 Hey folks welcome bookshelf mounting a grid of Linux boxes. That's the thumbnail for this update You know I'm trying to reach out to new folks and to do that to be more accessible by being as Specific as I can about what the content is in each particular video. So that's what this one's mostly gonna be about I mean, you know the t2 tile project is about Reinventing computing and it has you know consequences from hardware to software to artificial life to humans in society But that's a little bit of a longer story. So I'm gonna try to stick to this. We're gonna have a video, you know, I'm mostly a software person not a Electrical engineer not a mechanical engineer not a furniture designer interior decorator or any of these things I'm a total amateur at all of that. So be prepared So we're gonna start by watching a little 90-second demo of the kind of things that we could do with this grid of Linux boxes And then we'll circle back around to introduce the main video And as always we'll start with the computing team creed First be robust then as correct as possible and as a Okay, this is the demo a Eugene time-lapse Standard laws of physics that come with the tiles at the moment that weird shape was the two lotuses 38 t2 tiles each one running Linux It's the same that you can see in this thumbnail for the video. So again, let me just warn you This what you're gonna see here, maybe isn't pretty, you know trigger warnings for incompetence, but you know, I Do software I do this other stuff because that's what it seems to be necessary to advance the T2 tile project, which is about actually building a real thing not not just in simulation But to explore the issues from physics to Philosophy so let's take a look at the the video Gonna have to raise this whole thing up basically from here To the wall is six feet and that would be enough if we could continue The support pipe as long as we could go up some extensions that would Screw into the bookshelf One small one and one big one. Oh great What do you say, huh? Do by 25 This piece away too bad. It's good at seven and three-eighths Exactly how we're gonna cut that. I Don't know. We got to figure out how to make this Diagular cut use the Makita instead of the for a penny exceptionally terrible. Oh Man, I want to fire up the jointer even this out all right, well, I Cleaned it up a Little bit For not showing you me cleaning up when the jointer was I'm actually short on Electrical outlets and I've got the Fuji on the battery replacement thing because that's its main lot in life At least I thought far enough ahead to make these non critical. These are the non critical measurements ranked down to 1100 rpm Assuming this plywood counts as soft wood It's not deep enough made a super mess of it. Now. We've got Piece of wood clean on the back Clean enough from the front at the two shelves the two risers I'm calling them more or less lined up by the thing that's going to be against the wall and the most important thing is that the holes line up And I have no idea how I did that, but I am skilled in the art of screwing up measurements Said he was a sigh. I'm a very bad whittler I whittled the ship till it's small as a boat now. I will hold it how it float So he threw it away and cut a throat When he saw his head was gone. He whittled another and put that on John shorty just for the record There's the joint there. It's going to go a little bit deep So we messed up we always mess, but I don't think it's fatal at least not yet Extra crappy looking stuff is going to be against the wall on the left side So that we can Lift the pipe out in position For the rod as close to the bookcase for the shells will support when we push on it This is the maintenance position when we want to get in behind. All right, but why don't we just Buzz off Big ranch and cut this in place You think I've got to be able to just wail on it hard enough it could go at it with With the saber saw and I'm gonna unplug the camera. It's quite a bit slower going Then I thought with this blade, but it did get through and then complete this mission by raising the bar It doesn't seem Lens Fixed focal length so we'd have to get lucky so we might be able to Get the whole grid in this shot We'll see Laura has been raised is the picture of what it looks like, you know from further back We see the the Fuji X-T4 There as well, you know, there's a ton of problems with this It doesn't actually lay that well against the bookshelves part of that I think is due to the bookshelves themselves Not exactly being lined up with each other these things could be tweaked in various ways in addition You know gonna have to pull at least a few more shelves worth of books out to clear Get some clearance behind there. So part of the idea of this is that, you know, we can get Get like, you know, some big slow fans that we would stick on the edges of the bookcases behind it to suck air out So it would pull air in through the face of the grid and then send it out the sides and that might be a cooling solution Which I am increasingly concerned about I see what 38 Lidding boxes actually feels like kind of toasty. So step by step That is the development process that we have for this update Last time it was, you know, here's where we were at. We're still at the same number of tiles All the stuff is still the same except the next step was rethink locations I mean, it was looking like maybe we were gonna need to go to a whole, you know It'll rent an office space or something like that. But now we raised the bar I think the camera will be able to get a half decent shot on it That's good Coming forward, I still want to actually build Lotus 3 and I have some kind of confidence that it might actually be able to stay there And not have the whole thing get disassembled again. All right, that's the development story. So very quickly The 2022 project over T2 the T2 tile project is to redo the event engine In a much more robust fashion and part of that idea was to redo the mechanism But whereby adjacent tiles compete to see who can have the next event and to instead of Doing a long complicated protocol with packets being sent back and forth and resolving races between them That maybe we could use this ring oscillator idea that I've talked about in the last couple of Updates now I want to show something really quick here Okay, this is the time. This is what it's about. It's just two weeks. So Let's take a look at it and see I mean there's not much to see it's because it's all still very low-level stuff, but you'll see Okay, so here we've got three tiles that are Are running? We've seen this sort of thing before but in this particular case these tiles are not actually connected at the moment We have this one has a power line jet power injector of his own These two over here are sharing a power injector even though in fact, you know This whole thing is really just coming from a from a Y splitter over here But the point is that allows these things to be connected or And so now let's let's connect This is the key master the white one Let's hook it up to its Northwest Okay, and we start seeing update G-drow in e2 1 and e2 0 and e2 1 What that is is they The key master and it's Northwest Northeast, which is the Southwest to the back Have started to do a ring oscillator between them And so they're passing a zero around and then when it gets back It's getting flipped to a ones that's going around and so forth now It's going incredibly slow because all of this stuff is slowed down for debugging purposes The important point for me is that it's it's very robust You know, so if we pull the thing out, you know So now we pulled it out There's a bunch of other messages that come and go through there as well because still the higher the old event engine is still running at The moment as I'm working on this stuff But the G-drow ones are the ones that were things Overheating it does a lot of that with all the lights on but let's plug it in again and see it pick right back up All right and boom right there It's it's right back to passing the shared clock around But now the cool thing is the thing that's really is new for me is let's hook up another one There we go So now we get east But east 2 and east 3 if you see that So what's going on is each of these pair-wise tile pairs is has a shared ring oscillator going just between them That's why it's two because there's two tiles involved But now in addition there is a three-way shared clock going around this whole corner going You know from from the to the key master to the guy to his northeast the guy to his southeast To the guy who who's his west and back around and that's the east three One that's also happening simultaneously and and again, you know, these are all very robust We can pull them out we can plug them back in it's happening at a lower level Then the whole packet exchange that the stuff is built on and and so all right So that's that's really mostly what I wanted to see You know we could start up a start up a huge gene here as well cuz wine But it's not our point for the moment all right, so there's the demo and You know don't forget about this this this is the circuit Schematic diagram for the circuit board that is on the back of each t2 tile that that ties the whole thing together But what I want to call your attention to is this is one of the intertile connectors the ITC's and you know, there are 16 pins, you know, so VDD that's power. There's three power GND. That's ground. There's three grounds that's six pins we've got a bunch more and You know, so what are they? Well the The are the RX and TX RX and TX ready those four pins are the things that are actually Sending the packet signals back and forth that the whole thing is built on but there's also these Lock pins that if you dig way back in the t2 trial project archives You can find discussions of what these were being used for but I never really got it to work and I never really got it To work reliably and I never really got it to work in a way that would play along with the packet transfers Which have all of these delays due to buffering so I have repurposed these pins To do the generalized distributed ring oscillator, so the the Eiger lock and the I rec lock, you know, they are actually doing the Four stroke cycling while in four stroke in the case of going across the pair And then four stroke all the way around the ring Using the other ones like that. So there's a chance and again I don't know that any of this is really going to end up working out in the end But it is in the spirit of you know be you know UDP ish don't assume that there's perfect transit Down below but you have build up a ring now One of the problems with the ring oscillators in fact is that it doesn't guarantee that there's only one Edge between a one and a zero going around if you have a ring oscillator with lots and lots of stages You could have an island of one that's so cycling around with an island of zeros behind it And then another island one another island of zeros and they would just keep going around and around as two separate things And if you were counting on the edges between zero and one and one and zero as being a mark of uniqueness Something in the grid that wouldn't be true. And so that's a real problem in a distributed systems in general We're going to avoid that we're going to mitigate it anyway because in each step We're deliberately randomizing the slightly randomizing the amount of delay time through the stage How long you get to hold the turd the the token when it's your turn and in that case I think you could do some theory probably pretty easy to show that the islands will collapse into each other rather quickly And once we've gotten to a single Single ring a single island of what all ones to all zeros to all ones It'll stay that way unless there's further, you know errors and disruption in the system in which case You know that's what self-stabilizing systems is for so that's the current status of that I just wanted to You know get it marked down. Okay, and finally, you know outreach So in the last episode BJ Manon Submitted the science fiction story to as a mobs science fiction magazine. It's under review You know my my feeling at this point, you know, we're waiting at least probably another three weeks And maybe even a couple of months beyond that, you know My feeling is most likely they're gonna pass and the reason that they're probably gonna pass is it's not accessible enough, you know, this is like Back to the Thumbnails and making everything more accessible. It's a problem. I have So we'll see I mean the story From the first few readers that I got feedback from the story has gotten tons more accessible A lot easier to understand what's going on But still it's packed pretty dense with computer stuff and all kinds of crazy things So I'm thinking they'll probably pass We'll find out. All right, and the living computation foundation for folks that don't know They is backing the entire t2 project the t2 tile project and A bunch of other stuff that's in the broader mission of living computation. So, uh, the second half 2021 second half summary is is now out. Uh, uh, you know, there it is under the the news item and you know I get depressed. I get down sometimes that I feel like I'm not making progress enough progress fast enough, you know But you know, you look at it over six months, you know, a ton of stuff that actually got done, you know, so That's good. Um, and We've got some new living computation foundation nerds if you don't know for for a five dollar donation to the foundation Which is a nonprofit You can get your own living computation foundation nerd number Um, and so we've we've we have a bunch of them including a Core of you know, folks who are actually recurring donors, which I didn't expect at all and it's like I don't thank you guys enough But well, I don't thank y'all enough. So thank you Uh, uh, it really it makes a huge difference. And we've got a couple of new ones Uh, uh, one that I've owed a thank you mail for a while and another one more recently in the most recent one It was actually pretty substantial contribution like three digits Uh, uh, specifically calling out computing up the computing up podcast, which I super appreciated Uh, uh, so there that's the computing up homepage. In fact, we have another we're due out to have another one Uh, uh, well yesterday, but it's probably gonna come out today or tomorrow depending on when I get some sleep So, uh, uh, that is it. Um My goals for next time is to let let let's let's welcome our new nerds do it or die Uh, and then to try to try to take the generalized distributed ring oscillator the ringo idea that I just took a brief little look at Uh, uh, and take it up to, uh To user space so that we can at least see it working. I'd like to build Lotus 3 Uh, uh on the grid and once again always remind myself have some fun You do too. Uh, uh, thanks so much for coming. I hope to see you next time