 I will present to you the problematic of high performance applications. I would like to ask you, each of you this question, what are your main challenges and concern when selecting a microcontroller for your high performance application? These key points are what we have summarized from customer feedback that we have heard. A lot of time when people talk about high performance application, the main challenge to think of is real-time performance. What is the performance? What's the real-time? How fast my code can execute it? The other one is memory constraints, and this is the key point for today's workshop, which memory types can we use and scalability and memory interfaces for external or limitation of the internal. Product integrations, the ecosystem, how to build your software, how to build your product time to market, so the whole ecosystems and the product information, product temperature range, and security because nowadays we see more and more people care about security. How can I secure my code? Because nowadays people work with maybe third-party or design house. Now it become more and more important how to address the security to prevent middleware or prevent more tech. For some customers they care about scalability and if large package offer is available, I need maybe you need 100 pins in your products, you need a number of IOs, but somehow it's in the market only small package available. For some people they care about large package offer. Longivity, yes. We committed 10 years longevity and renew it every year, so it's basically endless to be honest. So longevity and supply chains, stability, this we will address also. People concerned about using select in a micro. Development support and solution costs and family consistency. For some people if they want to select a micro, they think maybe I can reuse my code, this experience for my next project. So people care about family consistencies, maybe they can easily migrate between families, within the families but different packages and different setup. We conclude these 12 challenges. You may consider then how to address these challenges. So let's talk about family consistency first. In our extensive STM32H7 product portfolio, in these sessions I want to focus the new product here STM32H7II and H722333, H72535 and H730. This is the new product H7 series. It's with the Cortex M7 core running up to 550 megahertz. You may wonder what is the difference between 23 and 33. 33 is with the hardware crypto and 23 is without hardware crypto. So if you need the hardware crypto AES, you may need to select the 33. The same rule apply here. So H725 and 3535 is with hardware crypto and 25 is without hardware crypto. You may wonder what is also the difference between 23 and 25, one is with LDO and the other one is with S&P as option. And the flash is up to one megabyte for what we call so-called SS line here 23253335. We have this so-called value line. Value line is always cost optimized. What does it mean? It means basically cheaper. And how can we achieve it by limiting the flash size up to 128 kilobyte? So in a market, there are some other microcontrollers for high performance. They call it flashless microcontroller. So we use the similar concept here in the value line. So if you see H7 in the end is zero, it's always the value line. So STM32, when we name the microcontroller, we have the certain system. So if in the end it's zero, it's value line. And you can see that the flash is limited up to 128 kilobyte. And the idea is you use external flash, external RAM for your applications. So the next topic, we would like to talk about memory constraints and scalability before we jump into more details. I would like to ask you normally, how big is your code size? You see that a lot of time, let's say 90%, your code is between 50K to like within one megabyte. In these slides, we would like to show you the memory architecture for your choices with STM32, H7, 2, and 3. So you may, for people who just answer maybe A to D, you can use internal memory only. So you can optimize everything or in one. So H7, 2, 3 is supporting up to one megabyte internal flash and 564 kilobyte internal RAM. For some people, maybe you are not so time critical. Maybe you don't really care about the latency here, but you need to optimize the cost. In this case, you can use the value line and external memory only. So value line, we supported up to 128 kilobyte internal flash. So for people who answer A, maybe this could fit small code, maybe quite sensitive or people answer BCD, you can use external memory as well, put your code in the external flash, but with octal SPI interfaces, so you can have a fast code execution and we support on-the-fly decryption. So this can dramatically optimize your code execution from external memory as well. And thanks to the cache, of course, and on-the-fly decryption. For some people, maybe you want to have the mixed memory architecture. You said, yeah, I have some critical one, it's real-time, I care about performance, I can't lose anything. So you can use the SS line H7223332535 with the internal flash for your code. If you have some data, maybe you need to store, maybe just some real-time sensor data, and that's tremendous. Maybe you can put that non-critical part also in an external flash, depending on what is your requirements here. So thanks to, again, octal SPI interfaces, we have two octal SPI and on-the-fly decryptions, so you can also optimize the combination of external flash with internal flash. And the last one, external memory only with internal RAM. So this is with the value line. So for some people, maybe you have a critical code you want to protect, you can use a secure boot within this 128 kilobyte internal flash. And with the octal SPI interfaces, you can have, and thanks to the on-the-fly decryptions, you can have also your code executed from external flash. So basically, we provide wide range of memory architecture flexibility here based on your choice. So it's up to you basically. This is the memory types that H7-2 and 3 support it. A lot of time, people would like to use PS-RAM and Hyper-RAM because of the low cost and AP data width. So you can connect it through the two octal SPI interfaces and it's a small microcontroller package. Some people, maybe you don't like PS-RAM. You want to use some flexible memory controllers. You may choose SRAM, name flash. In this case, you can have your flexible controllers and different parallel memory configuration. And we support up to two times 256 megabyte. Nor flash here. So you can have your code execution in the nor flash connecting using the two octal SPI interfaces or quad SPI. So basically it's supporting octal SPI, supporting quad SPI, of course, and on-the-fly decryptions. So it provides a flexible architecture if you would like to use the nor flash. The last one, EMMC. So for some use cases, you may need some SD card interfaces or SD EMMC interfaces. So we support two SD EMMC interface and SD card interfaces. We provide a wide range of memory types for external memories. So depending on your requirements, these are all you can choose. We would like to show you some examples of low cost external memory setup with H7, 2, and 3. So you can have, for example, connected the nor flash and cost-optimizing low-cost PS RAM here for your frame buffer, for your graphics. But in the nor flash here, you can put your code as a Q-dit through the octal SPI and the other octal SPI with a PS RAM. So this is one of the mixed memory architecture here for the low-cost external memory. If you need bigger flash, more RAM for your graphics. So two octal SPI can realize it this setup. Or you can use the external flash only and execute a code from here. So this is highlighting H730, the value line. So provide up to 128 kilobyte flash. So you can have your boot loader, your secure boot here. Secure boot, secure. And then you can execute it, your code from nor flash. So you can do the authentications with a secure boot. So without people hacking it and adding a PS RAM here for the low-cost and for your frame buffer, for display, for your graphics requirements. So we have tackled the memory constraints and scalability. Let's talk about performance because this is today's main topic. H7203, this slide shows the CoreMark result. So the CoreMark is a well-recognized benchmark in the world. So we use this CoreMark as a reference. You can check their website, triple w.coremark.org. If I remember correctly, you can see all the microcontroller performance result there. Not just the microcontroller, but also the microprocessor. What I'm trying to show you is H720 running at 550 megahertz with Cortex M7. We're able to achieve 2778 CoreMark, single core with embedded flash. So this is the highest CoreMark. Actually, if you look at the coremark.org, highest CoreMark in single core microcontroller. So thanks to the embedded flash. And today's hands-on, you will see the differences between internal flash and external flash. You'll see some performance difference. But we are able to achieve 2778 CoreMark. The other one I would like to highlight is deterministic behavior for the code execution for internal memory. So basically, thanks to the 64 kilobyte L1 catch for data instructions. So basically, you can have very good fast code execution from internal memory. And thanks to the L1 cache as well. And multiple bus metrics for autonomous operations. So the complex DMA, the scalability. And we don't compromise in this H7, providing this complex DMA controllers. So the D1 domain with ARM Cortex M7 and provide maximum performance at 550 megahertz. And we also provide a high-speed peripherals. And we have this what we so-called D2 domain. And it can offload the CPU. So it has its own RAM memory. So basically, you don't need to go through the core, but you can have a DMA in this D2 domain and offload the CPU activities here. So this end, we have this low-powered D3 domain. To be honest, this is the slower mode. So if you are looking for the high-speed peripherals, please use D1 to be honest with you. But the D3, basically, if you care about power consumptions, your application is basically sometimes in a sleep mode, in a stop mode, you can use this D3 domain, still have some IOs up, and you can still have some data acquisition. Of course, you can offload the CPU as well. So depending on your requirements, we provide the flexible architecture on H7-2 specifically to address this performance. Let's talk about product integration here. So we have a rich integrated peripherals. Basically, you provide a terminus peripherals with DMA access to all of the CPU. So you can have a DMA direct memory access. So between the peripherals and memories, this can offload the CPU activity. And we provide the three busses metrics to maximize the CPU performance as well. And the three FD-CAN for time-triggered functions. So if you need some FD-CAN interfaces, we provide up to three. If you have some camera LCD requirements, you need to connect to the camera. You need to connect to a display, LCD display. We have a dedicated DMA for it. So starting with the smallest package. So for some microcontroller in the market, you may see that they are not so flexible because dedicated DMA may be only in a bigger package. But H7-2 and 3, we have starting with the smallest package with the dedicated DMA for camera and LCD. It has its own Ethernet controllers. If you have some Ethernet requirements here, external memory with on-the-fly decryptions and up to 16-bit ADC with 19 channels each. So if you have some requirements to have some analog interfaces, you need a high ADC. So it's a 12-bit ADC in a low-power domain and a 16-advanced analog peripheral in a two 16-bit ADC. We have also application-oriented features like motor control timers, PWM, state-of-the-art security, like we support, and also USB support it. The other point, let's talk is product temperature range and large package offer because you may wonder about what is the temperature range and which package do you have like 100 pins package, what package is available. So let's talk these two points. This slide shows our H7 package offer, H7-2 and H7-3, starting from 68 pins VFQFPN to 176 pins LQFP or USBGA. And you can also scale your flash memory size and a RAM size starting from 128 kilobyte flash. So this is what we so-called value line here, which is cost-optimized. And you can also select 512K kilobyte flash and or one megabyte flash, depending on your code size, depending on your requirements. You can easily scale down, scale up between the same packages or even between different packages for the code reusability. And also we provide lots of the pin-to-pin compatibility between different flash variants. So maybe you start with the value line first and you notice that it doesn't work. I really need more flash. I don't want to get an external one, it's too much trouble. In this case, you can use pin-to-pin compatibility and then migrate easily to others. So the light blue one here is H7-2,3 and 3,3 lines. So 3,3 is with the hardware crypto and 2,3 is without hardware crypto here. And this dark blue is H7-2,5 and 3,5. And so if you see this light pink box, so it's highlighting with hardware crypto or without hardware crypto. For some people, you may wonder, I need power consumption. I care about it a lot. I need S&P as for my power supply. For this rate one dot here, that has internal S&Ps. And the temperature range we support from minus 40 degree to plus 125 degree Celsius. So for most of the industrial applications, it's minus 40 to 85. For some typical critical applications, you may need to go to plus 125. So we provide a wide range of temperature range here. For solution costs, you may wonder, how come this device can save the cost? So basically we offer the package four layer design. So in the manufacturing of the PCB, this four layer design can dramatically reduce your PCB cost. And the wide range of the input voltage from 1.62 volt to 3.6 volt with or without power management unit or external activity components. So this means you can save also the PMU unit here. We can optimize the current consumptions and product efficiency with DC-DC converter. So you can save external DC-DC switch. So that's why you can save the cost using the H7-2 and 3. It provides state of the art internal peripherals like an ultra low power battery domains and precise RTC. So you can save the crystal using the internal clock. Depending on your requirements. So it can save the cost. H7 is a one stop shop. That means basically it's all in one everything you need. You don't need the two micro solutions. It's all in one to address your cost reductions. And if you have some continuously projects, maybe now you need a single core. The next one, maybe the dual core is needed. You can easily migrate it within H7 family. So basically this can save a lot of your development costs because your code can reuse it a lot of time. Your development time, which is the cost. Let's talk about the security part. H7 supports AES 128 or 256 encryptions and a key provisioning for STM32 authentication. I want to highlight it's part of STM32 trust security for cyber protections. So basically support on a fly decryptions and secure SFI what we so-called secure firmware install. So on the fly decryption OTF DEC, that means on a fly decryption that enables you to encrypt the code and execute it from external memory. The SFI means that it allows you to order the standard products anywhere in the world. And you have an EMS, which you don't trust. Maybe somewhere in Asia, you don't trust it. It's a third party. You don't want people to have access for your code, for your firmware. In this case, you can use SFI secure firmware install with H7 and securely program your devices. Okay, so let's talk last point, the longevity and supply chain. So we committed 10 years longevity. So now it's 2021, so 10 years, 2031. And we renew it every year. What does it mean? It means that in the 2022 next year, we renew it and then committed another 10 years. So we guarantee 10 years longevity and renew it every year. And we have a dual source. So dual sourcing in supply chain. So we always have a dual sourcing for the fab, for the front end and back end to guarantee the stable supply chain here and also the quality, of course. And the product available at ST or the e-store, you can immediately get it from our e-store or from the distributors. We can, it can always vary fast at your hand now. Yeah. The development size, so we will talk more in today's workshop in the afternoon session. The QBemEx and QBid, you will use it a lot in today's hands-on sessions and the Q packages, the nuclear board and discovery kits, it's all available with H7. So the software support, the hardware, you can use the nuclear as a reference design for your product. For the support size, so we have a worldwide FA support offered directly by us, ST. So, and also we have community. So like if someone posts it in a community, something that's on work, you'll get an answer as well. Also ST technical support using your MyST.com account. That's also valid. We also have this MOOC MOOC online training. It's a YouTube channel. So if you want to learn something about security, you want to learn something about that micro, you want to learn something about maybe a Bluetooth stack, how to get it work with STM32WB, it's all available in a MOOC online training. You just need to Google it, ST Space MOOC, you can see it. So we offer massively online trainings and webinars, workshops. So we have continuously workshop online training and we have more than 300 partners in our ST product portfolio. So the conclusion here, we have toggled these 12 challenges in this problematic of high performance applications with STM32H7. So some takeaway here, I want to highlight is H7, STM32H7 is the one stop shop for you for your high performance application because we just answer these 12 challenges in this slide. It deliver the highest performance microcontroller in the market with embedded flash. So it provide the up to one megabyte flash with H7 2 and 3. Even for other family, we provide up to two megabyte of flash, we will show you more in this afternoon. So and with the highest performance. We have a rich memory and IP integration. We keep innovating our new IPs and we always try to allow a very small package as well. So that can dramatically limit, optimize your PCB size if needed. STM32, we are not limiting our customer with memory architecture. So we address for memory architecture in this presentation. So you may wonder, you want to mix memory architecture with internal and with external memory. We're not limiting you. So you can choose your preference here. And ease of use, STM32 cube ecosystem. In today's afternoon sessions, I will introduce you more about cube, what is the cube environment, what we offer in a large ecosystems, how to help you to develop your software tend to market quickly. Okay, so that's all for the problematic high performance applications presentation. Thank you all.