A beginner's guide to the OpenHW Foundation community and commercial ecosystem

This talk focuses on RISC-V, a reduced instruction set computing architecture that is the fifth iteration designed to improve processor efficiency. The speaker discusses the community-driven nature of RISC-V and its significance as the first commercially successful open-source instruction set. He highlights its historical context compared to established instruction sets like ARM and Intel, emphasizing RISC-V's suitability for silicon research without licensing fees. The Open Hardware Foundation plays a vital role in developing verified RISC-V cores, enabling innovation in chip design and production. By offering open-source resources, the foundation seeks to democratize chip development and provide a comprehensive ecosystem that spans hardware to software solutions.

[music] >> I am not Telle Ti Eldridge. Ti cannot be with us today and I've been asked to step step in for her. Of course, I've never seen those slides and I will be making up everything. So, let's discover what's there there there is in this deck. No, not true. I'm Frederic de Beaucourt. I help out with program management at the Open Hardware Foundation among other things.

I am a staffer of the Eclipse Foundation. All right. So, this presentation is about RISC-V. So, when we say RISC-V, what what are we talking about? Well, essentially this is the fifth iteration of what we call the RISC computing architecture. RISC being reduced instruction set computing. So, historically speaking, processors were using in the past very long instructions, very complicated and at some point researchers started to have

this idea that maybe we can design processors with short, very predictable, the same size instructions and improve computing because of that. So, RISC-V is the fifth iteration, so to speak, of idea. And and really, it's the first community developed commercially successful instruction set in history. So, that's quite an accomplishment from a technical perspective. And really, it's it's a huge community. So, there is RISC-V International as the

steward of the RISC-V instruction set. And really, it's a huge community. Eclipse Foundation is a member of that community as well. And of course, we are trying to well, push uh little projects there. So, when we look at the history of RISC-V, you can see that this is still a very young let's say initiative in the sense that you think about established instruction sets. The ARM instruction

set goes back to the '80s. The Intel instruction set has its roots in the Intel 4 4004 in the early '70s. So, they are fairly more mature. So, in this case, first version of the RISC-V ISA was developed as a research project in Berkeley by a professor and a team of students. And then, over time it gained popularity because, of course, the problem when you want to

do research in processors uh dealing with commercial commercial instruction sets mean you know, you have limitations to what you can do, or you need to pay for it, and of course academics have no money. Uh so, essentially RISC-V was a way to circumvent many roadblocks on the way of doing proper silicon research. And then, uh there were slowly, uh you know, commercial entities that started participating. Uh

here in 2017, for example, Western Digital announced their first core based on this. So, people could pick up the instruction set because it was open source or open. Uh so, no licensing fee, and they started building commercial little chips all over the place. But by the way, maybe you don't know that, but your Nvidia graphics card or your iPhone is full of little RISC-V cores. So, uh

in the case of the iPhone, for uh of course, the main instruction set is ARM on that, and Apple is licensing that from ARM. But for many little microcontrollers that are on those chips, like the temperature control and all of that, it's always five because it's simpler, cheaper, faster to iterate that way. So, there are tons, billions of RISC-V cores already in the wild, uh believe me.

Anyway, you can You can see the rest of the history there, but let's say that this is a very healthy uh open source initiative uh when we talk about the RISC-V instruction set. Now, you will ask me, "Why is this important? I don't care about the instruction set on my phone, laptop, microcontroller, or whatever. Well, yes, you care, and here's why. Um, essentially, first there's this freedom

aspect. We are all involved in an open-source community, and the fact that we can now innovate in open-source at the silicon level is tremendously important for the future of the whole industry. Then, of course, when everything is open and in the open, you avoid patent troll problems, right? That's another big consideration, and strictly you you've seen all of those lawsuits between between chipmakers, and all of that,

trying [snorts] to gain a competitive advantage through lawyering rather than, you know, innovating in tech, right? So, in RISC-V, we avoid that because everything is on the same vendor-neutral level with the instruction set. And not only that, but uh, the simple fact that it is open means that it is much more flexible and configurable than any other alternative. So, in other words, you don't have to beg

an ARM product manager, let's say, to do whatever you want, because you can innovate yourself, you can contribute to those specifications, and of course, this means chip development because becomes much more democratized. Of course, producing a chip is a magnitude, and uh, or several after of magnitudes more expensive and longer in terms of effort than building software, okay? Even a small microcontroller would be 10 million euros

of investment over two or three years for a team of 10 people, okay? But still, it's possible now to have very viable early-stage startups working on silicon because RISC-V is there and provides them a ton of things that they can build upon, right? And this is, once again, the spirit of open-source. And finally, there's the sovereignty aspect, right? All of those existing ISAs are American. So, from

this aspect, RISC-V International is based in Switzerland, which means from intellectual property perspective, you avoid certain pitfalls related to geopolitics, okay? And so, RISC-V instruction set is nice, but to me as a software developer, because this is my my my true trade, right? It's a bit useless. I mean, it's nice to read all of those manuals and specs about instruction sets, but it doesn't give me a

platform to work on. And this is why Open Hardware Foundation has been created. And the goal of Open Hardware Foundation is really to make RISC-V as an instruction set and take it a step further by having true processor cores and other hardware designs based on it that you can pick up and build actual processors out of it. Some of our members do just that, right? So, and

and the keyword on all this text blurb here is really professionally verified. I think I think it's all somewhere. >> [laughter] >> You did not read it. Because we have several versions of that of that slide. But essentially, the distinction between many of the open source RISC-V stuff you will find on the internet and what Open Hardware Foundation does is that we have verification engineers on staff,

like my friend Kai over there, who was the star >> So, you know, people who are deeply knowledgeable about chips and will take a simulator on one side, they will take a processor design on the other side, and instruction by instruction will run programs to compare the results and and and ensure that the chip does whatever it it is supposed to do. So, we have professionals like

that, which means that what we offer is not totally production-ready in the sense you you wouldn't, you know, in a matter of weeks get a product to market, but maybe in a matter of months, which is incredibly faster than anything you can do, you know, traditionally in the silicon space. And so, what we have, of course, are, you know, those processor cores and this processor IP, which

is aligned on the risk five ISA. So, we we really collaborate with them, and we are part of this global innovation ecosystem. So, when we look at Open Hardware Foundation, it's called a foundation, but in fact, it's an uh working group of the Eclipse Foundation. So, they are part of our big happy family here um at Eclipse. And and really everything is permissively licensed. So, you can

pick that up, build product, sell products. You don't owe a cent to anyone from an IP perspective. So, that's a game changer. And really, we are the world leaders, the only place where you have this kind of professionally verified open source risk five cores anywhere, okay? And really, we have many, many global partners from industry and academia working on this with us. So, our two strategic members,

both of them from Europe, and specifically from France, are CEA and Thales. And then, we have tons of members uh all over the world uh working on those technologies with us. Some of them are universities because of the academic roots, of course, of the risk five instruction set. Some others are software companies because we need good software to run on those chips. And some of them are

hardware designers and other people involved in silicon. So, it's really a big happy family, and you are well than welcome to join. Of course, we need more friends and more money. Ah. Um but really, what's great about Open Hardware Foundation is that it is this foundational technology that was the missing piece in the puzzle at Eclipse. Eclipse was pretty much software-centric and it still is. But when

you consider the addition of Open Hardware Foundation, it is this foundational layer that now means we can deliver everything in open source from the chip to the real-time operating system, Eclipse ThreadX, other operating systems like Oniro. We have dev tools, we have application building blocks in SDV and IoT and other vertical domains. And then, of course, we have Open VSX as a channel to distribute our uh

developer tools as it were. So, it's really the most comprehensive ecosystem in open source that you can think of in the industry. And you're all part of it, so congratulations. of course, um Open Hardware Foundation is really a place where people who care deeply about chip design come to collaborate, but this is not just about the cores. We need a Linux port, we need compilers, we need

RTOSes and other pieces all around this. And of course, good documentation. Some companies like Circuit Sutra, for example, one of our members, they are improving massively some of the documentation on some of our projects. Uh and and that's fantastic contributions because without a good developer manual, it's hard to write an OS for a new chip, right? So, from this um our community is very diversified. And of

course, uh as I said, uh we need more friends, so feel free to join. Uh when we look at the portfolio release, we have a whole uh very broad family, of course, to offer. Starting with the CV2, that's Kahlil's favorite because he works so much on it, I suppose. So, CV2 is the equivalent of a Cortex-M0, very small, low power for deeply embedded application. Stepping up, we

have CV4, a bit more powerful, but still good for microcontrollers. And then, we have the most powerful one in uh CV-A Wally and CVA 6. CVA 6 is the most widely used, I think, a risk five open source processor in the industry, even if people won't say they are using it. There are already millions and millions of actual chips in real physical products based on this technology.

A Margot Tours by Thales. They won't tell us what it is, but at least they are saying that they are using it. So, good for them. Um, and then there is CV Wally. It's interesting because there is this upcoming textbook in June where CV Wally is the narrative, let's say, used to structure the textbook to teach about chip design. So, the next generation of chip designers and

verification engineers will have trained on this fantastic technology. And then we've got CV A5, which is a version of one of our cores that works on FPGAs, so you can experiment in that environment as well. So, anyway, we are part of broad collaborations. Open Hardware Foundation is involved in research projects at the EU level. Of course, we collaborate with RISC-V International. And now, partly because of me,

I'm the software guy bridging the gap between Open Hardware Foundation and the rest of the software ecosystem at Eclipse. So, we need this because well, silicon are important. You've seen during the pandemic, no cars because no chips. So, we're trying to solve this by having a sovereign chip facility production in Europe, and we need to grow this. Uh, and there there are several European initiatives once again

around this. And I think I will probably be out of time soon. How much time? I'm done. I'm done. I'm done. So, thank you very much. In any case, if you have questions, we have an Open Hardware pod. I'm going back there, so feel free to visit, and I will try to make up other facts about Open Hardware Foundation. Thank you very much. >> [applause]