 respects freedom. Our next speaker is Nicole. She's currently the CTO of purism and she has this role because she's been involved in embedded systems engineering for a long long time and now she will talk about how to make a phone that is actually truly free. Please give her a warm round of applause. Yes hello thank you very much for having me here. We have a lot of ground to cover so please excuse me if I'm a little bit frost. So what I want to explain is the story that has different facets and it's about a quest to create a mobile phone that respects users privacy and uses freedoms and that is really truly free and I will explain what I mean with free and it's also a story about a journey that we started over two years ago now and which is still ongoing then because making a smartphone or mobile phone these days is incredibly hard and that's also why it's called hardware. So the kind of freedom that I mean is the freedom not as in free matter but more as in freedom and it's similar to what Paul explained just a few minutes ago what he called sovereignty what I call is more freedom it's similar to the ideas of freedom that the Free Software Foundation has which is the freedom to use it for any purpose that you like to study it to change it and also to be able to share these changes because we at Purism and many others in the open source and free software communities we believe that only if these are guaranteed then a system can be truly trusted and it really can be made in a way that it preserves the user's privacy. So to understand what the challenges are you have to look at the mobile phone market the current market every year I don't know how many of you are aware of that about 1.2 billion smartphones or phones are built worldwide so this means that every seven years every human being on the planet Earth could have a new phone which is really quite amazing. So what this also means is the mobile phone and smartphone market is gigantic it's really huge about 30% of these of all these smartphones of one of those 1.2 billion are made in just one single city which is Shenzhen in China. So the largest wealth of experience in making mobile phones is in this location which is also quite amazing and this also led to this enormous experience and amazing experience that we just had by building the phone we went to Shenzhen and asked arbitrary companies for our project to build this phone and within half a day we got a quote for designing the PCB for our phone and do this in the Western Hemisphere and Western World engineers will ask the hell out of you. So this is totally amazing what is happening down there they have the biggest experience and they are really fast to do things. Looking further at the market you will quickly find out that the majority of the market is currently just governed and ruled by two major systems and two major companies which is Google running the Android platform which is about 85% of the overall global market and Apple iOS which is almost 15%. So basically almost all the rest has run into insignificance so where we used to have Symbian and all these other different platforms there are none these days so it's mostly dominated by Android and Apple iOS and this has a huge impact on the hardware market also. For making a phone you need chips and these chips are made by chipset makers and there are quite a number of chipset makers worldwide and also in the chipset market there you can see the dominance of certain companies and one of the major players here is Qualcomm. Qualcomm owns more than 40% of the smartphone chipset market currently then 20% about by Apple 15% by MediaTek which is also an Asian company mostly Chinese little bit of Samsung and then we are 15% others like Huawei and others. So there are only a very few companies actually creating these solutions. If you look inside just the Android ecosystem because the Apple iOS system is pretty close inside the Android system this ecosystem is dominated by just two which is Qualcomm and MediaTek and if you then think that 85% of the global market is Android this means in other words about two thirds of the smartphones that you find out there is just dominated by two or three companies if you can't Google them which is really really amazing so that's a huge market and very very few companies actually ruling this market but then there's not only hardware but also the 3GPP standards so if you want to introduce a device which runs on those sell in the networks you have to comply to those standards and these standards are governed by the 3GPP organization which is the third generation partnership project in short 3GPP is a standards organization which develops protocols for mobile telephony. These 3GPP specifications are in general open so you can download and read them and try to understand them it's huge but they are not so free to use and here we come into the pretty ugly area of patents and especially software patents and I will come in a moment to why this is all important. So 3BPP standards are covered by literally thousands of patents and every new generation from 2G to 3G to 4G and now 5G the number of patents grows almost by a magnitude so we're talking about thousands of patents and due to the patent legislation and the legal restriction or the legal legislation for patents a patent is granted in country or jurisdictions and this is why we have about 80 million patents worldwide on 3GPP standards and about 250,000 patents which are standard relevant that means if you want to create something that runs on a cellular network which complies to 3GPP standards then it's about 250,000 patents for 4G which are potentially in your way or kind of in your way which you have to which you have to use and the larger patent holders are in this order basically Qualcomm, Interdigital, Ericsson, Nokia, Huawei, Samsung and a few others and the first two or three have more than 50% and by coincidence Qualcomm is also one of the number one chipset makers especially in the Android area so this means we have a market which is controlled and dominated by very few companies so why is this important because of this the smartphone is used today everywhere by everyone so it's a dominant platform everywhere on the planet and growing many people or the smartphone has by far surpassed the relevance of ordinary computers many people have ditched their PCs for in traded at this in for a smartphone or a tablet and even tablets are now running out of importance smartphones are ubiquitous and universal and yeah so smartphones have become the platform the computing platform of basically almost everyone's everyday life and they store everything about us our whole digital life communication all means of communication text emails voice calls encrypted are then not encrypted of course also location movement everything secrets we store on our smartphones like passwords or access tokens to almost everything our bank accounts so this is exactly why it matters to create something that we own that we control and like Paul said that we are a sovereign of because this is our digital life that is controlled by these devices and we entrust these to just a handful of companies that's critical so how to change this and how to engineer a smartphone or a mobile phone so the things you need is basically a CPU which is in the embedded devices usually in SOC system on a chip which controls memory storage RAM things like that it also drives the screen through GPU IPU and all these things controls the cameras and different peripherals peripheral interfaces so that's the main CPU and then you also need those wireless interfaces that everyone currently expects from a smartphone these days which is wireless a wi-fi bluetooth and of course the cellular modem it's a phone so you need a cellular modem for all the current technologies this sounds kind of simple but it isn't so the main challenge in the smartphone is first of all the size it is incredibly small and everyone expects to be very very small these days and many of these components the CPU the SOC and also the radios consume a lot of power and so this itself is a challenge and this challenge is shared by all hardware manufacturers they want to make a mobile phone which is also the reason why the chipset makers like Qualcomm MediaTek and the others have created solutions for that and remember the market is gigantic so there's a huge demand and the demand is of course answered by those manufacturers and they created system on chips which integrate all of these things including also the radios onto one piece of silicon which is perfectly neat and perfectly fine for many of those smartphones very tightly integrated and conserving power and storage and all these things but this has some problems the advantage is that for example the cellular modem these days has become incredibly complex and you have seen this or I talked about this with the patent situation there is a lot of knowledge that goes into developing the firmware for such a modem to control the radio and all the protocols that go on so the modem firmware itself and controlling the radios is extremely complex needs a huge firmware and if you have it into the same SOC silicon you can share a lot of the resources that you would usually just also use for your main CPU that drives your applications so you can share the RAM you can share the flash and a lot of other things and you also also have fast data interconnection from your host CPU to this modem so that's very convenient same saving of course also applies to Wi-Fi and Bluetooth but it's less than with the cellular modem so these 250,000 standard relevant patents in 3GPP 4G are pretty much not in vain because it's so complex and you can see this also if you look at the the binary firmware of a cellular modem these days these are usually 30 megabytes just binary firmware of these modems so it's really huge piece of software which then runs in your SOC and has access to the same Rambos for example as your applications which is a concern as also Paul laid out you're probably not sovereign of your own data anymore because the modem could spy on the Rambos and there was fun fact many of those modems actually run the linux system and there was a very good talk at the 33c3 by Harald Wielter and Heuger Freiter which you can look at at this URL and they actually dissected one of those binary blob firmwares and now even if you wanted to do something like a free modem implementation implement these 30 megabytes of binary thing and do this it's really really hard because then you have to license all those patents the 3GPP patents must be licensed there is legislation for that by the patent holders under so-called friend terms which means fair reasonable and non-discriminatory but these patent holders still use the patents to control the chip market and this also led to one of the lessons learned that we did when we first started out the research for doing our phone that we learned that there are only very few back in 2017 very few modem modules that you could buy which would actually allow you to make voice calls because at that time modem manufacturers especially a company called Qualcomm wanted to sell their smartphone chips for smartphones that do actual voice calls and they didn't want to sell modems to others to also do modem to also do voice calls and build phones with that so it was pretty hard at that time to find decent modem and to have something customized for us even that we are talking about 10 000 devices that we want to build modem manufacturers like these wouldn't talk to us at all so integrating a modem and asking for custom firmware something like that for us at our size even talking about 10k not an option and so here you see the the grip through software patterns from manufacturers on the market and what the market can actually do so what we wanted and now coming to the actual realization of what we do is we wanted to create a phone and a smartphone which is binary blob free at runtime that means that the host CPU does not have to handle any binary mystery code that to download to any devices we don't want to use any proprietary drivers no close source drivers no close source firmware and we wanted to have the radios cellular the wi-fi bluetooth separated from the main cpu also to reason to what what paul explained to be sovereign over there over the data and it should be based on free software only we're striving to get um respects your freedom certification from the free software foundation and of course it should be a open design hackable for everyone schematics will be available to everyone without nda so it's completely you can own it and you can change it and you can do whatever you like with that what we came up in 2017 as the solution on the best solution that we could find at that time was to use the nxpi.mx8 quad as the main and host cpu which features for cortex a53 cores running at 1.5 megahertz or gigahertz not megahertz we have enough peripheral interfaces for driving screens and cameras and all these things and all the peripherals needed and also for sensors but most importantly the idot mx8m at that time was the only soc that supported free drivers for the gpu the the graphics engine that drives the screen in the end and which allows for graphics acceleration and this is the gc7000 light from vivante for which there is a free software implementation driver which is called etnaviv which is vivante backwards and for this you have upstream mainline kernel drivers and including and the drivers also have been pushed upstream to mesa already and this is working pretty nicely already today there are more choices unfortunately now two years ago they haven't been just to be fair the mali gpu is also working quite nicely now and also the the port of the free drain or driver to the adrenals gpu's of qualcomm also become pretty much stable and available what we also wanted to do is to have the radio separated out of the cpu so that they don't cannot interfere with the data that is stored on the cpu and then and the applications that are running on the cpu on the main cpu so we separated out the the wifer and bluetooth on an m.2 card and this will also be implemented in this way in the final phone um m.2 is a pretty much well known standard it's connected using sdao the firmware sits on the card the host cpu doesn't touch it for the cellular 4g we have two options also sitting on an m.2 card um one is the gemalto pls8 which is made in germany and partially in the us the more the card would be made in the us and also the broadmobi bm818 which is made in china um the broadmobi is available for more regions of the world and that's why we chose the second one and in theory for the phone later on any m.2 card that that can use the same interfaces as we use them now and are supported in the firmware and in the hardware can be plugged into the device and also other radios if the antennas of course will be tuned for the 4g bands and for wifi and bluetooth and if you have another radio they're going to use on there of course just plug it in for example for these uh hf radios that also palm mentioned would be a possibility and it will be very interesting to see then what we also implemented in the hardware our hardware kill switches um so that you have a switch that really physically severs the hardware like the cellular baseband the wifi cameras microphone and also the sensors from the host cpu so that so that even a malicious acting um application cannot get access to the to this data and do some bad things for you so your privacy can be really safely and secure what we also did is we integrated a smart card rear so that you can use a smart card like for example the open pgp card or arbitrary other smart card to store your secrets in a very safe place like um encryption keys um and also the the smart card reader for this we have a separate project we're collaborating with vana kochen niby utaka here um for implementing a free sir a free firmware for this smart card reader which is based on an stm32 l432 kc which is a microcontroller and this will be accessible for everyone so what we did in the first place is we started to create the development kit end of last year and this is part of the pcb i will show the real device in a few seconds um and this we did based on a so-called system or module which implements all the heavy lifting for ram and flash and all these things and we designed the baseboard for that so that we get something out some hardware out to developers that are interested in developing the software and helping with developing the the software as soon as possible and we started to distribute these at the end of last year and this is the real hardware probably cannot really properly see the picture but you can see the separated modules here for the modem and the wi-fi and the system on module in the middle and the phone screen on top and this is actually working now um that's pretty nicely working the software is making progress and next i can show you very proudly this is the first boards that we just got received from from china that's the first time that these are shown to the public for the final device which has been developed in parallel in china uh this location where they have this incredible huge experience and this is one side of the board this is the other side of the board with the cpu in the middle and you can also see the two m2 slots on there the gnss module and a little further stuff so we are getting there but hardware is hard everything takes much longer than anticipated of course and to give you a idea of the complexity of that board we have on this board more than 160 different components and that's over 1200 parts placements on these boards so that's an incredibly dense board the pcb has 10 layers so 10 layers of routing the smallest parts are 0201 for the electronic geeks now the smallest ball pitch is 0.4 millimeter that is the the distance from one ball to the next of those bga devices the smallest diameter of drills and holes for example for vias is 0.2454 millimeters which is incredibly small and all of that is condensed onto a pcb which is about the the size of two credit cards and this is what hopefully the final device don't don't name me down on those size numbers in the bottom right corner but this is about the the design that we intend to have in the end so we're getting there so this is the hardware part on the software side of course as already mentioned we want to have everything open everything shall be free software on that so everything what we do inside of purism for the project will will and is copy left license so gpl or l gpl v3 and all the contributions that we do to upstream projects which are quite a lot are of course according to the upstream projects our motto is upstream first that is if we can push our changes into the upstream projects then we will definitely do so this has several reasons one of them of course is maintenance but the other is of course also to avoid obsolescence of the of the platform and of the software we want this software and this platform to to strive and to live on and this is only possible at least in our opinion when the things are mainlines the bootloader of course of the system will not be locked in any way it's a standard u-boot that also comes when with the nxp cpus and the bsp for that the u-boot will will live in the emmc of the board which is the embedded multimedia card which is similar to an sd card but sold out onto the board and we have one tiny issue and this is what where freedom is limited by the hardware and today there's almost no way around it for the ddr forum you need initialization somehow ddr4 is pretty fast and it's parallel signals on the board so the the attachment part that is inside the cpu has to very carefully tune the signal levels to talk to the ddr forum and there are a lot of patterns on doing this and this is also the reason why these ddr4 initialization sequence for almost no platform at all currently is open source the ddr4 controller is a small microcontroller and you have to download the firmware and just mention it here these are struggles that you have when you start to develop things you suddenly find oh there's a piece of binary code that we have to use and there's no way to avoid this so we handle this by a second cpu that is also inside of the idot mx8 it's a cortex on forecore which then pulls this firmware out of an spi flash and puts it into the phi and the phi can initialize dram and we can continue to boot but that's pretty stubborn and i don't like that approach but there's no way around it the kernel that we're running is a mainline kernel currently it's a 5.2.0 kernel all the patches that we do we are heavily working on the vivante driver and also a lot of other things trying to mainline all the drivers that are needed for the idot mx8 we're doing a lot of work that nxp should be doing but well we're doing that porting all the stuff forward and all the patches are sent upstream into the kernel community and get merged over time then on the operating system side we're running a debian derivative so purism also makes laptops and on the laptops we have a debian distribution which is called pure s and we're using the same pure s basically just compiled for arm64 on the phones too which means you have the same packages and and everything available on the phone that you're also used to on the laptops which is pretty awesome which also means that you can install arbitrary debian arm64 packages on this phone we also want soon to introduce flat pack for application distribution this has several reasons one of them is of course the the possibility to sandbox applications and to ease also application development because dependencies can can be handled a little bit smoother with flat packs and the flat pack distribution but this is still in the works for the graphic user interface environment we use a system based on wayland and we had to write our own compositor and shell on top of wayland using wr roots which is which is a toolkit to write compositors and shells and these are called foc and fosh phone compositor and phone shell these are two of the very few parts that we wrote from scratch because there was nothing that we could contribute to for the application environment we chose to go with the free desktop org standard and we chose the the gnome approach and we're using glab in gtk for application development for native applications which works pretty nicely and one of the reasons also to choose this approach is that we are now working very closely together with the gnome community to enable of gtk and gnome application applications and the whole gnome environment to run smoothly on mobile phones also using a responsive defined design methods so that applications that are written for the desktop can be made in a way that they also run on a small screen or smaller screen like on a mobile phone without having to reprogram or recompile them and we're developing specialized widgets for that and also all of this is going upstream currently it's in a separated library but we're working with the gnome community to accept this upstream for those that are interested in participating or looking at this development we have quite some development resources developer purely as mlibram 5 this is the the main side where the development also documentation is stored they can also find instructions how to download a virtual machine image for running on your laptop or pc so that you can try the software on your pc without the device and also develop against this virtual machine of course source pure as m this is the the side where all the source code management and also continuous integration management happens it's all in the public there we work together with gnome and with with the kernel forks then here are some pictures of the actual applications you maybe wonder how this actually could look like and this is how it looks like this is a modified version of the gnome settings dialogue and you can already see that on the left hand side you have main menu of the gnome settings which now folds kind of in this responsive way it's the different views of the different settings on the right hand side you can also see the first implementation of a virtual keyboard here then here are some other examples how to do this so let's see initial setup customized for the smaller screen which then also works in the same way on desktop which is pretty nice here we have the composite and shell working and it's not just the shell it's really a shell shell on the right hand side of course you want to have a shell on a mobile phone here is contacts setting up so a lot of things already working and being adapted here's a web browser we use the the gnome web application for that which is based on webkit on the left hand side there's a web page the purism web page in that case on the right hand side you have the the the tab handling how open tabs are handled in the browser which is actually pretty neat you can flick through the open tabs and do something and now this interesting time oh we are running out of time i'm afraid but in general i have a dev kit here which is this thing and in theory someone could call it but we don't have the time now to wait for that i'm afraid but it's working so the whole software stack is running we can do this later on you can find me in the cows west village and we can do this there if you're interested in now summarizing a little bit of the challenges so we have seen there's a lot of parts that we have to do to create this mobile phone and the one or a few of the major challenges are of course handling suppliers um because if you want to do a hardware you have to work on all these suppliers and work to work with them to get the supplies which is parts PCBs parts for the case and all of that this is an immense effort so developing something is one thing but getting all those suppliers in line is really a huge undertaking then the other part is if you want to do something with computer consumer electronics and to put this into production and seers production and mass production at the moment there's almost no way around china because the whole production facilities are there there are the people that have the experience there are the people who can crank out stuff really quickly and at a decent price it's not cheap anymore i can tell you but it's at least a decent price but it's very tough to work with china because of the language barrier only very few people speak english down there so you always have those translators and interpreters in between and this really makes collaboration really hard additional to the time zone differences and then also consulting suppliers the components forcing for an electronics product like this and especially if you want to build a thousand or five thousand or even ten thousand you really get into trouble we have on some components we have lead times up to 20 weeks so this means that you have to plan all your logistics way in advance before you start the mass production and this is a huge challenge in undertaking and then of course there are all those regulations and certifications especially if you have an intentional radiator like the cellular phone or the cellular modem which does radiate of course in certain frequencies and to be compliant to all those regulations that's a huge undertaking you have to go to laboratories doing all those certifications and things so that's really challenging especially for a company that hasn't done this before the other challenge that we also see is that due to the lack of hackable hardware in the mobile space for the past years because android has become so dominant there's only very little knowledge out there in the free software communities about these devices and how those base technologies that are needed for those devices can be done and there's only very little free code out there so this is also a big challenge and was a big challenge for us and we are now working together with the different communities and empowering people handing out those deaf kids and trying to create this community because we think that we must change this we need to free this knowledge and enable the hacker communities to liberate people from this oligopoly of a very few corporations who currently run and govern our digital life and clearly at some point they also exploited just look at this Cambridge Analytica scandal so and with this i want to close we need power to the people free the mobile and hack the planet thank you very much well thank you for this great talk so everybody we have about 10 minutes for q and a you know the drill there is a microphone angel somewhere in the middle i literally can't see anything but i swear he's somewhere there and just wave at him or queue up behind him and then we will get to you and if you're on the stream hop into the isc or on twitter and then the signal angel can read aloud your questions do we have people wow do we have any questions from the audience yes okay please go ahead thanks a lot for your talk welcome i think a big challenge is the application environment only i've heard that only china has more than one million android application developers and we need a lot of good applications on such a nice platform what do you think about that and maybe there is an option of running android open source project or selfish or as on your platform what do you think about it okay so that's of course one of the number one questions how do we want to handle applications so of course first this device is not intended to be the replacement of the average android phone so we don't expect to have those millions millions applications there what we want to achieve is to create a platform that is conserving and respecting the user's privacy and also be secure and these for for this we only need a handful of really good applications like a web browser something that does the calls handles messaging and all these things the the most basic applications and what we also see in the in the community around the the liberum five right now is that a lot of people are very interested in working for this platform creating applications of course we will not be quickly at this one million application or developer stage no way no but we're getting there and for the first time or for for the beginning it will be dual use you have your android platform with different applications and then you have this other platform which is probably the liberum five which conserves your privacy and where it can be secure about your digital life and probably at some point we can also talk about enabling running android applications there are some open source and free software projects about this which could be possible sailfish i'm not so sure if this works in parallel we can support qt applications in general but a whole sailfish i don't know how this can be encapsulated thanks do we have another question i think there's someone walking up please go ahead yes hello thank you for the talk i'm this is a very exciting project um i was wondering if there's people running the development software on other devices and if you're if purism is supporting that at all or interested in having this not just be for the upcoming hardware which does look cool but there's a lot of old devices out there yeah of course so um you have to keep in mind that purism is a quite small company so um supporting other projects is tough for us but we do our very best so if it's something that we can explain about things that we did of course we're totally open to that but of course the the liberum five is currently the the main driver for us and this is also the thing making hardware costs money employing all those people developing this this software stack and also the hardware costs a lot of money and we have to take we have to make some money and this is why the liberum five is of course our main target but anyone else can join our communities on the matrix channels and get in touch with us we do our very best and we want this platform to to be yeah usable as widely as possible and now a question from the internet yes um is there any information on the cellular modems firmware a source available user adaptable or protect against updates by digital signatures that's a very good question so um the short answer is no because we're dependent on module makers and modem makers that provide these modules and modems to us and with the 10 000 devices that we're intending to make we don't have much leverage on those modem makers and so the firmware is binary only we cannot do anything about it we just can try to separate it as much and as possible and to control the interface to the modem but there's no much way to control the firmware itself and this is also the reason why we have the separation and um yeah the separation yeah sorry okay from the room hi thank you that was a very interesting project very noble intent um paid to be the beam counter how much are you expecting to charge for your devices um oh i'm not the right person to ask for about finances currently you can pre-order the phone right now from the purism web page and if i'm not mistaken it's seven hundred dollars right now six ninety nine does anyone know approximately seven hundred apparently um do we have another question hi thanks for the talk and i want to ask you some clarifications i don't know if they a hundred percent configure as a questions but they could be answered as yes or no so maybe i'll get away with it like um i from what i understood in the end you sat down for um ips like cpu and gpus and communications that were not open right in the sense of the schematics were available and the other thing was did you do the integration into so c of the ips by yourself or did you have to contract subcontracted to people that maybe because what i was interesting is if it's something that is doable with only open source tools and the very last thing is the gpu drivers that you pointed out for the gpus um are they reverse engineered open source implementations or are they actually maintained by the vendors that was one day thanks sure um so i hope that i got the the the question right um a lot of the knowledge that we used to to um upstream the the things and implement them in the kernel entity into mesa is based on the on the bot support package that comes from nxp so there are a lot of drivers they're kind of free software um the problem with those bsp is not only from nxp is usually that they have a huge wealth of code which is not suitable for upstreaming so what we do is we kind of cherry pick a lot of things from this um from this vendor kernel and upstream this into the into the mainline kernel and there are some other parts like the gpu driver that you also mentioned which is closed sourced in the vendor kernel the vendor distribution this is an free software reverse engineering project the vivante project which has already been going on for many years started with the imx8 if i'm informed correctly and that's a huge amount of effort to reverse engineer this and by the uh first reverse engineering of the vivante some gpus for the idot mx8 a lot of knowledge was gained so that the reverse engineering for the idot mx8 the gc7 house light that's in there um has been lots easier and faster than it was for the idot mx6 so it's in part it's partly partially reverse engineering partially using code that comes from the vendor kernel porting this upstream and yes it's doable for experienced hackers from the open source community to dive into such a development project but it's hard work and you have to know a lot about hardware internal workings of specific hardware for reverse engineering a gpu you have to know after know a lot about open gl and vulcan and all these drivers how there's work how these work to understand the binary and try to disassemble and understand how this works and trying to develop this open open source driver and integrate this into the kernel framework all right a question from the front i think uh yes thank you for the talk you've shown in the beginning that the actual problem is not that much of a technical problem but it's more like a political problem that there are a couple of companies uh most of them in china that have actually a monopoly or oligo oligo oligolopoli uh of the uh technology so my question is what must happen uh from a political point of view to solve this issue oh dear from a ability point of view it's very very complicated because if you start to rule into companies and try to tell them how they're supposed to work um it gets very very messy very quickly um from regulation political point of view i don't actually know how to solve this problem i think one of the well what politics could do is to um much more support research efforts for developing these things at universities and in a free manner to support free software communities to develop all these things let's say for example the risk five cpu core is an awesome platform for doing that and for integrating a gpu a free gpu implementation into that and for integrating all the other peripherals that are needed for those embedded devices this would help a lot and the other interesting part that politics probably could also do and this is something that's maybe immense to what paul said is to make it easier for for communities and also universities and in general the the ecosystem let's say to develop free radio protocols and access to its radios because those bands that are used for that are highly regulated so it's very complicated to develop something in these bands to develop a free implementation of a modem for example or some alternative to a cellular modem or a built-in network which is alternative to the commercial networks um this would be a very good thing that politics could do to deregulate some things an area where we can freely live and hack and work all right do we have another question or do you want to use this last minute to accept a call on your phone we could probably oh now it's off i don't know why it works perfectly well now we're benchmarking the boot up time under stress yeah but we can do this later on in the chaos west village if you want to it's pretty hot so power management is still an area we have to work on all right then please thank our speaker