Logo TIB AV-Portal Logo TIB AV-Portal

How to design your own chip?

Video in TIB AV-Portal: How to design your own chip?

Formal Metadata

How to design your own chip?
Title of Series
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Release Date

Content Metadata

Subject Area
Buy hardware, write software -- this is the basic rule we in the FLOSS community followed for many years. But things are changing. Today it is easier than ever before to create own digital hardware, aka. "chips." In this talk I'll show give a introduction into what (in terms of tools, knowledge and other factors) is required to get a digital hardware design up and running. I'll also show how to get started: where can I find the community to get help and existing code? What existing projects can I contribute to?
Authentication Slide rule Computer animation Software developer
Statistics Code State of matter System administrator Software developer Source code Database Bit Line (geometry) Coprocessor Mathematics Kernel (computing) Process (computing) Computer animation Lecture/Conference Query language Reduction of order
Area Digital electronics Constraint (mathematics) Open source Word Computer animation Peripheral Lecture/Conference Term (mathematics) Computer hardware Summierbarkeit Computer-assisted translation Speichermodell
Computer program Programming language Dataflow Scheduling (computing) Open source Code Real number Software developer Source code Projective plane Bit Cyberspace Hardware description language Category of being Word Computer animation Computer hardware Energy level Library (computing) Speichermodell
Computer program Programming language Open source View (database) Electronic mailing list Computer simulation Energy level Function (mathematics) Logic gate Descriptive statistics Physical system
Computer program Functional (mathematics) Implementation Real number Range (statistics) Microcontroller Regular graph Mereology Message passing Befehlsprozessor Bit rate Lecture/Conference Computer network
Server (computing) Theory of relativity Computer animation Open source Different (Kate Ryan album) Real number Projective plane 1 (number) Bit Form (programming)
Axiom of choice NP-hard Group action Open source Code Multiplication sign View (database) Insertion loss Student's t-test Coprocessor Revision control Hardware description language Computer hardware Formal verification Software testing Extension (kinesiology) Physical system Area Programming language Software developer Model theory Computer simulation Bit Unit testing System call Process (computing) Computer animation Software
Computer program Randomization Functional (mathematics) Implementation Computer file Open source Code Java applet Real number Execution unit 1 (number) Water vapor Cyberspace Function (mathematics) Student's t-test Run-time system Code Revision control Lecture/Conference Different (Kate Ryan album) Formal verification Computer hardware Cuboid Software testing Software framework Compilation album Area Projective plane Keyboard shortcut Computer simulation Bit Unit testing Compiler Proof theory Word Exterior algebra Computer animation Software Formal grammar output Waveform Whiteboard Freeware Resultant Library (computing) Gradient descent
Point (geometry) Implementation Standard deviation Projective plane Electronic mailing list Cyberspace Field programmable gate array Neuroinformatik Product (business) Arithmetic mean Word Exterior algebra Computer animation Lattice (group) Computer network Position operator
Revision control Multiplication sign Order (biology) 1 (number) Bit Whiteboard Coprocessor
Process (computing) Computer animation Open source Hypermedia Model theory Data storage device Control flow Whiteboard Resultant
Area Point (geometry) Pattern recognition Artificial neural network Direction (geometry) Projective plane Bit Data analysis Coprocessor Convolution Befehlsprozessor Semiconductor memory Term (mathematics) Computer hardware
Implementation Greatest element Group action Observational study Open source Software developer Virtual machine Tube (container) Set (mathematics) Parameter (computer programming) Coprocessor Rule of inference Dimensional analysis Connected space Power (physics) Revision control Process (computing) Computer animation Circle Logic gate Resultant Library (computing) Software development kit
Scripting language Dataflow Open source State of matter Code Multiplication sign Compiler Product (business) Number Revision control Goodness of fit Computer animation Computer configuration Physical system
Metre NP-hard Open source Direction (geometry) Multiplication sign Source code Black box Mass Order of magnitude Product (business) Number Facebook Population density Formal verification Square number Software testing Logic gate Error message Area Pairwise comparison Software developer Projective plane Tableau Bit Line (geometry) Connected space Computer animation Personal digital assistant Object (grammar) Musical ensemble Resultant
Game controller Mathematics Computer animation Open source Semiconductor memory Natural number Real number Planning Quicksort Multi-core processor Coprocessor
Computer animation Projective plane Computer-assisted translation Rule of inference
Axiom of choice Group action Standard deviation Software developer Projective plane Source code 1 (number) Set (mathematics) Rule of inference Theory Type theory Computer animation Computer hardware Vulnerability (computing) Social class
Axiom of choice Module (mathematics) Programming language Open source Computer file Code National Physical Laboratory Multiplication sign Closed set Projective plane Mereology Coprocessor Revision control Root Term (mathematics) Computer hardware Website Right angle Position operator Speichermodell Library (computing) Social class
Functional (mathematics) Lecture/Conference State of matter Keyboard shortcut Website Quicksort Open set Computer-assisted translation Library (computing)
Domain name Code Projective plane Numbering scheme Line (geometry) Open set Telebanking System call Product (business) Revision control Computer animation Repository (publishing) Blog Website Summierbarkeit Right angle
Group action Open source State of matter Multiplication sign Open set Disk read-and-write head Coprocessor Goodness of fit Lecture/Conference Term (mathematics) Computer hardware YouTube Social class Form (programming) Stability theory Focus (optics) Block (periodic table) Software developer Projective plane Content (media) Process (computing) Computer animation Repository (publishing) Video game Website
Point (geometry) Axiom of choice Addition Standard deviation Open source Information State of matter Projective plane Cyberspace Product (business) Programmierstil Intrusion detection system Graphics software Website Figurate number Family
Computer animation Fuzzy logic
done carefully I feel this Michael Irvin Newton's of furnaces to going over the authenticity thread-local was 113 um flies need for speed so others was um
I go switch to English and OWL and if everybody's OK with that or do we have anybody who does not
understand English or the other way around not German so you can ask a question either way and let's see the but the question and then I put on us 1st slide is how to design your own ship and to do that let's say the start of the story and actually a story that's pretty familiar for those of you that to softer development I guess that's
actually most people here so it's the story of
Cholly you're a guy that is a actually DBA admin he's doing some database work and the and 1 day he finds this is really weird to
query doesn't just run right it's slow for the unknown reasons that he has no idea was actually wrong so the um and looks at some statistics that guess of the crystals all kind of weirdly wrong but he looks declare itself and should be fast actually so he goes deeper in the state guarantee goes through the whole of the respect he goes to the kernel layer and in the end the it is the use of CAM although the C code looks right the stuff that's executed looks right it's just for some reason so you digs down further and th looks at the C code that he sees and um he's he's well this kind of mural 1 line of C code that is actually the hot spot in this uh database most query is actually turning out to be a lot of assembly instructions some some weird you but rolling something that he knows he could do much faster if you just had a bit of a bit of modification to the processor that yes so it as well a problem just check out the process source code do some changes do a simple simulation and 1st it's doable requests and just 2 weeks later he goes to the shop advisers new processor the that looks unrealistic yeah does and actually it's in the name of this guy in and it doesn't work that way the question is why doesn't work that way why can we do on a development why can we do to development and should we reduce offer development why is it so much slower what is so much more expensive and that's what we're going to have a look at today the the and and the the sorry I'm
gonna start with some words and I'm
seems dating from the slides and sorry but this cat always kind of irritating the the but it's a frost contributions so it has to be there and and the rest of the sum of of words and the word I will use today most digital hardware design and use digital hardware design to kind of separated from the non-digital hardware design and how it but yeah um that is doing PCB that's doing that the things like make a where you have our during the war said you put together that you program and but we will have a look at today is really to the harvested design that means um you create your own chips in the end so physical devices that's what we hope for and the the the term we use is a free and open-source silicon I guess that's a good compromise to get rid of this hardware vs. digital hardware confusion and um yeah open-source hardware is sometimes used for 2 December sometimes used also for this this all make a area and I try to distinguish that because it's uh easily could have used because the techniques and the constraints that we have when designing are very much different for so I don't
know how many of you have and it's hard to read but it's not relevant what's actually on their anyways how many of you have ever read or written some of Barrie Locke code some the HDL code some SystemVerilog code to things that just raise your hand OK that's 20 % 30 % something like that for all the others of user we start um our design with the source code just we do as we do exhaustive development the source code is actually not real source code but there the design guys actually insist on that the hardware description language it's not a programming language so what would you do in the end is you describe how you transistors that make up your chip are connected or how they function in this way the you do them a bit higher level but you're not on a as high level as if you were to say C + + space programming but things like that so for all of you
that have never heard of hardware design I will quickly go through the design flow does steps that you need to take as actually going from your ideas that you expressed in the source code to the final the and you start on the left with you're um HDL sources that the that the schedule source code and you use some other libraries that you get from vendors that you get from others um sources and that's called IP costs the doesn't this is a weird word because intellectual property doesn't need to be very intellectual or property of anybody but that's how it is and yeah that's what we stick with and what you got in the end is essentially your design project the description of of your chip and then you have a couple ways to
go 1st you can do simulation you always get the simulation and that's what you want to do and um
will you get as output of simulations is for a the nice view of a lot of ways so you got of all the signals that you have in your system and it's not visible right now but it's usually also if you have it on a screen it's not much more understandable so it's dividing is really really annoying at this level too but that's the output you get out of the simulation for a while that's nice that's what I always wanted to have a look at the and but
that's not really dead satisfying um Mandela way to go is you take the same sources and to the synthesis so meaning you go from this description in this programming like language to in that list and that list is essentially a collection also um Uvira transistors but and and OR gates and things like that that are connected by wires and if you have the Laplace essentially it's just a grab bag of unplaced wire so you have is that you just like official letters so you have ended adult stars that you're and and OR gates and that they're
really connected to you got this to so we your net of things that is coming out of synthesis and taking the synthesis you can either neatly placed this necklace on a piece of silicon or as you can program EEG on an FPGA and who has ever heard of refugees before just raise your hand again 1 of the majority that's great and so at the epigenesis as had the ships that you can reprogram so they have some fixed the in there and you can have put different functionality onto the piece of this great and at the FPGA is very nice because it in the end get gives you a real piece of hardware store you black thing needed to cancel drawn to something and that as your function that you program so that's actually rather nice Dallas is is that if the G 8 especially as they get larger other expensive we'll have a look at that and um also that the clock rates and the functionality that you can put a message is rather limited so on a standard FPGA in the regular price range of a couple hundred dollars and or euros or but pounds anymore if M you can uh go up to like 2 100 megahertz speeds like that so that's you would use and you can program a couple small um CPU course on their like similar to cortex and 3 cost things like that a very small tiny microcontroller cost so in the end what you get as that is the efficient implementation that's your the 1st part of a chip
and the unfortunately can of those chips are much like the picture we have
here and it looks at the ink on the project to even worse than it does in reality uh so that's all natural including 3 and actually
they advertise those chips as servings of vegetables and never seen that before that of each 1 of Poland's Japan as so that's the chips you get out of that it's not the real nice form of chips that you want to have in
the end so the real guys go for an a back you take the same design sources to a obtain tiny bit different sentences and then the you want to design your own chip to do that you need to do a couple more steps and we'll have a look at that also and in the end what you
get is that the relation you know that affected all the ones that was dig a lot of work but to go jogging afterwards but that's the stuff you want the so let's have a
look at this reality let's start the simulation what do you need to do a simulation of future design how much of that can you do with open-source tools when we need actually to use commercial tools and what's the stuff that's available around the the the the and 1st so for simulation its code obviously you need a simulator and you obviously want to do some testing uh in the software world we have a lot of unit testing of 2 system tests whatever used in agile methods that the extent that you'll find the hardware mold they're much more in the old view model style of development and it has a gives good reasons for that and we'll also have a look at that so testing is a bit different in hard entities students of so but starts write new code writing your code there is essentially true programming languages the hardware description languages that are made used we have very long or its newer versions system area for all of you that have the system very and thought well this Kansas view stuff there don't ever most gross for hardware times have changed SystemVerilog is now very much able to do on a designs it unfortunately has a different language this name the same but that cannot be put in synthesis and put on a chip but that just can be used for verification testing and they met this that choice of naming so anyways that's where we are system is a good way to go the or digital as at the very obvious if you like see you gotta like very lucky if you like + call you gotta like it's still uh so that's kind of the uh uh the holder languages look both look very reviewed an old but yeah well the these are there is a company wants there is respect it's not widely used at all there is chiseled which is a bit more widely used today because there is very interesting process resigned the risk 5 processor that's coming out of Berkeley a sheet of group of David Patterson maybe some of you have heard of the this guy's the guy that kind of 1st the invented the risk ideal abroad it uh to wider knowledge they
doing a prosody is an not was 5 that's losses this coming into play and my HTML it's also there but it's kind of not been widely used in the the on industry so you
want to use and simulation you need to do or you need to run assimilated to do this and there is some places where commercial software is the only way to go and um that's why I mentioned the big 3 commercial simulators you there's 1 from Cadence 1 Fomento graphics and 1 from synopsis the and when you do just simulation you don't actually need to use those you can use whatever you want if you want to do in a in the end you that definitely you will need to use 1 of those 3 and their health expenses and there is open source solutions and actually great open source solutions there is a cursory look forward of simulation there's GDL for um if you want to simulate the HDL code and there is very little which is very very nice project that translates very code into see code and then compiles that and then you get essentially a a runnable program that contains use simulation and it's actually rather fast so good ones and indeed away from you because that's the output of the simulation don't forget that that's the joy that comes out of simulation is this great waveform um the so well and testing a verification 1st of all maybe to things I mention testing and verifications of testing is essentially what you all the stuff you do is offered as uh just run a unit test see OK it works it's nice an verification goes a bit beyond that because you want to make sure dead dead stuff that you produce the output you get from your simulation or Indians of a further steps are actually what you um spacefaring would you wanted to have so they actually apply a lot of formal proofs they use um things like constrained random uh verification things like that we have a lot of random input so you try to get as much coverage as possible using as many techniques that you can get your hands on before you produce a chip and so there is some of the tools that are available in the open-source world that can uh go towards this goal there's Kokaji thank you very nice on project that's Python-based tests as such it is rather that is that industry quite widely I am done by a couple guys in the UK uh 1 there is a unit the and also analyzed very low artistic frame of BH at utility and there is some I O is the methodology which gives you a couple of wider scope all for verification all of those 3 if you look at actually existing open source projects that do whatever design you want you to find any of those you usually don't find any testing it's just like yourself about you should write your unit tests and all that stuff but in the end you just look it away from OK it's fine and that's stupid the the and OK let's have a look at some some results that we get out of simulation so it's it's pretty be accessible for hobbyists and I mean by lobbyists it's um you can get the software for free as abuse you don't need to pay for it but it's not open source you the so you can get your hands on that stuff it's cheap you don't need any hardware except the piece you you have and um so for size the um there's also a lot of open source tools so yeah let's go over that it's not that much fun India and I don't know who has fun looking at waveforms or simulating a chip on your PC yet that's not the defined starts its the but anyways we have refugees and with that the chase you can actually gets as was that kind of some you the self but like the needed does your functionality so what you need again coach simulator testing you could skip testing depends on you and In isn't is a tool you need an implementation tool and you need an FPGA board and we'll have a look at all those 3 in a 2nd 1st of all sentences so you go from and then the the HDL code of area codes student at least that's what the synthesis thus there is um 1 or a couple of major commercial tools 1 of them is a synopsis of I um there is usually all the FPGA vendors also sell or give you for free depending on the exact better um tools that can do so does this there's tools spikes EIling that's currency vital water for the older guys diocese the and there is still frontier that's close quarters the I mainly on unexciting side is kind of just what you have available and I what you should know about all those tools it's usually a a 5 to 6 gigabyte download and they usually contain a couple versions of your Java runtime environments that's what you need and they usually contain a doesn't of C libraries different compilers different compiler versions the usually pegged at all of descent files which for some reason don't know symlinks so you got a lot of files that our . S 0 . is 0 . 1 . is 0 . 1 . 1 things like that and that makes you download 6 gigabytes or something like that get to it's a real modern way of packages offered and In the end kind of every word has its own box you want to use the keyboard or the stack of them and switch between them so it it works it's kind of the winners 3 need to use that the stuff that you want to
use but there's 3 alternatives and there is a very locked rounding um which is free tool uh I think it's been developed in France mainly and so so there is another project it's called users if the guy that Clifford Woolf it's a space Austria somewhere and actually that's a very interesting thing to look at and we'll have a a c a 2nd why that is interesting OK FPGA implementation tool so you
got in that list you got this this fish net of
things and now you want to place the fishnet neatly onto the space that you have a way Bologna FPGA so essentially you need to figure out how to teachers or positive things together uh that's what implementation is essentially again there is standard tools every um FPGA develop a vendor also has an interest in and giving you the tools to make actually that piece of silicon that is say uh so useful to you it so again we have ICT about we have quarters but there is also 1 Major free alternative and we call greater but as the that's the only free alternatives that works nicely and that's uh um project ice storm the when that's a implementation tool for a lattice eyes 40 FPGA that's a very small FPGA so I say for because 40 K uh gated governance and meaning ends in Austin so that so you can get very much on those just but you can actually get some small designs onto those chips and add them we'll see in a 2nd he always Cereno yeah the I stick the all was even less and whenever no if it should be and it's really GD it's very small and I think it's still sold out um Clifford presented that a project at no cost computer Congress of Fig the yeah the kind of this Christmas and uh just shortly after that they were completely sold out of those and M. I think they are rarely will so they're they're enabling again but not that much that Odinga ramping up production again because it's really the only 3 that of a off getting a full toolchain yeah 58 so if you are interested in that and dutifully should have a look at that because of the great starting point and there's some other words there is an amount aboard did you know rather cheap and and and there is also a small exciting sport mitotic 7 as the j so that
kind of as a set on those smaller chips you can get maybe 1 small processor on the eyes you can get very a bit of switching you can't get you know full prosody reside on those effigies so if you want to do a bit more serious interesting things few article with a bit larger boards the nexus for is a very good uh version to go with uh it's actually in education pricing of things 178 it's like 2 or 300 URIs and in the regular pricing but just find some guidance that is still go find somebody but out here emergent order that um takes ones rituals nice small German company that manufacture those boards I think they have also been under under Bitcoin marketing that get that get started getting off dead now the
so the larger your ideas get the larger your aborts need to go the there is exciting somewhere and there was another times were and there's also cite emotion there are some deftly more than a thousand euros there are 2 3 thousand things like that and they grow as much as you want and so it's ever look hobbies
accessibility score is that you need to spend some money and how much that is depends probably on what you have um sold for a couple of 1st maybe thousand euros Otto you find and for others not so but still we're astrology break your come it's it's J. store really cheap so 3 and 5 open source support media are so as 40 merchandise everything else not really so you need to use the vendor tools there are usually free knew well the so you can some of them for free and they worked out to reasonable FPGA size so that you the emitter the size of defecated jaded you can program with them and for
most of the boards we have shown at they work rather nice so it's a great results there but it does stop you can get your things and funds for its it through its definite answer uh you get something really neurons and that's what you want to kind of in the end you want to have you this this couple who weeks of work and then you show your girlfriend or your wife or your mom or whatever can well this a know goes on at the true model but it works in there you're a job there that's what you want in yet and the so some advertisements
and you can do obviously all this can start from scratch and do your own project but there is a couple projects out there in the wild that already do some um interesting things that you can extend so you don't want to start from nothing and someone 1 project that I'm involved in is called often and what we do is um we build a assistantships so I chip design consisting of a couple CPU cost uh memory on their interconnected things like that and um filled it all together and then you can compiler but only if the key and extended to your own ideas so that may goes a bit in the direction that off the story that we have seen in the beginning and where you want to just take an existing the processor lost design and extended to do maybe a special hardware accelerated you think good at solve the problem faster I
think there but the heart to the thing to look at today in terms of hardware accelerators is things that are looking at deeper learning things that look at convolutional neural networks things said that to do image recognition to do and the other big data analysis things that's where a lot of this hardware-accelerated things comes into play and if you want to go into this direction I guess things like that are a good starting point well the going to the area where the real fun is basic design they think it's essentially yeah the chips
that want to have any and the so cover things some are more easy to obtain than others you need to know all of the above that we had before and you need a thing that's called a design kit we get to that need an implementation tool again and you need to money the and the without the design that's actually the thing that is the depending on your connections to industry more or less easily to obtain but usually you can get the so what is this it's a standard cell library design rules electrical parameters things like that essentially everything that's a design or a foundry process MIT is made of of so you know the you can put your processor whatever design to complete the Global Fund resources and C or probably not Intel and and they so you can well but I want to fabricate this chip in days 14 things that technology or I want to have it in a 14 nanomedicines at low-power technology or and there's a couple of versions of those technologies usually depending on if you target like a mobile market where you want to optimize your chip design for low power or where you have a each perceived a style processor we want to optimize it for performance so even though within this technology noted as 49 various and you have a couple of the subcategories that you want to optimize a design for the and all those parameters
essentially the things you need to choose and to get the results in a 14 animator things that no power is what's inside the designed it and a standard cell I was essentially tell you um I won the men's gates and those so you kind of this is the place to put those transistors in this dimensioning right there um design rules how many layers of metal you can have so essentially at the bottom you always have your transistors and then you have a couple of metal layers how why do they need to be things like that all those stuff the electrical parameters all of you get that from a foundering you usually get it for free but you need to sign in and usually not just 1 of those yeah just don't tell anybody about in an economy find India's but if you hit 1 of those DAC um yeah you're allowed to do that and so this on exactly 1 machine you're not allowed to exercise machine remotely from anywhere or I've seen 1 where it says you can access it remotely but just 20 kilometres in a circle around this machine I have no idea where this rule comes from the and and need group that all things again so then is really not nice and M. but definitely not good fit for for most open source and the study of a set of development and and it's usually free but it in the end you need to buy a tube so they don't care about that much and by the amount of
money that you need doesn't stop there as you need that is accomplished will become usually and you choose 1 or the other but that's about where being stopped use either system design compiler a synopsis and combined or used cadence tool and I think it's called encountered tools that now it's they've been renamed their products kind of every 2 weeks or so don't know and I always choose whatever market nice but in the end you know it's a 6 or 10 or 20 gigabytes downloaded from his old old versions of old old code in there so whenever tell you we designed all you the maybe the you it's new and the at by the way all that stuff is usually a scripted in article script so take a holds all the tools that you use together so if you're not up to speed with you tickle knowledge that's where you need to start right now and good because those 2 only adding a couple 10 thousand you should do it depending on what you and the there is a son uh open source options and that's the 1 it's q flow and the other 1 is grows to um the bad thing about those is you need to be ready to take risks because no company whatever will give you any guarantees or even say you that it's a good idea um and to use those tools and I think for Q flow they have done a small uh designed a few taped out FIL Corollas I'm not so sure so unless you got a really huge state of money he got a police dividends those because you don't want to spend too many a couple times I would talk about money
here you need money and let's have a look at some numbers the so doing a
full chip design essentially uh just 1 table so you got your source code you compile it essentially once isn't as as once you do your um way for mass and things like that and then you want to do a full production it's in a couple million euros range that's obviously a lot so you need to share this cost and that's where a technique of most called multi-project wafers comes into play and that is essentially a way of sharing the the chip area that you have with others and you can get this from a couple we use these kind of Wenger's or collective I think that in Europe the most active what is called Europe practice they do this for universities but they also do it for other research um facilities and they also do it for whoever gives them some money so that's actually available for everybody and for example so you want to get you to you want to get 50 chips of 50 or someone in 40 and you pay per square millimeters and and he's nice mm a command and so that's around 50 thousand euros this gave you essentially a piece of silicon that's no black box no packaging no testing whatsoever so you that comes on top of that and um out of those 50 pieces you usually happy if you have kind of 10 or 20 that xt work um depending on your locked in the time of the day and things like that and you can get in this technology for another music elegant um thousand 500 uh gates per square millimeter so if you compare that back to the um if numbers and please don't compared indirectly but kind of a rough 0 order of magnitude comparison it just you can get so much more on 1 square millimeter off-chip then you can get on a reasonably costly FPGA um so yeah if you want to go on all the technologies that you probably would you want to do so 65 mm I think that was with a penchant for error or something like that but typically or maybe and it gets much cheaper and you still can go a bit higher like 129 amusing so that and so this sounds kind of unlikely that any open source project would do that actually there let's and there's 1 more object they they call it IEEE it's essentially a a basic just that if a j that's that's already done um but they allow you to the that 1 customer layer on top of that so you got your transistors you got you're going to connect new wiring layers and you can add 1 um additional a wiry layer on top so you can do a couple of new connections so you get essentially there think of the density of um the although the really sick you got to uh less costs and you got a bit of the problems that you have for the 58 so for some things that might be an idea but also there is we can the design tools and all that it's it's kind draw from really need in all the stages so the following nothing you will go for but the if you everything in this direction have a look at that good yes as a result the results therein I I think for most of us that having just sold Facebook to an investor and it's gonna be a bit hard on my side and it's not really available to open source developers are unless you want to take a lot of risk meaning money and but the fund depends it's kind of if it works in the and it's got the greatest experience you can have because then you can just showing a red LED to your mom but the a black box come on that's what you want the funding pens because there is no real guarantee that things will work out the 1st place that's what the companies do verification and a lot of verification for so I just heard from some guy that 2 a large semiconductor manufacture the does modem chips so you have a 100 % line coverage when
they on their source code before data about anything and achieving 100 % line coverage uh is with pretty much work so that's what they do before data about anything and even then uh when you talk to other guys that exceeded the design when they had the time when they send it out and until they get it back in put it on test and see if the test cases run fountain so that's where everybody standing around this test run just hope it works uh so that that's what fun means in integral and OK so yes definitely if works and as the before it sounds impossible to do this in open source way or to do as a community but there is actually 1 project that I know
of and that is actively doing that so they are an open source community there them actually coming out of the University of Cambridge and M. they are doing a chip design a kind of a microcontroller-based chip design based on a risk 5 processor that's there current a hype in of processor designs that is winnable um as it coming of they use this a processor core at all the stuff around it did you need to get a full my of memory controllers um I old things like that and some other ideas the the end of the plan is to put that on a real piece of silicon by the end of the year the nature looks really good so you can go there right now and get a designed that runs under the GAD runs in 1 of those makes sports and and you can you to that they actually got there and did have repository and as you can get just sort of that and Indians the changes you make will hopefully end up in in a real chips so that's at again so good compromise uh if you don't want to spend the money will yourself that is I think financed by EU money so go for it and OK what is
seen right now it is what you can do to get the involved to get started to do your own chip designs doing it all by yourself is probably
going to be not that much fun so foreign sorry it's protectress just but the cat is really sad the so yeah that's drawing upon let's see what the community does what you can
do to find others that share your joy and looking at waveforms and so the check let's have a look at the
community and what we need to
kind of what what holds a community together what's there you can suffer roles and here we want to have mechanisms rules that others can participate and we want to build on existing work we've seen a couple projects already and we want to exchange ideas um to make sure that everybody profits from what we found out
the and yet how can those to live together action that can live together quite happily and living together is that a set of rules that helps our community and we express those use rules usually a license choice so that's um such you rules that we
have as you there's kind of theory the types of licenses and they're not gonna judge standards to your personal choice it's the way how you want to structure you community or if you want to go in existing ones except the rules are not uh there's usually permissive licensing weak copyleft of strong copyleft licenses and I think this is a topic that's pretty well settled in the development wells everybody has their opinions on that but it's good because it has been discussed a lot In the Honda well that's not that clear what you should choose a what you can choose and what the implications are so let's have a quick look at that so permissive licensing without a patent class in itself the world MIT in these the licenses and this also works great in the hardware also essentially you can do whatever with that Europe source code that that you want that is a set of rules that it well enough understood to be applicable to just about anything um and there is a project that uses to be the lessons it's risk 5 I guess that's 1
of the major parties ran up and there is also if you want to have a close like the of allies like the Apache license you can go with the sold the alliances and actually if you look at licenses for hardware guys and have problems with that as this guy I think he is in a unique position of having understood hardware design and being a lawyer at the same time so um that definitely if you have any questions on how to design a licensing that's the way to go to and and he is based in the UK and really realize guy so we copyleft
um lgpl is usually comes into mind already um was in a public license um essentially you don't have copyleft for your whole project but just for a small part of your project just 1 module or just 1 file so far this could be left there's still which license it's based on the NPL but you specs are also nice guy the then ends yes library copyleft lgpl question is what what a library and softer terms has been the root out over time so it's well defined what is a library in hardware chance the it's not that well defined and that's what Abdullah starts there is no clear definition of a library uh in hardware designs you could make up 1 but there's just 1 right today date and that makes actually using the LTP L a pretty tough choices because essentially you use a license that is not clear anybody what it means I still commonly used open course so that's kind of the the site where you find a lot of existing at the cost and but I would definite recommend stay away from it because and you make it extremely hard for somebody else to use your code if it's just not clear what you meant by using this classes and the again there is strong copyleft there's UPL GPL B 3 users hardware-friendly language as a wrote there uh actually it was drafted along with then so when they open source to SPARC processor at son they um took this experience and put it into the Jacobi 3 and but if it's the use but the implications are not fully understood or not what you wanted to be an if you have a chip design and as a told you before kind of there's no way around in the current world to design achieved without using IP costs without using the some stuff that others created and that's usually not available under GPL and they would do anything but a release it under such license so essentially what you do when using GPL today if you either restrict yourself through the usage of this and poor in a way that creates incentive to sell you commercial versions that's what guys let us with the only on a course as entities that want you to buy a commercial once and other than that you can use to be the the the there I think that could be a space for such a strong copyleft license another world and it's just not there yet so let's see
OK so we had a look at licensing now we want to get as some um yet some existing IP core some libraries somewhat use things like that and to place to go to the days is a site called open course or and if you look at that site you understand what the cat is doing it's actually sleeping on the keyboard so it's in the very dormant state it's not doing much but it also prevents you from doing anything because sleeping on a keyboard that's what is happening to open course today sort exactly backed by a company have lost interest in that but it's still wearing a lot of global Edwards money from that site so that prevent anybody actually from Torino anything about its simply
state it the In a not really functional um state anymore so sometimes you can download things sometimes not sometimes you can register new account sometimes not it's uh sometimes this the ends works which means you can download things with a couple of bytes per 2nd um sometimes it doesn't but still the the a soul we and I'll tell you in a
bid to what we actually meant it is figured out so we need some some community have so we need to replace open course that something that's actually living in the 21st century and is actually working and maybe we have achieved some of those calls so what you want to do is but the people as a way to publish their work to make it should visible to the world and 1 2 um have a way to uh find code that you can reuse and because right now it's essentially reading blog posts off of things that grow we may be turned up to you then it's interesting I've been in this material for a couple years now and it's still a sometimes that somebody comes up well there's this great project that implemented on hold GPU that actually runs an FPGA and that has a couple thousand or 10 thousand lines of code this and never heard of that have it suggests that the sum product is they're just done by 1 guy and uh in his home banking Texas or so and it's never sold well that's a shame it's really a shame because there's a lot of stuff out there and and so what we
did it is a creating new side it's fair following the OpenOffice LibreOffice naming scheme does Collibra costs and what you see here if a screenshot um xt off the version that we will release in October so you won't find this online right now and that essentially replaces open course with something that fulfills the goals have just been talking about and so remember domain name right now you can actually go to the site you find some um things that I talked about licensing thing and that's on the side to already today but the project repository to the heart of the
site will go all of life in October the and I'll tell you why um so what in mind in terms of goals so 1st of all we need the Trevor how aside where some documentation it's really hard to get started in hard
resigned today you usually are able to take courses at university that teach you how things were done roughly 10 or 20 years ago depending on the how will well updated course whatever is that you talk if you never visited any an electrical engineering classes and the morning go to university and I think you will have a really hard time getting started and because there's just not much content out there did you can just read a couple of posts watch a couple YouTube videos and then get started at that's all the stuff that software development is so much easier so that imitation and in use in discussion and there's already 1 thing out there that's a block planets so yeah it's still in the last century so and there's the blocks of their handle with but over the past or is anybody's still using Google + well it's it's develop what half that a head of state the good it's it's funny how long blocks of remained the ear as stable yeah and this project was at best ideas and as that of documentation 1st Condon out there today and his discussion you can find a um aliased it's still around the and so will Google + and as a mouse it'll things like that and the project repository to be released in October so who's behind all that paper I think 1 of the most active um open-source hardware groups that are um behind the the open risk project a project of processor it's a risk-based approach to processor design and and some guys that we've seen before the due respect sir and um all of the grand for example there's a couple of this that are very active in this open course and and form of risk at the community and while the focus today is not so much anymore lately developing process resign the um it's those guys that are can have a lot of experience in the industry as well as in um open source so we that kind
of met each other uh said well can we need to have solved this the starting point was this is open course uh the sleeping state problem and so on we got together we actually 1st of all wanted to solve the major problem that them open course has today because the trademarks for Open Course Library 1 company they don't give it up as less you pay them a couple 10 thousand euros so we wanted to solve this once for all and said we want to have this foundation essentially the so feared legal entity that kind of contains mainly those straight trademarks and to make sure that they don't uh are taken away from the community again we saw this in the UK moved which now all maybe turns out not to be a that smart choice since I uh follow can just transferring the money to the UK is kind of a huge mess and you can get an IDE then if you add a Western Union addition to you banking account welcome to 19 27 no idiot but but yeah that's where we are the then the the forget anything on us let go
and so yeah it's 2 years old now it works right well we've figure out all the legal stuff they're pretty good on that on the site so going into question how design your own chip so have a look at the existing product it's gonna be much easier for you and there is a key people that you can talk to and ask questions and in the end just fire up your editor at the is a very nice to us for um very look editing because I think the standard coding style of every locus just use whatever he makes the um very up again outputs you and that that used most projects families inventing by 3 spaces but with and in and yet FPGA designs you can do that a 6 year if you got the money or if you join or low risk so that we have a look at that so I told you we have a look at the site in October and that's actually what annexed or or cause it was formerly called conference but it's not just getting more so last year we were its source the like 120 people that attended um from all over the world we actually had a lot of people from the UK that are involved in risk 5 so there's a lot of interesting discussions in-depth technical and also on the licensing communities side is what so if you look at the 200 is and that's the conference to go to October 7 tonight in Bologna nicety get some the pizza deer and obviously some spaghetti and no admission please register any ways to make sure that we got enough space for you and or or for more information the that's about all I have to say
and I put some contact details of their fuzzy foundations and that's near the left side and if you have questions I guess we're ready to come at the Moscow thanks hi if