 So we're here at the Embedded World 2015 and who are you? Well, my name is Tomas Agay, I am the product manager of the Cortex-M family of processors. And who are you? My name is Dia Soubra, I'm the product manager for the Cortex-M3. So the Embedded World and the Cortex-M, what is the Cortex-M for the Embedded World? Well, today is becoming the standard core for all our partners, MCUs or SOCs. If you look today, both of the stands have at least one or more Cortex-M based products on the show. Maybe a new one here, we have just breached the 10 billion parts being shipped. How many billion? 10 billion parts being shipped. Cortex-M based products, yeah, shipped by our partners. So it's a good achievement knowing we just joined the MCU play about even 10 years ago. And it's just going faster and faster and more and more? Yes, it's actually going faster and faster because now for Embedded it's all about software. So either you have an army of people writing software on their old MCUs or you just do now sophisticated software on top of a 32-bit microcontroller like the Cortex-M family allows you to create. So it's basically all about software. And so around here basically every booth has ARM stuff, pretty much? Yeah, pretty much. And it's a great standard for the Embedded World, right? Yeah, it's all about making the development easier and faster. The differentiation on the core didn't show much value in the past and our partners, especially their customers, can develop much faster, can innovate much faster using the wide ecosystem we have around Cortex-M in terms of tools, in terms of libraries, all the shared solutions. And the prices are great, right? I mean, what I'm hearing, I'm not sure, but some of them might be saying ah, we wish we would have gone to ARM quicker earlier, some of them, right? Yeah, I think if you look at the partners who have been very early to join the Cortex-M party I think they are the ones today who have the biggest portfolio, who have the biggest market share and indeed I think they have a good traction behind it. So how is the Cortex-M compared to some of the other microcontroller stuff that maybe has been there before or that's still there? How can you compare them? Yes, I think from the start when we designed the Cortex-M the very first one was called M3 almost 10 years ago. They were key drivers, ease of use, one of them very important. How can I easily debug my code, how can I easily debug my code? Low power and still very high level of performance. I mean, 10 years ago it was not obvious that people would be using or most people would be using 32-bit processors and now it's coming right here. So how about 64-bit embedded non-tracking? No, we've talked to many partners in this space. 64-bit for embedded seems to be far, far in the future. Right now everything is 32-bit. Remember there's a large volume of embedded systems all shipping with 8-bit microcontrollers. So first you have to take these hundreds of thousands of designs and move them up to 32-bit and who knows maybe in the future we'll talk again about that. And that's a big jump already. I'm seeing some people showing really interesting app stores for Cortex-M stuff and there's a new stuff coming and some new like giant amounts of internet things coming out soon. Yes, definitely. I think as mentioned by Dia, it's all about software. I mean, we are leading the foundation of ease of use, the engine, we can run software and we'll see more and more people bringing more and more value around software. So you see more and more complex tasks. You have seen maybe the Embed OS for example which is being announced by ARM. There are other things we are creating more value just beyond the pure silicon, beyond the pure peripherals. It's here how can I create an application in the stuff of a finger. And it's going to be secure? There's some people coming with new and the thread? For IoT, there won't be any IoT without security. So definitely when the key aspect of IoT will be security and low power. It should be the two big pilots. And one of the cool highlights I've been looking at around here is a couple of companies showing Cortex-M7. Yes. So this is a high-performance stuff, right? Yes, so here we have a happy opinion. We've just released the core, announced the core in fact a few months ago and now we have already three partners going public with their products. So we have the product from ST Micro, the product from AdMail, and also FreeSkill announcing they have a KV5 family. So a very fast traction, very fast deployment and giving access to the parts of the customers to the latest ARM core. And what's the idea behind the M7? Is it just more performance? Yeah, because basically you have seen a huge success around Cortex-M. And people love the architecture, people love the low power, people love the low latency, the fast real time. And they wanted to give the next level of performance. Some people were already using all the performance around the M4 and wanted to get more without having to jump to Cortex-R or Cortex-A processor. They wanted to see the same family and this is exactly what we are offering with Cortex-M7 is a boost of performance for the next generation of applications. And I've also been seeing some really cool Cortex-M0 Plus stuff and people saying that they have 10-year battery life. It's pretty cool, no? Yeah, I actually keep in mind the M0 and M0 Plus are the lowest possible power, 32-bit processor on the market today. So this thing was designed from the start to be the lowest, highest energy efficient processor you can get in a 32-bit envelope. So you can still do the same sort of software. And this is where, as Tamar was saying, so you can take that same software now from the M0 all the way up to the M7 and you're still in that same environment. You don't have to go switch architectures or have to go switch processors or anything. You're in that same family and you're taking your software along with you. Some apps just work on everything. All the apps are forward compatible, which means if you take a binary application from the Cortex-M0 it will also run on the M7 because what you're doing is you're getting more sophisticated instructions but they all share the same memory definition. So it was all done on purpose by design to allow again people to use and reuse their software which is now the key element that people care about. So how do you do this design to make it the lowest power 32-bit? What went into that? These are our engineers. These are our genius. I think first of all is to understand the system because the core is one thing but your core has to interact very well with the system and I think also where I would like to highlight here on the M0 plus for example what we see is a lot of innovation from our partners how to make the complete system low power. It's not about the processor, it's about the peripherals, how they interact, how the peripherals can work while the processor is being in lower power modes. It's not about some partners going forward in the process of technology who have companies like Ambic for example releasing the devices using special technologies so it's all this differentiation around the core to attack and to solve the low power problem from different angles. So yes there's a lot of know-how from ARM since more than 20 years about making a very low power processor design to avoid a given gate to switch and to use power but it's also working closely with our partners to tackle the low power challenge altogether. And the new nanometers smaller and smaller also helps, right? It does, yes and no. I think if you look at the parts which are the lowest power today they are not always at the forefront of the process so the reason behind is with the smaller processes your leakage will go higher. So follow power and as most of the applications will be active only a few percent of the time what's almost more important is the leakages when the processor is in slip mode for example and today I think the switch parts are still around maybe 90 nanometers or 65 nanometers and this is where you will find most of the lowest power M0 plus being produced. Is the advantage also that all the Cortex-A software is kind of compatible with the M software? Some of it somehow? So you don't really have a di-compatibility but in fact there are a lot of parallelism. Cortex-M is a subset of the initial set of the Cortex-A a subset in terms of the functionalities. The good thing is we are enjoying the fact that we have so many partners, some mid-tool partners around Cortex-A which also adopted very quickly Cortex-M so here it's the same environment you can target either Cortex-A or Cortex-M but it's not directly linked to the pure binary compatibility. I also saw some cool stuff here at the show some companies are combining A and M like I saw the free-scale boost it's pretty cool right? I think it's building both of the best world together on the Cortex-A you can run very rich OS like Android, for example, or Linux you can give a very rich customer experience with very nice human interfaces while the Cortex-M is much more optimized for real-time so hard real-time can be very fast interacting for communication protocol so you will see them combining both of them to get the best of both worlds maybe more and more of that oh yeah, definitely and the Cortex-M could be also still active with the Cortex-A, it's put in a power-down mode to save power at the SOC level and what's the Cortex-R? oh that's another segment? yes, yes, so you have the Cortex-R it's around here, it's fair also? yes, it is so we have the three families the Cortex-A for application processors the Cortex-R for, let's say, high performance real-time and the Cortex-M meant for embedded or let's say macro controllers the Cortex-R, for example, is heavily used in applications like storage like automotive powertrain or for example, like telecommunication infrastructure like base stations this is a processor able to run at very high frequencies so much higher than Cortex-M and this processor also better fitted for real-time than the Cortex-A alright, so there's also some of that being used in combination with the A or with the M if you can mix what they want currently you will find most of them indeed providing a combination of Cortex-R and Cortex-M in storage applications for example in SSD so maybe this show is not the best fair for solid state storage devices but you find them in automotive, for example so you've been very busy around here this fair and this is really cool and lots of meetings and lots of things are happening definitely, we're just running for one meeting the next one and I think we're almost there for the next one