 So I only have like a few slides because the idea here is we're meant to discuss So there are people that haven't been to a bop before stands for birds of a feller Basically just meant to be a discussion session. So I think we have a couple of mics So the idea is you should stop me at any time and ask questions make comments and that sort of thing Let's do a slideshow and then to start off with I do have some links in here So if you go to tinyurl.com ELCE 22 dash bof you'll see these slides So I just spoke about risk 5 but just for people that weren't there Just to kind of level set the conversation So risk 5 is a free and open I say it was started back in 2010 at Berkeley The V stands for the Roman numeral 5 because it's the fifth instruction set to come out of Berkeley And the reason why it's free and open is because the specifications are published under an open source license So a few things that are different about risk 5 because it's not the first open ISA, but it's become quite popular One of the reasons it's a simple clean slate design So based on all the experience they had at Berkeley with creating instruction sets They designed this one kind of putting aside all the past things that they had done It's also modular. So the idea here is that they're extensions that allow you to scale from small things like microcontrollers All the way up to supercomputers potentially But the key thing here is that it's stable So we have a base integer ISA and then we have standard extensions which are frozen now Those won't change and then we had the way we had new things to risk 5 is through optional extensions So back at the end of 2021 we had 15 new specifications with 40 extensions Some of the really interesting things were vector which is scalable vector instructions hypervisor So we can do virtual machines and then some things to accelerate cryptography operations and also bit manipulation So if you want to know more about the instruction set and some of the standard extensions This book's only about a hundred pages and it's available in several languages. It's called the risk 5 reader And then I just quickly wanted to go over risk 5 international So originally it started at Berkeley, but now the standards the specifications are developed by risk 5 international Which is at risk v.org And there's a lot of members including companies universities UK as an individual or a non-profit organization can join free of cost I have a link there to a wiki which has a bunch of resources on it a lot of the Specifications and extensions are developed on mailing lists for the different topic areas And you can view the public archives there at that link and also a lot of these Working groups and special interest groups have regular meetings So there's a meetings calendar that you can check out to see when those happen Now in terms of hardware if you want a board to start developing on risk 5 international has something called the developer boards Programs the idea here is they want to get boards out to open source developers So there is a form you can fill out on the risk v.org website and I will bring up those Links again as for the slides because if you want to pull up the links you can do that I also put the PDF on the sketch page So I just spoke like half an hour ago about Linnings risk Linux on risk 5 So if you're interested in that you can click on that and look at the slides or my talk Which should also be on the conference platform There was a microconference risk 5 at plumbers Back on Tuesday and the live stream of that is online And then there was also several kind of interesting talks to kind of go over some of the different Ecosystem things that happened earlier this year and I host something called open hours every two weeks Is this meant to be kind of a informal virtual meet-up or we talk about like risk 5 software and reside dev boards? So the next one will be on October 12th That one will be not good for Europe that'll be for the Asia time zone But then on the 26th that'll be early evening in Europe So risk 5 itself is not open source processor. It's just a set of open source It's just a set of specifications under an open source license So if it says it's first five it could be open source or it could be proprietary depends on the implementation But for me the thing I'm excited about is that open specifications make open source implementations possible So an open ISA makes it possible to have an open source processor design And there are actually several open source cores that are available so we have Rocket and boom from Berkeley and Paul from ETH Zurich, which is a whole family of cores Western digital created something called swerve, which they were using for like storage controllers Nose are open source as well. And then the open hardware group has been trying to create Verified IP that you could drop into your like companies could drop into their own ASIC designs called the core five And then I don't know if they're here But there's actually someone here from low risk and they've been involved in the open Titan project Which is a silicon root of trust project and that's using an open source core as well Does anyone use any of these cores or cool? Does anyone have any comments or anything they want to say about it? Oh, can we have a microphone over here? Yeah, I'm gonna say both Thanks just five and by go of it bit RB 32 have all been taped out So I'm not putting them under FPGA stop. Ah, okay. Yes. That's a good point Yeah, I should say FPGA friendly. Yeah, so that is a good point that they are both in Actually, I'm gonna I think of the next slide is about the open-source silicon stuff that's going on cool. Oh Yeah, so the one I didn't include on there because I had a separate slide for it was Alibaba has a chip design company called P head. So kind of similar to sci-fi, but it's part of Kind of like sci-fi where they designed cores and licensed them But the neat thing is they released for there was five cores is open source and one of them is actually in a So see from all winter called the D1 So if you want to look at the RTL, you can actually go to their github and pull it up. Yeah Well, the question was if the SOC was open source So in this situation Alibaba T had designed the core So that's the only thing that they open source and then the SOC is by all winter And I don't think all winter releases any of that. So Unfortunately, usually like a lot of the IP is licensed from other companies. So It's usually not. I was also gonna say be wary because it's all generated RTL Okay, it's generated very long and I don't think There's been discussions, but I don't think yet that they've put the Source and the tool out. Okay So like your ability to use this to tape out your own core Probably no not probably not to use all right not not great yet for a collaborative project Yeah, unfortunately, but they they all they are aware of the issue Okay The reason I thought this was kind of neat because you can get aboard with all winter D1 And then you can look at the RTL for the core and it might not match up like exactly right like what they taped That might be slightly different, but I saw some people looking at like the MMU RTL like the verilog or whatever it is for the MMU and they were working on the linear code I thought that was kind of neat because I don't know if too many situations where you you're working on the driver and you can Go look at the RTL for that part of the core. That's gonna be the game changer, right? Yeah. Yeah, I Mean well probably many of us have done that but it was like we were working for the company, right? So in this case, it's the public that's able to do that There's a high performance core That's being worked on at the Chinese Academy of Sciences They're trying to tape that out as well. So There'll probably be an update at the risk 5 summit, which is in December about this project But that was another one I've been looking at But yeah, not not like a commercially available thing yet Okay Yeah, so this was like the one that hopefully we can talk a lot about because there's a lot of interesting things happening here So one of the organizations that I follow a lot is called fuzzy, which is the free and open-source silicon foundation They have a monthly newsletter Which I like I always like to check out because it has all the latest things that are going on with like open-source FPGA tool chains And open-source silicon projects And then I think probably the most important thing that's happened in the last couple years With with like open hardware is that Google has teamed up with e-fabulous to create this Open-source or they're working with this Former Cypress fab in Minnesota called Skywater and they've been able to open-source their 130 nanometer Process design kit so that or PDK. So this is that these are the files that say like the how the transistor cells are designed and all the Design rule checks and all these sorts of things all the all the proprietary information that you need to actually Provide to like produce the mass that's gonna be used for lithography at the fab So this is like the lowest level of information that you need So this is now open-source for their 130 nanometer process And they just announced like I think last month the 90 that they're gonna do that for 90 nanometers as well So no, this is not like the latest greatest stuff, but it's actually Quite useful for microcontroller class devices And yeah Okay, yeah So the the comment was that the Pentium 4 was kind of this era So I mean there was a lot of things that I did on my Pentium 4. So, you know It's not, you know, it's not like what we have nowadays, but there's a lot of interesting things that you can do with it, right? and The to the the other really important thing here is that they have a free-of-cost Multi-project wafer so they combine a bunch of different designs on to a wafer and they're doing that for free The only requirement is that it's open-source. So I think what's the quarter they do one of these multi-project wafers So I think it's 40 projects per wafer. So you just you do your project and you send a poll request And you know, they there's actually not that many people doing it So if you are interested and you learn the tools, you actually have a pretty good chance of getting your tips made And the other big thing here was Like not that many people probably know how to do this So there's this interesting course called zero to ASIC Run by a person named Vathy van that takes you through how to use all these different open-source tools that you would need People to design your own chip kind of the use case here is you know for people that are maybe not already chip design engineers Maybe you're a software engineer and you're like it'd be super cool to make your own chip. Yeah, go ahead He's also just done the tiny tape out project, okay Which gives you I mean, it's only like you can do a tiny like 300 400 gate design And he's putting I can't remember loads of designs on one. Yeah, okay So if you don't need a really nice light weight and there's a nice kind of github interface there so you can You know, you can do a design using kind of a graphical Logic tool, but you can also put in your own Cool, so that's called tiny tape. Oh tiny tape out Calm that half. Okay. Yeah, and the other thing to mention is of course that there's also the global foundries have now gotten the game So there's now the 180 group global foundries Which is particularly exciting because the round trip time is gonna be a lot shorter than sky water so We did a tape out on sky water In the NPW 5 We're still waiting for it a year later Oh, so you you were you did one of the projects then, okay? Yeah, we didn't so so I'm chip flow dot IO Oh, cool. And Yes, we did our tests. We test our tool chain Doing an NP a kind of a microcontroller class on NPW 5 and a Linux capable So Oh, really? Yeah, I mean like only just about capable. Yeah. No, I know in simulation, right So yeah, what core was did you use on it? So we used extras five, okay, and then the sock was built with Amaranth. Oh cool. So Amaranth You want to by the way, since you're from chip flow. Do you want to talk a little bit more about that and yeah? People are interested. Yeah. Yeah. So basically what we're trying to do is with chip flow These like I can think the term that was we're coming to is fabulous semiconductor as a service So the idea is is like we have You so we're using Amaranth as the front end. So you've got a nice Python environment good debugging good simulation speed That you can use to do your design and do you and we got Strictures coming in for how you do your verification for that to get coverage and then you can do a test on the platform Then you can do a test tape test chip tape out and we'll be supporting proprietary processes as well as the open processes And and then be able to go to manufacturing full manufacturing with manufacturing test Nice of it. So I'm very excited I the previous talk was about Linux on risk 5 and like my deal thing would be able to run Linux on one of these open source chips Yeah, so I I saw that they replaced the Pico RV 32 on the MPW with risk 5 And I've used vectors 5 on FPGA to run Linux. So that's that's quite exciting. Yeah, that's just quite capable really. It's probably the same sort of capability as a You know an R1 or something from arm It's got pretty good throughput for the size and gates But yeah, yeah, I mean there's all sorts of stuff. There's so many things happening. So there's now an open power Design being built in Amaranth as well, which is gonna be fun Did you want to um, so I didn't really get into all the Python and all those things Did you want to mention? Yeah, I can give a little bit of a bit of history So there's like there's now been about 10 years worth of work of how to do hard word design in Python So it starts off with a my gen project about a decade ago and then Probably about five years ago now, I think That was forked off into n my gen which is kind of Not quite compatible. There's different. You can use my gen code, but you'll get different RTL outputs at the end of it But gave a lot more power and capability and then recently there was a bit of a trademark fight So n my gen became renamed Amaranth And and that's where we're so a chip flow we're gonna be heavily sponsoring Amaranth and bring it on so it can be a full toolkit for for both FPGA and Silicon great and like we're also doing a project and You know this will be marketing around this at some point But we're also working with a company called pragmatic who do plastic ICs and we'll do a very cheap turnaround plastic Plastic look like a polymer like yeah, yeah, it's about it's about a point seven micron process So it's like not you're gonna do anything mind-bending afraid. You know do the world, but it's a great way to actually make something Interesting and do it cheaply. What are the applications for that? Oh gosh Well, they're the market areas that the only if you look at what they're talking about publicly at the moment A lot of it's on the packaging like smart RFID, you know do things like Optimize what that you know the best before dates on a package, you know a little bit of a Any display there and you can measure like did this was this package within the right tolerances of Humanity and temperature of all of its supply chain. Okay, then we can do there We have a longer shelf life for that. Oh, wow. That's that's one of the areas I think health care is also gonna be an interesting area as well for that Yeah, but yeah, really interesting stuff. Wow. And so I didn't have it on here, but you also mentioned with the Google Multi-project wafers that no, there's not one global foundries involved Okay, so Google's now got a deal with global foundries. Okay, they have opened the 180 PDK for global foundries I was pretty excited that global foundries is involved because I mean I never heard of skywater before they're apparently an old Cypress fab, but like global foundries like they do IBM stuff, right? Yeah, that's that's pretty cool They were right there likes the old I this old AMD fab and the old IBM fab, right, right? Let's look at the other room and and financially engineered off and Yeah, so, you know, that's that's gonna be a big one and global foundries is kind of unique amongst the the faps The founders services companies because they're really focusing on older technologies And that's where we'll get the open source first, right? It's on the older technologies. It's gonna be a while before we can Seriously tape out Fin fat designs and open source, but I think you know, I mean we're aiming to try and get 65 nanometers Thank you and get some point in There's a lot. Yeah, I don't want to commit to yeah I Didn't list all the people involved on here but one of the people at Google that's been behind a lot of the hardware toolchain stuff his name's Tim and Tim Ansel and He knows I wanted to read Linux on a like a Libre silicon chip And he always told me like, you know, I want that to happen. I need to promote this So if we get people using 130 they get them using 90 and then Google sees Google sees that it's like a useful thing Then it'll it'll they'll pull to be able to like convince fabs to do more processes Absolutely, I think I think I love this space because like, I mean, I'm an old-hand open source guy, right? You know, I was building stuff commercially late 90s My first business over source was collaborate in 2005. Okay, so I love this because it feels like the old days for me Yeah, right, you know, it's like that you can you can know everyone in the community And there's a few heroes like really pulling everything along you're like your Alan Cox's and your linuses of the old So yeah, no, I love this space so much fun I think it's gonna be I think it's gonna be a massive green game changer Oh for how we do this like and how we do embedded systems and everything in the future. Yeah Yeah like the number of times when if any of us are on the device driver side like feel no the pain of When hardware design doesn't understand software and this gives us a chance to solve that. Yeah Like what like One of the things I thought was quite interesting in like in the FPGA world Like I've run Linux on system on chips that are in FPGA's that are using open source projects Like like light x in like in this instance, you control both the hardware peripheral and Linux So it's like one of these things they there was a conversations about upstreaming some of these drivers And it was kind of interesting. It's like do we change the peripheral to better fit with arting in Linux? Absolutely. Yeah, it was that it was an unusual conversation because normally normally that's not something you get to decide, right? Yeah quite interesting Did anyone else participate in the MPW or I know I've not had time myself But yeah, there's like a good number of projects that have been taped out on this process Yeah, which process are available on the Google MPW or what you mentioned before that you mean like the Technology note. So in this case, it's 130 nanometer and then they just announced 90 nanometer and then you were saying they also announced 180 with global foundries. Yeah, which is Right, right So yeah, I'm excited about the global foundries thing because that's they probably have a lot of technology They could potentially get involved, right Did you want to say anything about low-risk? I mentioned low-risk earlier and they were one of like kind of the first Organizations a few years ago to start talking about doing open source chips Yeah, so I mean I just started a few months ago But yeah, our main focus at the moment is open Titan, which will hopefully get us kind of open source SSE Going in this case for security purposes and and in security. It's a really cool area for the openness, right? Yeah And cryptography we've known for a long time that you kind of don't want security to go up security Yeah, so having an open source route of trust is just a really cool thing. Yeah, and that's what we're trying to get Yeah, and I didn't really go over the relationship here But the Ibex course interesting because it originally started at ETH Zurich So each time they Berkeley has been really involved with risk 5 but another group that has is ETH Zurich has the pulp team there And they created a family of course. I think originally it was called Pulpy right, you know, do you know which one became Ibex? I think it was Pulpino Pulpino, okay So yeah, I mean the nice thing about having taken that from ETH Zurich and putting it into a low risk Is that now we can provide the support and making sure that it's actually tape-outable and stuff And Ibex has been used in many different not just open tide many different projects for tape outs and stuff like that One of the things I remember from like the low-risk blog was talking about how to do like Kind of things like continuous integration and making things testable and in verification and those sorts of things So it looked like you were really kind of polishing like the initial design that came from ETH Zurich Yeah, and the main challenge in open tide and at the moment is just doing the whole digital verification design verification type stuff, right? So, yeah, I think I think low-risk actually can do like hopefully what we can do is also Provide some knowledge transfer about how we've done that DV and and make sure that it's transferable to other open SOC Right, right project Great. Does anyone have any questions, comments? We don't have the speakers you can throw, you know, we don't have the throw microphones, but we can hand them back I do. Could you give some kind of overview about the kind of roll out expected for risk five, you know kind of commercially and How it's going to go from maybe microcontrollers to application processes to whatever. I mean, what what do you see the future? Say yes, hence for risk five in the field I think right now for for risk five is most applicable to kind of embedded systems So I think the the advantage one of the advantages of risk five is because it's an open architecture Companies that already have their own design teams can implement their own architecture Anyway, they want like the way that you implement the ISA is completely up to you So that gives a lot of advantage to be able to like design things that are like really Optimized for your use case So there's I've seen companies doing things with like very low power designs like things like energy harvesting and Doing like novel things in terms of design Of like the things around it so so really like the the circuitry that's doing like that's like executing the instructions May not be the most interesting part of like a risk five chip, right? So the this whole thing started with a group at Berkeley It was doing design doing research into accelerators So what they were really interested it was like in accelerators like vector accelerators And they just needed like an ISA to start working on and they looked at some other options and they decided it would be Simplest or best in there for their case just to come up their own base ISA So the idea here is like risk five It's kind of simple. It's not very fancy in the whole idea here It's a great base to build things on top of so like whatever your area of specialization is you can build on top of that So I've seen some interesting things at the low end like with energy harvesting wearables and some things like that Yeah, so someone's making like an application like making a ASIC that's very optimized for like a particular use case like Having the risk five core in there for like the general Competition makes a lot of sense and then it's probably integrating with some other things on the die that do some more specialized Acceleration or something like that And in terms of the higher end there are three about three companies that are Startups in the risk five space that are targeting the data center So I think there's already pretty amount of stuff for the low end of risk five and in terms of like higher performance There's Ventana Esperanto and Reavos. They're kind of doing higher performance systems It's not a whole lot of information about what they're doing yet But the idea is like that they're trying to make like really high performance devices probably focused on accelerators for like the data center space So I'm quite interested to see what will happen there for like the high end The other thing that's quite interesting is Intel has gotten involved with risk five Specifically Intel foundry services. So Intel now offers their fabs as like you can go there and get your chip made So for them with their foundry services, I think they want things to show off And you know probably doing arm so C's is maybe not the best idea for that So I think risk five was really attractive for them for that So they've teamed up with one of the risk five IP providers one of the companies that does cores named sci-fi And they're doing an SOC on one of Intel's like advanced processes. I think seven nanometer And that's called horse Creek or at least that's the code name So I'll be really interested to see what comes out of horse Creek because right now We don't really have any high performance risk five hardware available commercially like there are like academic research groups and stuff that do higher performance stuff, but I'm really excited to see what will happen there The other thing that I think is possible is for like mid-range devices or like smart IoT devices And that's what all winner I think is targeting with their chip And I think there will be other Licenses of Ali Baba T head that will come out with like kind of mid-range processors One of the things I didn't mention is that Ali Baba T had ported Android to their system So I would expect to see like licensees like all winner come out with risk five based SOCs meant to run like low-end Android devices Part of there is probably like to save on cost, you know, they don't have to deal with licensing For the ISA right they probably most of these companies are still going to license the core I mean like he had but I think that'll happen at the mid-range Yeah, I guess Linus has made kind of a point recently around the arm side of things and you know Releasing Linux from an M1 Mac. Yeah, you know and kind of dog-fooding that Like there's arms been around for decades But there really is with the a shy Linux project there's kind of a momentum now that there's arm available You know consumer, you know performance desktop laptop package. Yeah, will it take that with risk five? Do you think that would and what's kind of the timeline there? You know like to break it out that embedded niche, you know, yeah, I guess this is for Linux desktop perspective And yeah, you're interested in the Libre stuff. I think Libre Linux people We really need that like the thing that's kind of gating so in terms of having like being able to like Compile Linux and release it on a risk five system The thing is like we don't really have any good SOCs for doing that yet Like the all-winner one is like long gigahertz single core There's a couple companies that are making multi-core ones, but they've not gone into full production yet But I think they will there's a couple like maybe like one like four core 1.5 gigahertz systems and But none of those are in full mass production yet but I think going into next year hopefully next year we'll start to see kind of like mid-range single-board computers like, you know a couple core like that They're not going to be very high performance But like they you could start running like a basic Linux desktop on it that sort of thing in fact the the Alibaba That he had they have a design called the C910, which is a higher performance one is a it's out of order core They say equivalent to maybe like an a76 And I think when we see companies making like SOCs or something like that So right now everything dealer like kind of test runs or like low quantity like e-vail runs, so Hopefully next year we'll start to see like full mass production for some of these things The other thing I think that would be interesting is what happens with the Intel Horse Creek So they announced this is a RISC-5 SOC with a sci-fi core That's going to be like they're seven minutes. I think seven nanometer process So it really comes down to like whether or not they made affordable dev boards Which like my hope is that they come out some dev boards and they maybe subsidize them because it's like trying to promote their foundry services So that would be really good for the open-source software developers to have like a much higher performance board. Yeah Also at the Linux Plumbers conference in the RISC-5 track one of these presenters had this RISC-5 laptop with him with the broken keyboard. He did? Yeah, I don't know What's his name was but like he had this RISC-5 based laptop with him. Yeah Well, I didn't I didn't put it in there But there is there is a company called Excelbyte that has announced a RISC-5 based laptop But there is there is one that's been announced they've just announced that it's going to happen They haven't really said when or how much but I I Think I would guess that it's probably gonna have like that Alibaba C910 which they claim is like an a76 performance But like I don't they didn't say how they claim that so I don't know I could say it seems like Maybe by this time next year we'll have like a mid-range RISC-5 laptop It'll probably be painful to use by like our normal standards But I think it's something that you could start doing like Actually, maybe use like kind of I'm thinking like arm net book ten years ago. You know, maybe that sort of thing Yeah, I was gonna say we've talked to Ben Dukes as well if you're doing that because I think have a good own build the like a whole good own building system now does our RISC-5 as well as It said he just said we've had a few breakages we tried to demo it and it failed so In terms of the So one of the reasons why none of the RISC-5 Linux distros are official right now is they don't really have Hardware to build them on other than cross-compiling. So like Fedora, Debian, Ubuntu like all have support, but it's not official yet Fedora tells me they won't make it official and then till they have hardware that they can do native Compiling on because that's one of the rules for Fedora. So I think we really need one of these startups like Ventana or Reavos to hopefully come out like really high-performance hardware that people could buy put in Iraq You know and for that Maybe one or two years. I'm hoping Yeah, go ahead. So yeah, no, I'm just going to Bob's comment We are building free desktop SDK and Nomos We have definitely had build it builds working It's just of course as soon as we want to bring our own match board over here Of course the build stops and it doesn't work It's like who where did we leave the last working? Because yes, the big problem for us for doing this sort of thing is the availability of hardware and the unmatched boards are just not fast enough and Whoever designed them needs to learn how to make non-noisy fans because And So for people that aren't familiar one of one of the board is people have been using a lot is the sci-fi of unmatched which they've unfortunately discontinued So Yeah, they would kind of will be waiting to see what they come out with next Cross-compiling guy here the yachter guy. So we don't have the hardware to build on problem, but we have a problem that Well, we need somebody to resource us to do cross builds on our Build cluster and the build cluster is currently fully booked with legacy architectures, you know x86 arm meeps and power pc all of them legacy So to add another one we simply need somebody like sci-fi to get us more server hardware That is arm or x86 doesn't matter it will cross compile it will run the test matrix in Qmo And then there will be official yachter Risk 5 not until then So a different problem, but so if anyone works for a company that cares that cares about risk 5 I would like to see support in yachto. You need to Supply some data center capacity, right? No, well the right thing to do would be to join the project as a member company and Come to the member company meeting and say we want this and we willing to resources and let's get this done Yeah, that's how it usually works like we have arm as a member Intel as a member But not any of the risk 5 until then it's a community effort which may or may not work. Yeah Intel is kind of interesting though because they are heavily involved in risk 5 now from the aspect of their foundry services business, so maybe maybe Well, maybe I could ask Thomas. I mean Thomas Gleisner to ask Intel about this well, so There will be the risk 5 summit in December So I will I'll keep this in my mind when I'm talking to people there because it'll be all the vendors there, right? Need to I'll keep on harassing them about joining the yachto project. Yeah. Yeah Yeah Yeah Definitely having an unmatched board you should probably be talking to sci-fi a lot because their default distribution For that board that they shit with it is a yachto image Yeah, and it's like I Agree with you so vendors that are using yachto for their SDK should join the yachto project and provide resources I think everyone in this room is it's a lot with that sentiment. Yeah We just need to find people with the budget Yeah Yes, yeah, so he was mentioning that there's the meta risk 5 layer So actually all the risk 5 we're stuck stuff for talking about all the risk 5 boards I think all of them are supported with Unofficially with yachto with meta dash risk 5 but it's kind of the heroic efforts as you said of of a Raj and maybe a couple of people. Yeah. Yeah Yeah Maybe it's worth giving me a card at some point or trying to talk about this Nice Of your card You were saying yeah, you could do exchange contact information. Yeah. Yeah. Yeah making connections Cool, I Think we only have a few minutes left anyone have any last questions comments. Oh Yeah If someone starts building one of these open chips, yeah, what happens with all the peripherals and components on top? This is a really good question, right? So like yeah, we're talking about cores but you need a lot more than a core to do it, right? So there are groups as part of this Google effort like to do things But we're very very simple beginnings, right? So like there's a group at UC Santa Barbara this doing open RAM as an open SRAM because we don't actually have a way With open source designs to create SRAM So they they're now able to create I think a couple of kilobytes of SRAM in that pi 5 that Michael willing's been doing in a couple of people Like they're using that open RAM from UC Santa Barbara to you to make their ramp But yeah, we're talking about like just having SRAM, right? So Yeah Early days. Yeah, exactly So, yeah, we're going to be them There is there's a start that's a few things starting to come together in the amaranth world for peripherals There's a bit of basis one of the things we're doing first is Being able to extract which buses the peripherals So we can have track buses we can go like what this do wishbone or excite But Yeah, there's a few peripherals there at the moment There's some there's some USB stuff like that you could do you a pretty simple Like, you know USB device with Dry, you know Video driver wise there's some there's bits and pieces There's I mean, there's there's there's really Yeah, it's like simple frame buffer stuff or driving HDMI all these sort of things but in terms of like like a capable SOC level Their level IP there's a few a few things still to get done, right? Yeah, I mean we we hack together It's you know, I've said that these these are the projects. I really want to get yeah, I will promote people into right? Yeah, so it looks like a basic hype around control basic SPI to SPI controllers and and you know, there will be a You know a section under the I'm around the project to build up these Capabilities in around there's also a few things are going on, but there's probably doesn't really one place Yet to go for all this stuff. No, I think that's a really interesting stage and which is actually talking with the step on Oh, yeah, I'll sell the with the final moment here I'll sell the Fosse foundation anyone trying to collect all these efforts together is the Fosse foundation. So Please check out their website. But yeah, we got to basically build everything from scratch So it's interesting times. It's a great time to get involved with open source Silicon open source chip. Yeah, this is like being involved in Linux version 1.0. Yeah In all the good ways and bad ways that is All right. Thank you everyone