 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2TILE project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. This is the 33rd T Tuesday Update. Let's get into it. Last time we had the trip to Vienna for pioneers, it was great. In this past week, it was focused on making the inner tile connectors. I have a little bit of video, a little time lapse making them, but I'll drop that in here. Since then, the other thing this week, it's been about making a software load for these tiles. I mean, you know, these things have Linux in them and they have Ethernet and all this kind of stuff. But all of that is not really even essential to the research part, which is the movable feast and the ULAM and SPLATN programs and all of the robust first best effort stuff that we're laying on top of it. But the way it works out, the Linux load is coming in through a little microSD card and it's a little bit hard to change. So I spent quite a bit of time this week trying to come up with a load that was reasonably fresh and had things in somewhat sane positions and so on. It's not great, but that's what it was. And I ended up with the E-Series Flasher, a little microSD card that we plug into these things and hopefully not have to redo that. We're going to be using the Common Data Manager, which we talked about a month or more ago, to spread the research stuff, but hopefully the underlying Linux stuff will be relatively stable. We'll see what happens. We also, in the past week on Saturday, we had the second T2Tile project public meeting about how to explain this stuff that was really very helpful for me, not a whole lot of people watching the live stream, at least in part because social network genius I am. I forgot to make the link public until the last possible minute, but still the folks that were there, we had a good discussion and it was helpful for me at least. So in the coming week, it's time to get the camera-ready copy, the final copy of the paper that's going to the A-Life conference in Newcastle upon Tyne in the UK this summer. So that'll go out. I think maybe there might be some kind of thing that it can be an official preprint and we can kind of like talk about it after the camera-ready copy goes in. I'm not sure we'll find out. So okay, first up we'll talk about the intertile connectors. We got 375, it's a 354, I don't know how that happened, assembled PCBs with the Y-Pay-less connectors on the bottom and then these 3D printed handles that we put on top. I've now got a bag full of them. There's plenty more to be made and more coming still. In addition this week, I made a deal with the same company PCB way they're now going to manufacture the other kind of tile, the other kind of intertile connector. This is the PD or the DP that shares data and power that's used within a single zone controlled by power but we also need to have a different kind of connector, which is this is the handle for it, although we don't have the circuit board for it yet, that just connects the data and the ground pins but doesn't connect the power pins so that separately powered zones hopefully won't interfere with each other. The order for that has now gone in, that'll probably show up in mid-June so before then we won't really be doing a whole lot of between power zones but we have so much to do before we get to that anyway, hopefully it'll be alright. Okay, we'll take a look at that little video. The main event today is I put together a lotus. I put together 19 tiles, one with six around that, with 12 around that. I have never booted the entire lotus yet. I have booted fragments of it to test it out and see if things are working but I have not yet put the whole thing together because I thought, you know, what the heck, let's do it live and see what happens. So let's do it. So this is the question. Ladies and gentlemen, children of all ages, I give you a T2 tile lotus. It's not quite complete because you see these guys down here are not yet connected but all of these guys up here are. I mean this is big. I'm not even sure how much of it the camera is actually getting. Hopefully it's getting this. So what I'm going to do is I'm going to hook up these bottom ones so that we have the full lotus. And then we'll power it up. And so the reason for some concern is that, you know, when you first power up a piece of electronics, it's like all of its tubes for electricity are all empty. And so they all have to kind of fill up before things can start working normally here. Are we okay? And there we go. And you know, it's just like filling up with water. It's not just like filling up with water, but it's kind of like filling up with water. And you know, if you don't have that much head of water behind you, if you just have a trickle of water coming in, then it'll take quite a bit of time for the pipes to fill up before everybody gets to the level that they're expecting to get at, which you'd say so what? But a lot of electronic circuitry is expecting the voltage to go from zero to whatever its working voltage is rather quickly and the more pipes you have to fill from a given power supply, the more current that is going to be drawn from the power supply. And if the power supply is not able to supply that much current quickly, then the voltage will rise slowly and we can end up having the electronics not actually boot properly. So I'm hoping it'll be all right. It'll probably be all right. But this is why we test over the heck is the other end of this? Here we go. All right. So you're ready for this? This is the first T2 tile lotus attempt to boot in the history of the universe. Here we go. All right. So we've got tile power everywhere. Those are the reds. We've got grid power. I'm sorry. We've got grid power. That's the reds. We've got tile power. Those are the greens. They look plausible. But we have one of them. We have this one here connected to the serial port. And where did the serial port go? Here we are. Okay. And it looks like it's kind of booting. So maybe we're all right. Come on. He's not going to boot. I mean that would be exactly the sort of thing that I would be worrying about if there was not enough current to supply this whole thing. That these guys might have gotten kind of starved. It looks like not everybody booted. So we have 17 out of 19. Well, so I'm going to try, well, I don't know. I guess I'm going to try, well, see, and this is the other kind of thing. You don't actually have to boot the entire universe all at once. You can hot plug these things. So I'm going to try power cycling these guys and leaving everybody else alone. First, I'll just try a little polite. What I'd really like to do is move the serial port over to one of these guys that's currently. Ah, oh look, so it's not, it wasn't even dead. So this may or may not actually start up okay. I think it isn't. I'm going to force the power cycle these guys. This is the kind of stuff I was talking about. It's terrible that due to my incompetence and the underlying software stuff and a shift to this thing called system D. The Linux boot process on these things. How do you like this for an irony? It is non-deterministic, but Linux is not ready to handle it. So, you know, it's not like, oh, so you like non-determinism, Dave. Stuff this in your pipe and smoke it. Yeah, I like non-determinism when the rest of the system was designed to handle it from the bottom up. Here, there's some kind of race conditions. Things might happen one way or they might happen another way. And especially, I don't really understand it that things will sometimes kind of get hung, either permanently hung or hung for a very long period of time and then eventually come up in the boot. All right. So, there we have it. We haven't quite gotten to... All right, we've got MFM running. We've got MFM running grand. I'm going to go back over here. Let's see if I can hook back up to this guy. I need to get some kind of little extender thing for the serial port. A little 3D printed gizmo that would extend these things so that it would fit in properly. There's a pain in the butt. All right, well, screw it. I'm not sure if we're getting anywhere. Can we display them talking to each other? I don't know what's going on there. I probably pressed the power button on him, didn't I? Let's get him booting back up. Well, you get the idea, huh? We're going to figure out how to make computations that work on top of all this and not even notice. We're going to try it out. Rebooting in the middle. The chunk missing and so on and so forth. And that's what's coming. Because we're giving so much freedom to the hardware. So that's a definite problem connecting the... I got in. So what I wanted to do, let's see if we can get everybody, is I wanted to drop a chunk of new content into one of these tiles while we're displaying the talk to each other and see if we could see it moving. So I made a copy. So if we move CDM 11 to CDM common, then eventually, I guess we're not getting that much debugging. So we'll just have to see. But CDM ought to notice that it's got a new file here. It should gradually pull it in, parse it up, find out what's in it, check some signatures and so on. And eventually start announcing it to the neighbors and we should see some bigger arrows going as data starts getting exchanged out of this guy and then gradually it should flow all the way to the farthest corner of Lotus. I don't know if we're actually going to want to sit around and watch all that where we at here. See, he's consistently sending relatively big packets to this guy. So that'll mean both of these guys will know about this new content and we should see some big arrows going. Oh, it's going the other way. Could have gone this way. And usually, so this is, I don't know, 10, 12,000 bytes or something. It's not big, big, but mostly we're thinking about moving MFZ files and configuration files that hopefully won't actually be all that big. There's some stuff moving. And hold on, not entirely sure if that would be the same thing. Maybe you need to enhance our display here to give us a little bit more information about what's moving. I mean this is all, again, the low level debugging stuff and as we saw a while ago, the movable feast is still running while all of this is happening. We just tell it to not do its display and then send our own stuff to the display instead, which is one of the reasons why it's all so slow because it's actually sending signals between processes and so on. All right, well, so sort of a mixed bag. We definitely got it all booted. It did not all quite come up cleanly, although it does seem like that may have actually been the raciness and system D and so forth that's just due to, you know, let's say just due to my incompetence in using system D and that let it go at that. But with a little bit of futzing around, which kind of drove home the point that you can futz around with this thing while it's running, we have the full lotus. We have 19 tiles, they're talking to each other. So, you know, basically what I want to do and actually next week I would like to set some goals, set another plan, another target to try to get all of this stuff lifted up to the movable feast level. But it took more work than I expected just to get to this this week. So we'll do that next week. This is very cool. And, you know, stuff is just going all over the place. Let's go back to MFM mode here. And if we can, if everybody see, and the response to this button is not as snappy as it could be because it's being done by shell scripts, pearl scripts. But again, these are things that can be tweaked once we get the basic nutrition working again. Oh, yeah. And, okay, so, yes, it will boot after a fashion. The next update will be out in a week. We'll have a plan for getting to movable feast events talking to each other across tiles. Once we get to that, we can benchmark this thing. We can set the initial record for average event rate on an indefinitely scalable computer. And that's our big goal. And then we're going to play with this thing. I mean, this is one lotus. We're aiming at a ring lotus. This thing with six more of these things around it. We're going to have to be doing some infrastructure work to figure out how to manage all this. And that'll be something to talk about next week as well. Thanks for being here. Have a good week.