 Prosesor ini, Ultra-Low Power Core ini Kerana saya nampak kamu mempunyai peluang yang lebih kecil Saya ingin melihat RCC Baiklah, modul ASB32 yang saya gunakan adalah dari Expressive itself Ia juga dibuat oleh EF Robot ASB32 itu mempunyai sebuah kursus dual dan juga sebuah kursus yang lebih kecil disebabkan sebuah kursus ultra-low power real-time clock Saya dari Expressive Saya dari Expressive Apa yang saya tahu apabila saya memutuskan ASB32 yang lebih kecil adalah bahawa ia masih bergantung pada CPU Ultra-Low Power Tetapi saya belum mempunyai kursus untuk kursus yang lebih kecil Jadi cara saya mempunyai kursus ASB32 itu adalah menggunakan kursus Arduino Saya bergantung pada itu Oh, saya minta maaf, saya harus mengharapkan diri Saya telah bergerak ke kursus ASB32 dan menggunakan kursus Arduino Baiklah, terima kasih lebih daripada kursus Arduino Jadi kursus real-time clock terus bergerak semasa kursus main ini memutuskan menggantungkan banyak kursus Tetapi ketika kursus main ini memutuskan kursus main SRAM-nya juga memutuskan kursus aplikasi Jadi ketika kursus real-time clock memutuskan menggunakan kursus ASB32 Ini adalah cara saya faham daripada memutuskan kursus ASB32 Dan kemudian kursus ASB32 memutuskan kursus ASB32 Bukan kursus ASB32 Saya minta maaf lagi tentang RTC Kau perlu menggunakan kursus kursus ASB32 Tidak Kenapa saya mempunyai kursus kursus ASB32 dalam projek saya yang memutuskan kursus ASB32 untuk memutuskan kursus ASB32 Saya tidak memilih kursus ASB32 dalam RTC Saya fikir itu sangat membantu kerana saya tidak memilih kursus ASB32 dalam kursus ASB32 dan saya kata, oh ok, saya baca tentang itu dan anda menggunakan kursus ASB32 Jadi itu sangat keren Saya ada pertanyaan dengan MBIOTS Bagaimana menggunakan Kursus Laura atau TMIWI Span? Itu hanya pilihan untuk kawan saya yang ingin menjelaskan kursus ASB32 TMIWI Span walaupun nanti lebih memuat TMIWI Span Kursus ASB32 tetapi anda menggunakan TMIWI Span dan anda menggunakan kursus ASB32 Saya menggunakan Kursus ASB32 dalam kursus ASB32 Saya tidak fikir itu sangat baik Saya tidak mengenal dengan semua ini Saya cuba menggunakan semua kursus ASB32 dalam kursus ASB32 dan saya melihat kursus ASB32 dan saya melihat kursus ASB32 dengan kursus ASB32 lebih dari 1,000 kursus ASB32 Saya hilangkan Saya menggantikan kemungkinan untuk memberikan yang saya mahu Saya ingin mengenal untuk berikan kemungkinan dan anda dapat melakukan kursus ASB32 dan saya juga membuat kursus ASB32 Kami membuat kursus ASB32 dan kursus ASB32 pada masa yang sama? Kami sudah mempergunakan bahagian yang kami membuat dan menggunakan dari kursus ASB32 Jadi apa lagi kursus ASB32? Kursus ASB32 adalah kursus ASB32 It can also have two million samples per second, but I recommend to go about 100kWh to 500kWh, but in our detailed experiment in schools, we usually normally don't go beyond that boundary. So the SLM is kind of okay for school experiment, but if you want to go to high range, there is a problem, but since this one is very cheap and portable, and it has so many features, I think we are kind of okay with the features that are going to be added. I actually really like the included current source on it, because it's like super useful for science and teaching. It's like, it's not enough current source. Everyone thinks in terms of both. We're fascinated by, when you talked about verifying pieces using reserves, you say something that's died in the industry, I've never heard about this before. So the question was, is the laser for verifying ASICs a thing that's common, and the answer is no. Is it a technique? As far as I know, I had to come up with it on my own, and actually, literally part of the reason I'm presenting it here is because I'm trying to keep other people from patenting it. The first thing I've gotten from several people I've talked to is like, are you going to patent it? And actually, I'm a little worried because I'm collaborating with some, by talking to some companies, with policies of patent, everything they touch, regardless of it invented or not. And so I'm trying hard to disclose, put in the blockchain that I have it there, get a prior date. I just want to not have this exit, like the domain of people to be able to use. Have you tried it yourself? I have not tried it myself. That is one of the things that needs to be done, but there are supers that have indicated, particularly optical phone induction for the purpose of maliciously attacking. And MCU has been demonstrated by Ross Anderson's lab in Cambridge, but it's on a much older generation of chip. And so one of the things that needs to be done is on the target geometry, like 40 nanometer, how does the extra carriers that are generated in photons get directed, as long as there's enough interaction to cause a bit to flip or something like that. So that's a core assumption that doesn't need to be validated. But I'm pretty optimistic because light does interact with silicon, will generate carriers, and then should at some strength cause a problem. Just the question my head is like, is there a strength at which it can cause a latch up instead of verification or something terrible like that? So all the open source projects I've seen up to when, they generally are characterized as products, nothing too dangerous. What if an open source project causes the time we protect ourselves? I'm going to have a related anecdote for that. That's going to be my next question. I was interning at an open source software for drones project a couple of years ago. And around that time, there were news about ISIS using chariot drones to send bombs over. Now, this was an ethical question even then. One of my mentors kind of gave me an answer then, that it's worth the cost. It probably did not phrase it correctly, but it's still better open sourcing it. So I just want to kind of ask that in that context. An open sourcing project for building drones results in not nice people using it. Is it still worthy? It's an open question. Delicious use of open hardware. Once we put something out into the world, the question is about, what if someone uses your project maliciously if it's open source? That'll happen whether it's open source or not. When you put something out into the world, even if it's patented, you have no control over what people do with it. If someone's going to do something malicious, they don't care if they're stringing out a patent. They don't care if they're violating an open source license either. If they want to harm you, they can say, yeah, Mitch Altman did this thing and I'm giving him credit for my malicious act. That's not something I'd be proud of. But once you put something out in the world, you have no control over what people do with it. I think it's really important to think about what's going to happen in the cultural and social context with your project, what will it best be suited for and that's what it will be used for. But anyone can use anything for harm. I can, not that I want to, but I can take this mic and start bonking. Only give people on the heads. That's not what microphones are intended for, but I could physically do that. But if I make a gun, clearly it's going to be used to harm people. If I make a drone, well, we already know what the US military has done with drones. It was started by a bunch of German hackers and it was presented publicly the first time in the first hacker conference that I went to in Germany in 2006, I think that was. And it was awesome. No one was thinking that this was going to be used to kill people in Yemen. I think I just put a couple of things. First thing is the most important part of every open source license software or hardware is the don't sue me clause. You use it at your own risk. So software in particular is like this, I don't know if the code has no fit because whatsoever basically, you can't quite do that with hardware and say this hardware is rick and therefore you should pay me for it. You have to warn that it does what it does. But there is a limitation of liability there. The other thing that I would add is that a lot of technology is dual use. Almost every technology, like when we harnessed fire, we could cook food and we could also burn people on the stage. This was just the moment technology came into the hands of humans, technology itself I think is neither good nor bad. It's like water, you need it, but you can also drown in it. And then the question is just in the process of disclosing and sharing what are the channels you use and how do you express it. And then yes, there's a step that has to be taken, for example, to weaponize a drone. There's a step you have to take to weaponize gunpowder. There's a step you have to take to weaponize an exploit. There's a step you have to take to weaponize any type of thing to make it particularly effective against humans. And so if you are trying to tweak the parameters of your technology to be optimally harmful to humans, then that's something you should ideally avoid. But if you're just trying, if you're like, oh, I want to play around with high voltage circuits that I think is really interesting. And you're not saying, well, these electrodes are really good for shocking people or something like this. But you're just saying, no, this is how you do high voltage and you have to be careful and all these sorts of things and you put a circuit that can generate a thousand volts and it can hurt and you tell people about that. I mean, there's a lot of really useful applications for high voltage sources and I think we should have that out there and make it available for people to use. I have a follow-up question. So, for example, GPS doesn't work above certain altitudes. Commercial-level GPS doesn't work beyond certain altitudes or speeds. And that's has a limitation so that you can't just make a missile out of your garage. So, there have been limitations that have been placed. The question is do we need those kind of things? We need certifications. For example, I come from India. We have passed a law which says that if I have to buy a drone which is bigger than a particular size, I have to register with the government. Right now, this kind of gets in the way of me building my own because then I have to follow but it's also not great for anonymity because then I have to register my government idea against a drone and all of those things. So, do you guys think these are good enough measures that we can put? Because finally we are the ones who are making these hardware and we are the ones that we have to come up with these ideas come up with these certifications faster. And then there are some open source gun designs so yes, you can. So, people are discussing in the audience about should there be government regulations, should there be limitations to technology to try to prevent malicious activity with projects that we create. And if we put the limitations in then that limits what we can do with them at the same time it limits the harm that can be done. So, there's trade-offs. In the United States, there are a bunch of people dying from guns all the time and there's laws that people need to register their weapons in order to use them and that has been effective at limiting the number of murders with guns. Of course, there are people who will just get a gun beyond the registration also but it at least prevented someone from being super bad going out buying a gun and killing someone while they're still hot. Maybe there can be government regulations. I mean, if we have governments that will have them doing something useful rather than just taking our tax money but these are all trade-offs. We don't live, whoever you are and whatever your points of view everyone will agree we don't live in a perfect world and no one can even define what that means or all that we actually live in and try to do our best. So, if you feel it's your calling to try to push for government regulations to limit certain kind of technologies that can harm people maybe that's something to do. But for myself, I've worked so many different places as a consultant in every single place I've worked. The US military has come in and said we want what you all have done this cool computer game can be used as a killer helicopter simulator and we just get a few customer tweaks here. I helped create virtual reality and people from the US military came in and they wanted to use it as a World War 3 training simulator. So, I don't want to help to do that kind of stuff but when you do anything if it's a new tool it can be used and therefore it will be used to harm people because we've got 7 billion people on our planet and some percentage of them are highly paid to look at any new tool that comes out, any new technology that comes out to be used for weapons of mass destruction and those people work for the US military and other militaries. So, what do we do to prevent that government regulation isn't going to stop that. This is a huge can of worms and we're talking about ethics and there's no absolute right and wrong but what do you feel like you're working on a project if it's something that you know is going to be used for harm but you're doing it anyways that's probably not a good idea but if you're enthusiastic about making drones like these German hackers and it's really cool and they didn't really think about it's going to be used to kill people in Yemen and other places eventually but we live in a world where these kind of things are going to happen so I think it's our responsibility as creators to think about that to the best of our abilities and put it out in a context of coolness rather than harm. I think it's not just we should rely on governments to release us but I think having this conversation in this crowd is very helpful and that if you see someone remarkable like hey maybe the AI processing you're doing is a little scary and creepy and you should think about pulling that in and you just telling your colleague that might actually be much more impactful than any government telling them might be wrong because ultimately it comes from community standards and we have to sort of ask individuals sometimes just say hey it's not like you're bad or something like that rather than maybe they could be creepy or they could be bad or something like this and then just make them think about it and then I'm sort of a fundamental optimist that if you just 99% of the humans plus if you tell them that they'll probably do the right thing there's a small percentage you really do have to enforce that but I think that's a good start Given the same trend of how far open source hardware is and each of you share the most complex or challenging open source i'm sorry i've been at your scene online it's challenging and complex can you take a share? yeah it was funny i concur no come on no no i started with open source given the positive level and it was pretty complicated and i i'm not much happy with what is left of at the moment PSL was my most complicated i think actually if you expand the definition of open source to include like the realm of like open source licenses themselves sort of like like all of the physics papers and stuff from the 60s and when people actually disclose and publish things there's some like incredibly cool stuff that people did like in the 60s 70s and 80s and you can download the schematics and look at it and just like you know i think you actually download like apollo moonlanders designing now that's like an incredible piece of hardware it's so cool there's so many people had to do it so little back then like they landed people on the moon with like 2 kilobytes and ran and then i tell people i won't put 4 megabytes and be trusted and if you can't do 9 people on the moon with 2 kilobytes and they roll it by hand with corn right i mean it's pretty basic actually that was going to be my question can you expand more on the nebulis the CPU and that subsystem and if you'll be trusted like what kind of thing you'll be looking at because i.m.e is going to use a lot of rand of course just like human language is like that so the question was can i talk a little about what i'm thinking about for the chip inside be trusted so the logic goes we define what we want to do with it so it's the text base for the i.m.e and sort of the voice and you work backwards in there to what the size of chip and you don't make it the biggest chip possible because it's too expensive and what you find when you work backwards is that the limiting factor is rand how much rand you can put in there and so this is why i got onto the verification of deeps of nanomere silicon because a bunch of guys are actually working in open fabs in 130 nanomere range but for like a chip like 4 megabytes of rand is like a chip this big in 130 nanometer and in 49nm it's a chip so it's a reasonable amount so then once you pick the note say 49nm you pick that because it's like where the mask kind of fall off and become really cheap anything beyond that the price starts to go up again then the trend just is all perform a certain rate so you know they're basically free as far as the CPUs can start to be 30 to be at risk 5 computer with some crypto accelerators it'll probably run at hundreds of megahertz and you ever need for the application but it's going to be about 4 megabytes of rand the flash will be off chip but then signature verified and pulled in so you can have arbitrary amounts of off mic storage and then probably also integrate with some ADC some housekeeping stuff but try to keep it very minimal and the idea is it's a very minimal design just a big pilot rand and a 30 to be at risk 5 CPU when architecture is really open and then on top of that the next module on top of that is in order to facilitate with the silicon inspection basically every register gets instrumented with the readout path so the idea is you basically stop the clock fire laser in there and you say okay we had this pattern before now after fire laser what's the pattern of all the registers you read that out and you say okay well the laser must have been in this set of logic because these are the affected paths so there needs to be some readout path and then on top of that there's another modules which they're blocks even as a designer so I'm going to be as a designer I get a map of the chip back showing where all the standard cells that I designed but for example the rand itself is a hard IP block which means that the foundry puts it in I don't get to see the rand I never actually see the rand design so they can put implants inside the rand so I have to harden against that and swap around the address space randomly and doesn't much stop to make it so if you put something inside the rand block it's not like oh I just passed over byte 8 and you always get the root vector it's now you have to put a much more complicated thing inside the rand which means I'm going to detect that as I scan through the device so there's a bunch of little tricks like that that it's really more about exploring those tricks and not so much about necessarily building a quad core or get a multitude of tricks you know whatever it is thing that a lot of people are really interested in so I think I ate it all the time Thank you very much Thank you very much to the panel for this session