 Welcome back. I will do a presentation on the H7, so how the H7 is positioned among the different family, the different series within the H7, what type of application we can do. So STM32 H7, as you have understood, it's all about the performance. And on the next slide, you will see the level of performance actually we are reaching with the H7. H7, thanks to the performance and its architecture, we have single or dual core architecture. It addresses really demanding application on the industrial segment or for security or even for AI. And you will see some example of AI application running on the H7. H7 is not only a core with high performance, but it also state-of-the-art peripherals, for instance, for graphic with to the 2.5d accelerators, complex DMA for fast data transfer and advanced peripherals. One thing which is more and more a concern that we wanted to tackle with the H7 is security. So there are a lot of features within the H7 family in terms of security like hardware, crypto, co-processor. But not only, we have implemented on the H7 some new security services such as SFI. SFI stands for Secure Firmware Install. So this is a native service that you have within the H7. SFI allows you actually, if you have an EMS that you don't completely trust and this EMS is doing your device and doing the programming. So the SFI allows you actually to be able to make securely the programming of the product in a secure way. So in other words, the EMS will not be able to access to the firmware. On top of that, SFI will be able to control the number of units that your EMS is supposed to do. So here with the SFI, which is again a native service that you have on the H7, here you have a way to control completely the programming of the device. Also related to the ecosystem, so you have seen this morning, so you have used quite extensively the SCM32 cube with a cube monitor. So you had the hands-on on it, the cube programmer and so on. But the ecosystem is not limited to simply the SCM32 cube, but also with a whole set of boards and partners and community for the support. So let's see now more in detail what the performance you can achieve with the H7. So here on this slide, you see actually the performance. The benchmark that we are using is a well-recognized benchmark and it's called CoreMark. So this is a kind of official benchmark. You can go through it on the website. So this is a consortium which is called, you can access to it through www.coremark.org. And here you will have the result of all the different architecture that you can find on the market. Not only limited to microcontroller, but also microprocessor. And definitely the H7 is the highest level of performance when it comes to microcontroller with embedded flash. High performance device, we actually started with high performance device with the F2 back in 2010. So 10 years ago, the level of CoreMark we had at that time was 298. 10 years after, we had more than 3,220 CoreMark. So you see the progression. So in less than 10 years, we increase of the ratio of 10, the level of performance of this kind of devices. And today with the H7, the level of performance you have today makes that in terms of limitation is not really the limitation is not really found on the product itself, but more on what you want to do actually with it. So here you have the level of performance we have with the different series of H7. The first H7 and this was back in 2016. So four years ago, we introduced the first H7 and it was the first H7 microcontroller actually using a 40 nanometer process. So at that time we came with a single core configuration Cortex-M7 running at 480 megahertz and this was bringing 2,400 CoreMark. We came after one year and a half after with the configuration of a double core. So here we are talking about the Cortex-M7 still running at 480 megahertz. But in addition with the Cortex-M7, you have Cortex-M4, which is running at 240 megahertz. And here we reached a level of CoreMark, which is over 3,200 CoreMark. And recently we came with the H7 2X. This is what you have in your hand. And here we had a single core configuration, but we increased the frequency up to 550 megahertz. On the lowest part, you have an H7 series, which is having a maximum clock of 280 megahertz. This is a special configuration. With this series, we tried actually to make a compromise between performance and low power. You know, with this type of micro, they are typically having a huge amount of RAM, a lot of peripherals. So what's happened is that in terms of consumption in static mode, specifically when we are in a stop mode, the stop mode is where the micro is not running anymore. You stop the clock, but everything is active. So meaning that the RAM peripherals can be active. So this is what we call the stop mode. Here consumption of a typical H7 could be over 100 microamps. With this specific version here, we find a way by reducing the maximum speed that you can reach. We have been able to reduce significantly the stop mode consumption below 30 microamps. In addition to the stop mode, you have the standby mode where here we reach a couple of microamps for any H7 series. But here, we simply cut the supply on the RAM and the peripherals. So here, when you restart from this mode, you have actually like a reset. You are losing the context, so you have to initialize everything. Okay, so this is for the performance. In terms of product line, as I said initially, we have two configurations, single core configuration. This is what you have today with the H725. But you have also a dual core configuration where you have a Cortex M7 and the Cortex M4. With this double core configuration, of course, you increase the performance. But the purpose here is not only to simply to increase the level of performance, but to allow you to make a system partitioning which makes sense to your application. And you will see just after a couple of examples of how you can take benefit of this double core architecture. Let's see some typical applications which can be done in a single core configuration. So I'm here on the slide 7. On the left side for factory automation, here a typical configuration where the H7 is handling the whole HMI. And in conjunction to that, thanks to the performance of the core, it can also handle the process control and the power management. Another example, connectivity security. Here, an example of alarm panels. So the H7 here in this case is handling again the whole HMI, but also handling the connectivity through wireless modules. So here, this is the different block diagram that could be used for quite a rich HMI. So you see here on the top side of the block diagram, the graphic part with the Chrome ART accelerators and the GPEG codec. So thanks to these accelerators, actually most of the graphic function can be done without the help of the core. While you are using the display interface, the ARM core can do an additional process here. On the example that we are showing here, so what we do in conjunction with the display, we do also audio decoding and output. So typically, something that you can handle in addition to the display. The display can be done either using internal memory or using external memory or a conjunction of both, of course. Here we saw again the security part. This is in dotted line, meaning that this is an option. Actually, in order to benefit from the security function, you need to consider a dedicated part number. And I will show you later on with the whole portfolio how you can distinguish product with crypto and product without crypto. So now let's see some configuration where here you are taking benefit of the double core configuration. So let's see an example for industrial. So here the M7 could handle the HMI, so the whole the graphic part potentially with audio. And the M4 in this case will be dedicated for communication doing a gateway function for instance. It can handle also motor control application where here you really need real-time capabilities and sensor preprocessing. So basically to be interface with the different type of sensors. Another configuration that we see more and more and this is related to AI, artificial intelligence. In this configuration, and this is something that you could see for instance for home automation, you have AI function handled by the Cortex M7. For instance, here in this AI, the function which is done is typically pattern recognition. So just to discover what are the users of the product by the different user. The Cortex M4 in this case could do gateway function again or real-time interface. Here we show with the block diagram an example of this implementation. So this is a kind of implementation where you want to do on one side with the Cortex M7 AI. And here typically voice recognition still handling the graphic part. And again, this graphic part is widely helped with the accelerators. So the M7 really here would almost be dedicated for AI and for voice recognition. On the right side, here you have the M4 domain where the M4 could be doing independently from the F7 different processes like motor control where real-time is highly requested here. Sensing and communication. H7 architecture is again all about performance, performance at the core level, but not only at the core level, but also on the peripheral side. You have seen this morning, we talk about the DMA, you have complex DMA, this type of DMA, a low interaction between different peripherals, between the memories and between the different core architecture when you have a double core configuration. Graphic, again, here thanks to the help of the Cromart accelerators or the usage of the MPEG codec, you can release the workload of the CPU up to 90%. And I think that you're on your discovery kit, you have some example of a graphic function that we do. And I think that you have the possibility to make it run with or without the hardware acceleration. And then you see here in this case how much bandwidth is used in these two configurations by the microcontrollers. Security, we talk about it. Again, a lot of crypto accelerators, hardware accelerators, which allow to release completely the CPU from encryption or decryption. And on top of that, it reduces the global consumption. And high and powerful peripherals. An example of a very interesting peripheral that we have is called a high-resolution timer. This high-resolution timer allow you to have a resolution on the PWM at the level of a few nanoseconds. But it's not only just doing this, it's able also to generate very, very complex waveforms controlled by other external event or internal event. How the H7 family is positioned among the whole portfolio of SM32 families. On the left side, you see the different categories of products we have. On the top of the line, we have MPUs. Here we are talking about pure microprocessor running using advanced operating systems such as Linux or Android. This family is called MP1. It's offered a double cortex A7 running at 800 megahertz and a cortex M4 for real-time operation using a real-time operating system and running at 200 megahertz. High performance product. So this is where we have the H7. And the H7 is our latest series. We started with the F2, then the F4. The F7 was the first cortex M7 product on the market. And now with the H7 in a 14 nanometer technology. Main things, MCU. This is where you find our lowest cost microcontroller. So you have F0 and G0 microcontrollers starting at 25 cents actually. So you have a 32-bit architecture, actually below 30 cents. And you have here F3 and G4 products, which are products where we have embedded a high level of analog peripherals such as D2A, OPAMP, LOV, ADC up to 5 ADC. And this type of architecture are well suited for motor control, highly complex motor control applications. Ultra power with L0, L1, L5, L4, L4 plus. So a whole range of families here starting from cortex M0 plus up to cortex M33 with trust zone on the L5. Wireless MCU. SCM32 WB cover actually protocol running at 2.4 GHz, so BLE 8.2.15.4. So this is what is used for ZigBee or Thread. And then WL, which has a radio covering the frequency below 1 GHz. And this family, by the way, the first system on chip device capable to support the LoRa modulation. So let's see more in detail the H7 family. So you have three different configurations, dual core line. Here we are talking about cortex M4 in conjunction with the cortex M4. And this family is called the H745. You have here you see the H755. So you see that whole here all the different line that we have on the H7 are always offered with or without crypto. So if we are considering the H747, this is without the crypto, you have it with the crypto with the H757. Same with the product that you have in your hand. So H723 is available with the crypto moving to the H733. H725 is with the crypto when you are moving to the H735. So the double line configuration, we have seen it, the single core configuration. Here with the H72X, the one that you have here in your hand. We started initially again four years back with the H743, H742, which is a low end version of the H743. And beginning of this year, we introduced the H7A. This is the configuration where the maximum frequency is 280. The purpose of this was ready to provide a stop mode below 30 microamps. Value line configuration. Value line means a low cost for us. So this is really a product line where the price is the lowest. We have been able to achieve that by a drastic reduction of the flash. So here the maximum flash that you have with inside the product is 128K. So that means that in this configuration, you will need actually to use external memory to run most of the application. This value line configuration is what we call a kind of pseudo flashless configuration. Actually, you have a couple of advantages of having a small portion of the flash. So in this configuration, which is a flashless configuration, your code will be outside. And it's very important actually for the application to make sure that the external flash is really the one that you really want for security reasons. So with this internal flash, you will typically have within this flash what we call a secure boot, which will allow you to do the authentication of the external flash, but also to make sure that the code which is inside the external flash is not corrupted. So this is important for security reasons, but it's important also for safety reasons. You can also have in this flash some portion of your code that you think it's very sensitive, either in terms of security or safety that you could have within the device. And finally, you can also embed into this flash the secure firmware upgrade portion in order to re-update the external flash. So this is again a different view of the different line we have on the H7. This is what you will have actually if you log on the www.st.com. So again, here on the top, the H7 with two lines, the main difference is related to the graphic. On the H747, H757, you have a MIPI DSI interface. So a serial interface to control your display. MIPI DSI is actually an interface which gets more and more on the market simply because it's an interface that has been developed for cellular phone. The display that you have on your discovery kit is actually a cellular phone display and the interface which is used there is MIPI DSI. So the 745 is without DSI. Again, the 55 here is with the crypto. If you move to the single core line, we have the H7A. Sorry, I forgot on the dual core configuration, here you can go up to two megabyte. This two megabyte flash is actually based on a dual bank. So you have two banks of one megabyte. So that means that you can program one bank and you can still execute with your second bank in real time. One megabyte of RAM, the H7A3 goes up to 1.4 megabyte of RAM. And this is quite interesting, specifically when you consider graphic application. H7A3, our first product line up to two megabyte single core, one megabyte RAM. The H725, what you have in your hand. And then on the low end side, the value line product. So now let's move to the HMI which are done on the H7. And typically here the type of HMI that you can do, the graphic display that you can do, is really almost at the level of what you can have with the cellular form. So here we are talking about high end display. So I will not launch the different video, but you have the presentation. You will be able to do it on your own. So this type of application, the resolution that you will be able to handle using the different type of H7, different series of H7. So the size of the screen that you will be able to handle will basically depend on the RAM size, which is available on the product. The H7A series are the product with the biggest RAM embedded into the product. So we can go up to 1.4 and we can sustain up to 5-inch display using double frame buffer and 24-bit per pixel in terms of color resolution. If you reduce the color resolution, of course you can increase the type of display you can handle and you can go up to 7-inch using the H7A configuration, meaning again 1.4 megabyte of flash. If you use external RAM, here in this case any product of the H7 can easily go up to 7-inch display. Going over this with a color depth of 24-bit per pixel is probably not in the scope of the H7. Here you will definitely have to move more to a microprocessor architecture. So to achieve a high level of graphic, what is important is to have the right tools. TouchGFX was actually a partner we used to promote with the H7, another product line by the way. TouchGFX now is a part of the ST offer. We bought TouchGFX more than a year ago and now the TouchGFX has been completely integrated into STM32 cube ecosystem. So TouchGFX offers you the fastest way and easiest way actually to develop a pretty advanced user interface. This development tool is like the whole STM32 cube ecosystem free of charge. One of the benefits of TouchGFX is that it shows the complete capabilities of the micro. And thanks to this it allows maximum performance with a minimum footprint. So the usage of the RAM and the flash is really reduced to the minimum and still having a high level of performances. Let's move to something which is now more and more considered, which is AI, Artificial Intelligence. Here's some examples of what has been done on the H7. People detection, food classification. And here we have any demo of all this type of implementation. You can have them, you can see them on our website going to sm32.ai.com. So food classification, voice recognition, industrial application. Here we mean preventive maintenance, doing vibration analysis for instance. So all this kind of implementation are examples that we provide. So how we implement artificial intelligence on STM32. By the way we can do that not only on H7 but also on our other product line. So here there are two parts that you need to consider. The first part which is the generation, neuronal network and the training of this neuronal network. All this part is not done on the STM32. This need to be done over the cloud using standard development tools. One of the most famous is TensorFlow. But you have some others like Keras, Linux, PyTorch, Caffe and others. So using these tools you will be able to generate a model. Then with the model you will train it with data you would acquire. Based on this when you have this pre-trained model you will use STM32 cube AI which is an element of the STM32 cube. Cube AI will allow you to generate a complete optimized C file implementing your neuronal network on the STM32 in a best and efficient configuration. And then you will be able to run the inference directly on the STM32. So this is the way where we can easily implement AI on STM32. Another things which is very important in the industrial world is a functional safety package. You have seen probably we have a package which allows you to do class B type of product. Here we are talking about a level which is much higher here. We are talking about seal 2, seal 3 functional safety packages. With this library you can get them at 1DA. They are not directly available like this. So this is actually a huge amount of knowledge here and this is really a valuable package. Here with this type of library we take actually benefit of some of the features the hardware features that we have within the product. This library by the way are not only available on the H7 but most of our product family starting even with the G0 which is our entry line product. So here we see on this slide some example of the hardware features that this library are using in order to reach the seal 2, seal 3 certification. Dual watchdog, an independent watchdog and the system window watchdog. So this is very important having a dual watchdog which monitor the behavior of the application. The CSS in case you have the failure on the clock, you switch automatically to the internal clock in order to secure the behavior of the product. CRC, the monitoring of the supply, audio function locking, memory protection unit to execute safely different processes. Flash protection, error code correction, these are new features which are specific to the H7 which we are not present on the F7 and F4 allow you to monitor the behavior of either the RAM or the flash in real time. So these were some examples of things that you will be able to use and implement easily on the H7. So here on this slide we are back to the ecosystem. So on the left side you see the cube and the different components of the cube. So again here the benefit of cube not only because it's free of charge but also the whole benefit of the SCM32 cube is that within one tools you have everything that you need in order to develop your application and this is really a huge benefit of the SCM32 cube. Things about middleware, so you have different packages about the middleware but here you see that there is the possibility to do what we call expansion packages. So typical expansion packages that we provide, that SCM32 provides are packages which allow you to drive some external components like sensors, like RF or whatever. But be aware that you can do your own expansion package that you will be able to add to the SCM32 cube. So this is something, so we have opened the cube environment with the possibility for anybody to do an expansion package which could be linked to the SCM32 cube. Hardware, there is a slide on it, I will just after, I will go back on that. Quick slide here, so you know cube. The thing that I wanted to stress is that within cube you have the cube IDE which is based on Eclipse and GNU CCC. But if you want you can still use another provider such as Kailh from ARM or IAR which can be linked directly to SCM32. The hardware offer, so different type of hardware. Our top class hardware is what we call evaluation board, they are pretty expensive, over $300. This is really needed in case you want to do a specific evaluation. In most of the case you don't really need this type of board. Discovery kit really makes a lot of sense, this is what you have in your hand. Depending on the configuration it's $97 down to $69. So you have a screen, you have a lot of additional components like a sensor, like a microphone. So it's provided with the whole set of applications. So this is really something that you can use as a reference design from a hardware and software point of view. And then on the lower side you have a nuclear board. So these are basic boards. But the benefit of this board is that here you have a Arduino connector. So you can add shields, adding new functions like your sensor or RF. And you can build like this a pretty quickly proof of concept. All the different versions of evaluation board, discovery kit and nuclear board for the different series. So this is for your reference. And I'm finishing with my last slide. So you have seen many things that we are able to offer but know that you can also rely on some partners. And here we have a dedicated program that we call partner program. Here with this partner program we are selecting our partner to make sure that they have their solution is fully optimized for the SCM32 platform. And you see here on the slide some examples of this partner. So here again the message is you can rely in addition to what ST is offering, you can rely for the dedicated application to a large, large set of partners having an expertise on SCM32. Here this is the last slide. So again we show all the resources available on SCM32 and specifically on the H7. There is a lot here. So it needs some time to get the right information but definitely there is a lot there. So this is the end of my presentation here.