 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2 tile project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. I'm Dave Ackley. This is the 16th T Tuesday Update. Let's get into it. There's four more weeks to the deadline to have all our commitments done and order out to manufacture at least 133 of the T2 tiles. Little computers that work together to make a big grid, a computational resource of a new kind. We're going to make at least 133 of them. That's what we're calling a ring lotus of T2 tiles. Hopefully we'll make some more than that. We'll see how it goes. This past week, I feel like I didn't get a whole lot done, but really there's a fair bit to talk about. So I'm going to get into it. I think basically I have sort of moved into a higher gear since starting these T Tuesday updates. So thank you folks for being out there. Thank you for asking questions and making comments and just letting me know you're there. It's helping make this happen. This week, today, I want to talk a little bit about hardware stuff. We've been talking a lot about hardware stuff, about software stuff. We've been talking a ton about hardware stuff. The goals for next week is to really move hard on to try to get as many of the ports that are necessary to manufacture the tiles in the pipeline to get here. And again, I was supposed to get the 3D stuff going again because I'm going to be looking for help if people have 3D printing power out there to build the cases and the various little gizmos that go to connect these things together. And we need to get that settled sooner rather than later. Okay, so where are we here? Oh yeah. So last week when we ended up, we had gone through a terrible, terrible wrenching trial about whether the underlying analog properties of the board were going to be good enough to use. I think we ended up thinking that they probably were okay, and I was probably just a little bit messed up in how I was interpreting the signals I was getting from the oscilloscope. Another reason why I hate using the oscilloscope is I don't get it. But I had hand-built a little version of an intertile connector that would have resistors on the data lines to help smooth out the signals. I gave a metaphor last week about shouting into a canyon and getting echoes back. Since then I have found a book called Write the First Time. Apparently you can get it on Amazon like $600 for it. It looks like it's kind of one of those things that people just use for money laundering or something. I don't know, but the book is available on the web all over the place and there are some places where you can fill out some information and then it lets you download it for free. It's got a lot of stuff. I sort of feel like, wow, I could have used a few months ago, but on the other hand you sort of need to get to where you need to get before you can even understand it. But it did let me to start understanding this whole idea of what the series resistors on transmission lines were about. And it really, so the idea of shouting into a canyon and hearing echoes is a bad metaphor because really what happens when you send a signal down a wire is it's more like filling the canyon up with water to make it a 1 and when you want to make it a 0 you have to drain it back down again. So the wave is not an echo wave, it's like a wave of water. Like, you know, when Moses released the Red Sea or the dam burst and you get this rush of water going down and hits the far side and the wave comes back, the series resistor, if it's all properly selected, helps the wave sort of finish up, cancel out, level out, cancel itself out quickly. So the entire line now gets to the voltage that corresponds to the 1 signal rather than having the water slosh back and forth and slosh back and forth. Which makes a lot of sense. Right the first time in the book is not thrilled about the idea of putting series resistors in the middle of the transmission line, which is what the intertile connectors would do. But these transmission lines are short, they probably don't need it, but it would be nice to know we could do it if we wanted to. So I made one of these things, it seemed to work. I also last week had designed a circuit board for the intertile connector that was going to have actual room on it for tiny little resistors if they were small enough. And we're thinking of using these little resistor arrays, this little thing that has eight resistors on it, which is what we need for eight signal lines. I got some of those in, they are wicked tiny. I mean look at this, it goes from, who is that, is it Roosevelt? I don't know, it goes from his ear to the edge of his mouth for eight resistors. I also got in the circuit boards that I showed you the picture of just a moment ago from Osh Park, here they are. They all look fine as usual, which is not a gimmie. So the resistor array does actually fit the footprint that I put down. I was able to solder it up by hand, even though that the thing was all incredibly tiny, just the whole idea of what's called drag soldering, you get a blob of solder on the solder mirror, and you kind of go urrrrrr, and you just go down all the pins. pins, and then if you're lucky or talented, the surface tension of the solder will make it wick up away from the in-between things, or if you're like me, you use the magic, magic ingredient solder wick that just soaks up all the extra stuff. And so, you know, it worked pretty good. So I mounted a couple of the female keyed headers on it, which is how we make an intertile connector to hook two tiles together. And it came out pretty good. I'll show you that in a little bit. In addition, on the manufacturing front, we ended up saying, well, if the signal integrity problem really wasn't the issue, we could use the latest circuit board that I had laid out, or we could have used the previous one. Why is there any reason why we wouldn't want to use the newest one? Well, the thing was, was that I had changed some of the parameters because I wanted to get the ground plane to flood in between the little holes in the ITC headers and the holes and the thing that connect to the beagle bone and so forth. And with the default values that come out of Kaikab, the flood is too thick to go through those little gaps. So I made it more liquidy, and it got in there. But I was afraid that maybe I'd gotten too close and it was going to be hard to solder. Well, so what I did is I looked around the internet. I went to Osh Park. I looked for open source hardware stuff that people had posted. Thank you. And I pulled down several of them to see what they had done. And a lot of folks had done what I had done. Somehow they had turned it down so that the ground plane would flood through .1 inch headers, which is what all this stuff uses, 10th of an inch headers. But then on the other hand, several of the sort of big designers who sort of I recognized their name, unlike all of these projects, which I really didn't know who was making them, did not do that. Use the default values in Kaikab so that you could recognize that the flood plane didn't go in between the headers. So is it safe? Is it? I said, screw it. Let's go for it. So we're using the blue board that we showed last time. The blue board is now the golden board. We now have an order, and there it is here, to make 200 of these things now in progress. And I don't know if I showed this. I like the back. Do you see this up here? So designed and assembled in Albuquerque, New Mexico. Because in my heart of hearts, I was kind of committed to using ETS, the local guys, rather than going with the circuit hub on the internet. So that has all happened. The actual boards that are going to be inside the production build, the ring loaders of T2 tiles, are being manufactured. Next time I checked, they were more than halfway through, or about halfway through the process, or 200 of them, and they're scheduled to ship in a couple of days. I think we will have the boards before the next T Tuesday update. In addition, I finally sent the email to Robert Evans, the owner, I think, of ETS, and said, can we get together and try to actually make this happen? And that happened yesterday. And in preparation for that, I made a little document with all the questions I could think of to try to get them sorted out. I had sent an email saying, I imagine my responsibility is to provide all the materials, and they have to be on all the surface mount stuff, have to be on reels, because the machine that automatically does the surface mount needs to pull them off of tape. Are the fiducials, and this is the result of the meeting yesterday with my notes, the answers to the question, yes, yes, it's okay, and so forth. Little changes and improvements, and so on. We have an agreement. ETS is going to assemble the boards, the golden board, which happens to be blue, from parts that I give them on reels, and so forth. It's all agreed. I now have to work with UNM and get them to make a purchase order and get it all set up, but I sort of think I understand how to do it. And bottom line, how much time does ETS need after all the parts are delivered two to three weeks? So if by episode 20, for example, we have got all the parts in and we can get them to ETS, and the purchase order is in place, we could be looking two to three weeks to having a ring lotus of tiles in hand. Now by that time, we're going to need hundreds of intertile connectors, and that's a separate manufacturing issue that we're going to have to deal with, but step by step. And so in general, the connectors, most of the surface mount parts are in good shape. The header parts, these things, the connectors, these guys that can all of the actual through hole parts that other things mount onto, some of them I'm up to quantity, but a lot of them I'm not, still going to have to be worked through. So goal in the coming week is to get at least orders out for all that. Some of that is going to be, I'll express probably because in some cases it's the only place I can, I know where to find the things. And we're going to have to deal with the time delays and the shipment and so forth. We'll talk about that next week. All right, software, yeah. It's been a while since I've been working, we've looked at the T2 tile running powered up a couple of times, but it had nothing much has been done to the software. I did work on it some last week. This is my GitHub repository. If you're not familiar with GitHub, it's a place where people can share software using this underlying revision control program called Git, which is really wonderful, which is also made by Linus Torvalds. So making Linux is not good enough. He also had to stop and make a tool, a wonderful tool called Git. It's a lot like Don Knuth, the god of computer science theory, was writing the master work on computer science theory from a lot of points of view. And he had to stop and write tech as a tool. So to me, Git is Linus' tech, building a tool to do what he needs, and it's just a wonderful thing. GitHub is a thing that just provides it for free. If you're doing stuff in public or you can pay to have private stuff and so on. One of the things GitHub does is it shows how often you've been making commits. And for the last year, this is mine. This is pretty anemic compared to a lot of people because I've been doing a lot of hardware stuff. But finally, there's stuff starting to appear again because I'm starting to work on the hardware. The movable feast machine stuff itself has finally gotten some bug fixes and so forth. And it's about time. And this is one of the things, oh yeah. In the simulator, the movable feast machine simulator, which is going to be a not simulator as soon as we load it into the T2 tiles, it has a choice of different tiles, square tiles, rectangular tiles, little ones, big ones, tall ones, small ones. And so on. But only specific ones. And so tile H was a sort of a medium-sized rectangle. And I changed its size to make, instead of 64 by 48, whoops, I made it 60 by 40 because that makes it the same aspect ratio as the T2 tile display. And I got a version of it built and running and check it out. And now I guess I should put it over here. Can anybody see that? Probably not. And I like that. So this one over here is the new, this is the new version of MFM. This was the old one that's just running Dragon Res. This is running a thing I just whipped up called Foo, which is essentially like a Super Dragon Res that instead of just deleting neighbors, it deletes big areas. Oh, and also, they're actually talking to each other. And here is the inner tile connector with the series resistor array. And it's working fine. In fact, we can't see it right now. We'd have to get in through SSH and talk to this thing, get in through the network connection. But in fact, these guys are exchanging lock information. So now it's no longer a ribbon cable. It's actually one tile talking to another and they've negotiated there. In fact, they know they're there and they can pull locks and stuff back and forth with them. That may be the first time that's happened. So that is kind of good. And I made it just a little frame to mount these things on to make it easier to handle. Because these guys are going to get, this is going to be about, you know, we're talking 133 of these things hanging on a wall, sitting on a table, something like that in a month or six weeks. Not well, probably two months to actually have it here and running. Who knows? It's going to start depending on all the tiles and so forth. So there's going to be a lot more software tweaks that need to be done. I'm going to be thrilled to get to be able to spend. Now see one of the things about this, by the way, which we're not going to be visible here because we haven't gotten enough room on the screen unless we get rid of all that stuff. If you look at this thing, I don't know, can you see it? It seems like, you know, you can see it sort of refreshing whereas here, it's going more little steadily. And that's I think mostly because this thing is redrawn a bigger portion of the display. And this display does not have a tremendous amount of bandwidth too. So the more that you change in a given frame, the longer it takes like that. Some of that may be improved, can be tweaked and worked on and so forth. Once we get the hardware stable and we have enough that we can actually work on it, we'll take a look at it. But it might be that we actually want the H tile to be a little bit smaller rather than whatever this is, 40 by 60, something like that. But that's the kind of thing that we really are not going to know until we actually have a bunch of these out and we see how much contention it takes grabbing locks between them and doing this. And we can tune all of this stuff as an engineering problem to try to get on the one hand a reasonably good air, you know, one air, two air, 10 air, 16 air, who knows. Given this, I'm guessing less than 10 air once we get all done, but who knows. So that's all what's coming. I will really be able to get back to being able to work on hardware and software most of the time. One interesting thing, by the way, is that this, the guy over here is, was using Ubuntu packages for the Mobile Feast Machine and ULAM. This guy is using code that it actually built, we checked out the GitHub repos and rebuilt everything from here on the Beaglebone itself. It took a long time, like, you know, I don't know, I didn't even follow it, probably a half an hour or something to build ULAM, but it worked. And that means we're going to be able to build entire new builds on a Beaglebone and then use the intertile connectors, I'm hoping. I wanted to be able to show code moving through an intertile connector, but I didn't quite get that far this week, so maybe we'll see next week. I'm not going to put myself on the hook for it. But all of that is coming soon. Please let me hear from you if you have questions, comments. If you're, I think, another public meeting would be a good idea, stuff that you might want to hear about there, stuff you might want to contribute there, all would be useful to know. All right, the next T Tuesday update will be out in a week. We're getting close here, folks. Thanks for watching.