 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. This is the 39th T Tuesday Update. Let's get into it. Next week is the 40th episode deadline to have a ring lotus. That's at 133 or more of these tiles connected together, running MFM events between them on software. No, we're not going to make that. We may be able to get to having multiple tiles, having MFM events connecting each other across tiles. Even that is going to be tough, you know, it's funny. I think of myself as a software guy, but that can lead me to overestimate how much progress I can make in software. I'm my own pinhead boss. When it comes to estimating how much software progress I'm going to make, you know, when it was hardware, okay, it was hardware. But then we've been doing this since October. So we'll see what we can actually do. We'll have made a lot of progress if it actually slips a week or two to getting MFM events running inter tile. Okay, so be it. Today I'm just going to do the news and next week we'll see where we're at. All right. So, whoops. All right. So, yes, our top stories this week. It's all about, well, it's not all about, it's all about a bunch of things, but one of the things it's about is getting out and spreading the word as best as possible. This week for me was a historical landmark because I did three podcasts in one week, including two in one day. This past Friday. So the main one I want to talk about is this biota podcast done by this guy named Tom Barbalette. He had one of his episodes with Anton McKaylo a few weeks ago. That's how I found out about Anton and Tom contacted me and said, what should we get together and we should talk about this stuff and so on. So we did that. And that's now up. I'll include a link here to listen to it. It's not bad. It's about 43 minutes. The recording time itself actually was more like 46 minutes, but he actually did a pretty skillful little job nipping and tucking and tightening things up and getting rid of my us that I do like that. And he liked that and so forth and just generally tightening it up. One of the things that I particularly liked about it was that he actually, you know, pushed a little bit about the idea, you know, why isn't cloud computing what we already have good enough and so forth. And, you know, when I'm lucky enough to have someone like Tom who say his objection, then we can have the back and forth and that's a part of that I think is one of the good parts of this podcast is talking back and forth and working through the issue. The problem that I normally run into is that people aren't willing to raise their objection because they whatever they can't frame it or they don't care or whatever it is, but they still just think it, you know, oh, this is just cloud, oh, this is just fault tolerance. And so therefore the discussion doesn't get had. So I really do appreciate Tom for going ahead and asking the question and I hope it was a good back and forth and so forth. In addition, on the same day last Friday we did another computing up Michael and I about the progress is it is really possibly progress or is it all just kind of screwing around when I feel depressed. I it's easy to think that, you know, progress is an illusion and so forth. But also there was yet a third one which was a podcast by some folks at Brown University that were connected through Michael that are doing starting a new podcast. I don't know if the episode is out or whatever when it's going to come out. I'll let you know if I find out about sort of computational topics and their ideas. They'll have experts and they'll have interviewers and they'll have sort of a normal person, a sort of man in the street person to try to provide a regular perspective on the conversation, which seems like a good idea. In our opening one, they didn't actually have a man in the street. They had their two interviewers and Michael and I and we were talking about autonomous vehicles, self driving cars. It's not something I actually know a tremendous amount but that didn't stop me. So I'll put a link up whenever we find out about that. Okay, hardware updates. At long, long last, long time, viewers will understand how I can feel about this. We have, you know, I have given the paying ETS for having built all the circuit boards for us. I was being a little negligent because I had gone over to pick up the last bunch of circuit boards and I hoped to talk to Robert ever since then at that point but he was on vacation or something and then it just kind of slipped. But now, they've sent in the invoice. I've signed off on it, $2,450 for 200 boards. I haven't tested them all but I've arranged with Robert that we have 90 days to actually test them. So that's one important drop-dead and that if there are hardware manufacturing problems, they'll make good to fix them up and so forth. They've already done it for one. So that's a landmark in the hardware development side of things. Oh, in addition, we need all the connectors to connect them together. And I've been talking about this a number of times. Here is one bag of the PDs, the power data connectors that would be the workhorse that we've been seeing all over the place here. I also, in the last week or so, have finished. There's another bag of those PDs but these are the DOs, the data only. I used gray plastic to do their handle so they're going to stand out visually a little bit from the tiles. This is it. This is, I think, all of the DOs that we need to do a ring lotus plus some stairs. But that's not all. There's actually one more thing that we need. We need the PIs, the power injectors, which, and I've got a, I've been 3D printing them, I'm considering doing the 3D printing in white. So we have black for the most and then gray for the middle and then white for the ones that actually feed in the power. These guys have a little snake inside because there is a cable that goes in, comes in here and snakes through them and so forth. We have a matching little snake bit that goes on on the other side. So then there's where the wire goes in. Oh, except I forgot. We also have to connect to our circuit board but we need the circuit board to do this. So this is the little latch that will go under the circuit board. It goes into matching holes here and then this snake cover goes on it. You can do that like that. And then finally we have a latch that goes over the whole bunch like that and you push it down to make it go. I made a new slight revision here because the ones that I have, it feels like this overhang guy tends to come off too easily. So now I've put a little flare in there. I don't want to actually seat this thing yet because it hasn't got the circuit board in it. It's been redesigned and there is the design. There's the snake I was talking about, the latch and so forth. And this little bit right here is the new part that's going to catch the ring going over and keep it on. This is the circuit board that's going to do it, the power injector circuit board. I've touched it up just a tiny little bit and ordered 21 of them from Osh Park. It's a four-layer board because it's got a big power plane going through the middle and a big ground plane going through the middle because these things are going to be carrying several amps of current once they're handling an entire lotus or an entire ragged rectangle, whatever we actually end up doing for the fundamental unit that we tile our wall with. So that's out. 50 bucks, I also donated a buck because PayPal offered it to the Internet Security Research Group. They're the easy and crisp people and they're doing great stuff. So I felt good about that. And the boards have been panelized, well, sent off to the fab with a bunch of other things. I expect Osh Park gives itself a good lead time so that they can usually be early, which I'm fine with that. But of course once you've done it a few times, you kind of start assuming that they're going to be early and so forth. So we'll see. I would hope that these things will ship sometime in the coming week. We'll see what happens. All right. In addition to having the circuit board and the connector, we actually need to have the wires that are going to come out of here. I ordered a bunch of these cables, 18 AWG supposedly. They'll carry five amps easily. And we'll use those to make the connectors up so that we can connect and disconnect to the power without actually needing to pull the connector out of the tiles themselves. In addition, I've been concerned about the heat coming off of these things. One of the problems with this particular chip, this particular silicon system on chip is that the internal temperature measurements for some reason are not that reliable. I don't exactly understand why TI has Texas Instruments who makes the chip. I don't know. I haven't tried to run it down. But the normal thing of being able to just ask the chip its own temperature doesn't seem to work all that well. And when you look at what they put out there, they say, oh, you should put a contact sort of thermistor to measure the temperature and so forth. I didn't do that. I have the thermistor nearby on the facing. Here's the Beaglebone with the chip on it. And here's our circuit board that went on here. It's got a temperature sensor near the thing. But I'm worrying because when these things have been running for a while, you can even kind of smell the fiberglass, the epoxy. It's getting a little bit too hot. That's going to just shorten. At least it's going to shorten the lifetime if not going to cause it to have more failures recently. More quickly, so any event, I ordered some little heat sinks as part of, you know, maybe we'll put a heat sink on the CPU. Maybe we can get some fans in back there on the acrylic panels or something to try to move a little more air and get them cooled down. So I ordered some of these. These are 15 by 15, which is millimeters, which is just about the size of the CPU chip, the silicon on the system on chip, 6.6 millimeters high. I was a little bit worried whether that was going to fit. I mean, you know, with this board and the Beaglebone facing each other, whether there was going to be 6.6 millimeters in there on top of the Beaglebone chip in order to get it. So I just ordered some from Amazon rather than doing an order from China. I'm getting a little tired of ordering stuff from China. It's just too much stress and aggravation, even if I end up saving money. In this case, there's, you know, it's sort of late in the game. A lot of money has been spent. It's easier to think. And again, this is just a preliminary test. I also got some of these other ones, these black ones will also have the peel and stick tape so that you can just sort of stick it down. These things are even higher. They are 7 millimeters high and so forth, but they're only 14 by 14. So I just got them to try. They showed up. They showed up in the video of it, but rather than let you enjoy it, I'm just going to give you a couple of individual frames. So there's one of them. I opened up. So of course, the terrible thing about all this, you have to take apart the board all the way down to nothing. You take the case off, you have to remove the screws on the Beaglebones. So I'm removing the screws and you have to pry the Beaglebones off. They have 96 of these pins buried in their headers on our board. And it takes quite a bit of energy. It's relatively easy to press them down, but getting them off, you kind of have to work it off by levering the corners as you go around. So it comes off and the Texas Instruments silicon system on ship is right in there. One of the temperature sensors on the facing board is right over there and so on. So I took one of the silver things and I just kind of dry fitted on top of the thing and then I was just using the liquid crystal display to kind of measure across whether it was going to go. It looks like there's really significant clearance. It's not bad at all. And there's the peel and stick tape, which I cleaned off the CPU chip with iSpoble 99% iSpoble alcohol. Some of the people in the reviews of these things were saying that the peel and stick stick wasn't all that strong and some of them had fallen off. I'm hoping that's not going to be a problem because we have no vibration here, at least not yet, but we'll see. So I wanted to try to get a clean surface for gluing. Peeling off the tape and sticking the thing down seems okay. Closing it back up, putting the display back on, putting the feet back on now. You know, these feet are at the moment, the metal socket screws that go through the case and the circuit board are just threading straight into the plastic of the foot with no metal to metal contact there. And I wasn't necessarily even expecting that they were going to last all that long, but they seem to be going in pretty well as long as you get them in the threads that they were when they originally bore the thread through. It seemed to work out all right. And really all they need to do is be strong enough so that when you pull a tile out of the backboard of the female connectors for those feet, the feet don't pull off the screw. As long as that holds, then they're pretty good. So maybe we're going to be all right. This was meant to be sort of just a temporary first cut, but maybe, you know, given the needs of time, it's going to be the final cut as Andrew Walpole said to me. We shall see. Screwing the thing back on using the little wrench, trying the black ones as well. And I'm like, I'm taking these black ones. Where the heck is the peel and stick tape? I can't find the peel and stick tape. And oh, they gave me a Velcro cable tie, which is very nice. I'm not sure what I will do with it, but they didn't actually stick the peel and stick tape on. Even though in the picture, if you go back and look, it's shown stuck on to the heat sink. So these guys are bogus. All right, software as well. That's what I was doing mostly this week. What I ended up producing is a new flasher chip, a new flasher disk, a whole new software load with all the updates and so forth rolled in. And getting that working was a huge pain, which I'll talk about a little bit, but running out of time already. One thing I discovered, this was kind of interesting, was that the common data manager, when it was shipping around the MFM distribution, it has now hard-coded stuff in it, a procedure to go ahead and install the new distribution of the Moogle Feast Machine when it arrives via the common data manager. And so it takes the existing tree and it renames it to the previous tree. It creates a new one. It unpacks the MFM distribution into this new tree. And if it all goes well, then it removes the previous one and so forth. It was just trying not to mess it up. And I've now discovered that, you know, I would see this happening over and over again. Structure needs cleaning, read-only file system. What does that mean? What that means is the underlying file system has gotten corrupted. Structure needs cleaning is an error message that can come from when you try to do a disk operation where the disk has problems. And read-only file system happens because once a disk corruption is detected, this thing is set up to automatically set the file system to be read-only to try to limit the damage. My installation procedure for installing the MFM would reliably corrupt the disk. Thank you very much. Very helpful. And of course, part of the problem with that is, I mean, this is a journaling file system that tries very hard to make. If it makes changes, they'll all sort of be consistent. And in fact, when you reboot, it would clean up the disk. It would, you know, go through it, discover problems, and fix them up. And it would say, okay, I'm happy it's fixed, but you're more all the changes you did recently since up to the time of the corruption are gone. And I had several cases where I had to redo the entire flash process over again, which takes a couple of hours, just because I had a corrupt disk that rolled back and undid the software changes I was trying to get in there. The answer to this, it turned out, was after every file operation, I had to explicitly do a sync command to force all the changes out to the disk before going on and doing the next operation. I really wouldn't have thought that was necessary. I really wouldn't have thought the system, the Linux system, would go ahead and corrupt the disk, but it did. And, all right, I'm going to skip all the, oh, well, you know, so then the installation itself was failing, following packages cannot be authenticated, ooh-la, well, it's kind of important that be authenticated. What turned out? The Ubuntu public key server wasn't responding. Okay, great. And getting proxy errors. A million things that we all depend on. I'm going to skip the story of the programmer in the interest of time. This is my programmer icon, except I made a new one. All of this was just to get at that you think this is one of the way programming ought to work, but in fact it goes through GitHub repos and the Ubuntu key server, and the packages go from one to the other, and then the common data manager, the first thing I was doing after I updated it once, I put it in and the old stuff was back again. Why? Because the neighbors were feeding the old stuff back in again. I think I have a new load. I think this guy, our current set up of these guys, these are all, have been all updated to the series F flasher. Hopefully we'll be all right. Okay. So we're skipping the circle of life. So in the community updates, yes, that's why I wanted to save some time. Marius Cezynski has been working with Anta Mikhailov to get this GPU based, super fast stuff going, working on Linux. It's turned out that it's not even just Linux, it's really just a matter about porting it from NVIDIA to anything. They've made a lot of progress, and I want to thank Marius, especially I haven't interacted with him that much, but he's just out there doing it. It's great. And finally, Golo Kohan had a question. It seems very interesting, but can't you do anything more complex? All you have was a sorter. I just wanted to end on this because, you know, we got new people that are always showing up and there are, you know, this is the earliest days and the stuff that we can do on this is very simple and very small because we're essentially starting over. So this is like assembly language, the lowest level of stuff. And you really don't want to be putting it up what we can do today against complex software like stuff that's already out there because we can't compete. We're not even trying to compete. We're trying to build a new approach from the ground up, step by step. So there's a bunch of other things. There's various versions of the router. There's a 2D printer. There's Game of Life. There's a little data packet router and so forth. I will put some links down in the bottom of this if folks can help us out in comments if you know other things because they are scattered around in videos. It would help if we could jump in and let people know where to look to find more stuff. Okay. I will put a little bit of Anton's stuff at the end because he's got some new stuff that's very cool looking and it makes for great looking thumbnails for the T Tuesday updates. Next week will be the 40th episode. It'll be special because it's our 40th episode deadline. Exactly what we're going to have for it. We'll find out. Thanks for coming. Hope to see you next week.