them. And that is where the technology becomes invisible. But it also is allowing you to become more immersed in your your personal experiences. And I think the the breakthrough in this, as it emerges will be augmented reality. So taking wearables to the stage where you get, you know, context information around you, that is allowing you to sort of experience the world around you augmented by those different services, but in a way that is, you know, totally seamless to you. So, if we can bring all of these things together, the underlying platform technology we’ve launched, should help you experience those services everywhere. So to make augmented reality, devices tangible, and something we can use, there’s a huge number of challenges. And we clearly want to bring with us all of those experiences we have today in our digital lives. And this is where total compute solutions really come in. They’re allowing us to take that developer ecosystem and bring it to a new device and a new world, and bring together all of the learnings we’ve created in security and software and tools. That’s going to just allow us to open our partners imagination, create incredible new devices and completely new experiences. And that’s what’s really exciting about the future of client compute.

Geof: Thanks so much for that, Paul. It sounds like we all have incredible and immersive future ahead of us. Thanks also to everyone listening today, and we hope you enjoyed the discussion. We’ll be back again soon with a new episode of arm viewpoints.