 Okay, so hi, welcome. My name is Luke Kenneth Casson Layton. I'm the guardian of the EOMA 68 standard It is a certification mark not a trademark. So I have the interest to branch out in as many areas as possible so now About three months ago. I On the mailing list for the EOMA 68 project somebody said Oh, there's this Indian Group doing a risk for their own risk five core You should contact them. So I went. Okay, all right wrote an email and said hello My name is Louie and I'm a free software developer over blah blah blah blah blah I'd be really interested to help you to develop a system on chip Which does risk five and bad guy's gonna yadda yadda 19 to the dozen sort of thing And this guy responded, 30 emails later, we got introduced and established that, yes, they have unlimited funding by the Indian government, and I'm quoting here, to piss all over intel and arm. The Indian government used to be called Paranoid. Now we know they're not Paranoid because the NSA really is spying on your process, an intel process, through the NSA spying coprocess of backdoor intel management engine, including the switch to called NSA off whatever it is, management engine off so that foreign powers do not spy on computers on NSA, as NSA's own premises. So, then we had Spectre and Meltdown and I'm very sure that people are sick to the back teeth of that. So, what I did was I talked with him and designed a pinmux for the proposed processor. Now, you can see that you're a Rumbus Tech, Risk 5 Shakti M class. Can we swap to the pinouts tab? Awesome. Okay. So, this is generated by a Python script, pinouts.py. It generates the documentation and the technical reference manual and includes further down some scenarios which test the viability of the whole pinmux design. Okay. So, if the pins don't map, then I know that I've got it wrong and I can make some changes. But this took me only two days to write. Okay. And adding a new scenario, this one for the Internet of Things one took me one hour. Okay. To explain to you how significant that is, five years ago I tried the same thing here with Keysad by actually designing the pins in Keysad and tried testing it after two months I gave up. Okay. Now, why a pinmux is important is because when you have a processor which has a system on a chip, this is unique to a system on chip because normally you do northward south bench and the south bridge take care of all the IO. But on a system on a chip, you've got to get it right. You're going to invest 30 million US dollars to produce a chip, one mistake, that's it, no customers. Because it does not satisfy the markets that the chip was supposed to be designed for. Okay. So it's very, very important to get it right. So normally you have, if you lined up all of the functions that were available on say, for example, an OMAP, Texas Instruments Processor, you would need 2,000 pins to cover all of the functions lined up back to back. All right. On a straight through. That is simply hopelessly impractical because what happens when you make a 2,000, 1,000 or 2,000 pin chip, the die is put on a circuit board where it hits bonding gold wires and uses fusing to melt the wire onto a thing. Now you've got to hit the chip. If you've got 1,000 pins, you have to hit the circuit board 2,000 times. And eventually it causes cracks. And that, once it's cracked, that's it, you have to throw the hung out. So the yields are very, very low. So the less pins you can get, the less hammering you can need to do and the better, the lower the cost, overall lower the cost of the chip. So pin mugs is incredibly important. So when you design a pin mugs, it has to be targeted at a specific scenario. So I went back and designed some scenario, some schematics, which helped me to test the peripheral schematics. So on this page I added a little section about what power management chips are commonly used for system-on-a-chips that I've seen over the six years of working with PCB designs. So there's a section for those, if you need to do one using effectively discreet components, et cetera, et cetera, different things. And then if your microSD card needs level shifting, this is how you would do it. You would buy these components and blah, blah, blah and other stuff and things. So there's all these different approaches and this allowed me to analyze the benefits of each approach when it comes to doing something as simple and mundane as connecting a microSD card slot to a system-on-a-chip. So sometimes you might not be able to switch the whole of the bank of GPIO down to 1.8 volts because you've got the bank is 16 pins. Only six of those are for microSD, the remaining 10 you need. So you can't do switch the whole bank down to 1.8 volts on demand. You have to do it as a level-shifting chip, external chip. So all these kinds of things, you have to think about them before you even commit and it's because this cost is so insane. $30 million is budget money for producing a chip. So of course now that's all that's changing with risk five. So this guy, he's so funny. Yeah, what the heck, I was saying, Arm, very good for you to try to bribe Madhu with $24 million offer. If you'd offered him $240 million, maybe he would have said yes. All right. This is what Arm is doing. Okay. All right. They think because he's Indian that he will take a bribe. Right. It's completely unethical of them. So Madhu's family members are founding investors in Google. He is the gateway to the entirety of India for a billion market smartphone. Billion smartphones, 100 million laptops in schools, 10 million, 100 million. It doesn't matter. The numbers are so insane. Synopsis and cadence have offered him free access to $80 million worth of tools, ASIC design tools. TSMC, you do not get access to TSMC. Okay. Unless you are someone at Taiwanese company or university or everybody else has to wait. All right. TSMC have offered him access to the multi-vendor wafer testing program at 20 nanometer, 28 nanometer and 40 nanometer. One free run of chips. The only condition that he's placing is that the entire source code of both the software and the hardware must be Libra licensed. And if any of you, any of you who have seen this talk, wish to make yourself an entirely Libra chip, contact me and I can provide you with the work in some way to the opportunity. Now they also have their own fab. I don't know the foundry part of the process. It's 180 nanometer and anybody wants to make their own processor. It doesn't matter what it is. As long as the caveat is it must be fully Libra licensed. I can put you in touch with them. They will make it for you and send you some chips for free. Okay. All right. This does not happen. This is empowerment for everybody who has been despondent about the costs of doing free silicon for Libra licensed silicon. Okay. It's just absolutely fantastic. So the only thing is it's a bit like it's a bit like riding a bucking Bronco because the one of the problems with the risk foundation is you need to say if you want to join the risk foundation, you need to sign a an agreement which has clear cognitive dissonance in its statements. Section 1.9 conflicts with section 5.2. Section 1.9 says you are required to push information out to free software people but you can only do it under NDA with our permission under section 5.2 because of confidentiality. So no free software developer will ever join the risk 5 foundation. No sensible one unless they want to become totally tainted. So there is this clique already a freaking clique in the risk foundation where only the contributing founding members are making the decisions and then doing one way push out to people like us in free software. All right. I hope you're watching. I hope you fix this please. Okay. So just talking to the camera. So the problem is that because Madhu's group is hardware focused then they're very very busy. They're not necessarily focused on the day to day main list decisions informal decisions being made on the risk 5 list. So the risk is that they just go you know what we got a billion population. We can fork the entire risk 5 hardware and software code base. Don't care. Consequences are then that cheap hardware commodity hardware starts coming out of India and starts Andrew and Charles manufactured in China and starts reaching the rest of the world. And they go I got the Bedebian or the Fedora binaries. Why don't they work on this processor? It's like well because the yeah you get the idea. So this is a very serious thing. I am sort of landed in the unfortunate position of being the messenger here to ask people in risk 5 to collaborate and keep together and make decisions in a unanimous decision making fashion because the consequences of doing of not doing so are too severe. All right. If the thing you've already seen what happened with arm arm 5 years ago I heard they had 720 licenses of different companies. All right. There's no way that you can bring that together. So you have Raspbian where it's arm arm arm hard float with something missing so we can't run arm hard float. And then you have army army L and how much I know you got on 64. And it's like this is just a chaos. Oh and you've got the device trees which don't actually fix the problem. They move the problem to device tree. So we need people to stick together on on the on this and not fragment community by not listening to all of the contributors and all of the parties involved. So anyway I apologize is quite a lot. Uncharge your talk off the cuff 134. Let's take it to questions. Any questions anyone. I apologize if if you weren't expecting this talk we're expecting the one before. What else could I talk about. Any suggestions. I know plug the 60 projects. Yes. Great. I'm on stand a w a w building the old one over there stand eight. The 68 is an eco-conscious computing project which the idea is that to solve the problems of mass volume manufacturing of design for manufacture and design for obsolescence and designed to allow governments back doors to your smartphone tablet whatever. Oh and Google to be able to remotely shut down your little home office system. It is considered to be highly and completely unethical because all of those products will end up in landfill. All right. So that's why I am doing the year 68 project. And I would welcome input and assistance in reaching more people to empower people with their computing devices. Thank you very much. Thanks for your talk.