 The T2 Tile project is building an indefinitely scalable computational stack. Follow our progress here on T Tuesday updates. Hey folks, it's been two weeks, our top stories. The T2 Tile project has reached 500 subscribers, it only took less than two years. I guess it's mostly been kind of slow and steady, and for sort of this level of pretty esoteric material, I'll take it. Thank you folks for checking it out when you drop in and see what's happening, and unfortunately I never have as much to show you as I would like. For example, so it's been the Wall of Science, it's been the A-Life 2020 Now Virtual Conference, its final deadline was May 1st, it's almost two weeks ago now, submitting down to the wire May 1st at 11.59pm, and okay, I submitted a paper, and you know, like I'm not supposed to talk about it until it gets published or it gets presented or something like that, so this is the title, Best Effort Computing with Spots and Spatial Threads, and it's pretty fun. I think it's got some pretty good potential actually. So we'll talk about that a little bit later. I guess what we know now is that after everything with the two-week delay and then the month delay on top of that, I still only managed to write two pages, so it's an extended abstract, it's actually really a teeny little paper crammed into two pages. This year they let you have your citations spill over onto a third page, so but these two pages are pretty tight. And I was submitting it down until four o'clock in the morning on May 2nd, the day after it was done. So I've got a pile of papers now that I'm going to be reviewing for other submissions and so forth, I've got to get on that, but that's the state of the wall of science. The other, well one of the other major things that's been going on for the last two weeks is pushing on the native engine, you know, remember that my original idea was we've had the simulator MFMS that has been running for years, and I imagine since the T2 tiles that were actually running these things on our little Linux boxes and MFMS was designed for Linux that we would just run MFMS on Linux, I'm sorry, on the tiles and it would be easy and we would get very compatible behavior between them. And that turned out really not to be true. The assumptions were different enough between the simulator which had to simulate a whole bunch of tiles and it did all of those multiple tiles all inside a single global random access memory and then used, you know, memory to memory communications to exchange data between the simulated tiles and the actual T2 tile where it's just simulating, it's just being one tile and the actual communication is not in memory but actually goes out, bits through wires, electrical voltages, wiggling and so forth. So as of a month or so ago now, I said, well, hell with it, we're just going to start over, we're going to build a new code base and then start pulling in bits from the old bit once we have it clear that we're making this thing exactly the way the T2 tile wants it, not the way we guessed the simulator wanted it five years ago or more now. So it's going well. So there's a lot of code being built up. I'll show you a few little pictures of it, but at the moment it's really not too exciting yet. So I always hope to have more progress, but this is where we are now. Mostly it's been about, as I've mentioned before, putting in stubs for bits, more bits of code and then getting the thing running well enough so that the next stub blows up, incomplete code, incomplete code, check circuit availability, incomplete code and get visible wrecked, incomplete code and just keep chasing the incomplete code, pushing it down, pushing it down and being able to see how the thing starts to come alive. The program starts to behave, it starts to act and then bam, it hits the export that doesn't exist is cash valid and so forth. Start getting a little punchy working on this stuff as much as I have. But, you know, there's just these whole regions, you know, T2 ITC custom state handlers manifest that just, you know, you as a programmer, you're just wicking them out of nowhere. You just up and then bam, we have custom state handlers. So the, a lot of progress on that. Here's a picture. The whole thing, the T2 native engine is built around a priority queue, a time based queue that just pulls off the next thing, the nearest thing that wants to go, although in fact, if you look closely here, so you can see several things in fact have expired, meaning they're now actually in the past, but we haven't gotten around to pulling them off the queue yet. So ITC and W intertile connector Northwest, that's one of the six intertile connector processes that's working independently that they all have representation on the queue. The, the, the analog to digital computers, converters, they're sampled off of the time queue. The EWs, EWs down here, those are the event windows. Right now, they're just all stuck in active in it. So they're not actually doing anything. Where is it down here? Oh yeah, the EW initiator, that's whole purpose of that process is just to keep starting event windows, tossing them out, letting them take however long they take, and they have multiple steps on the priority queue, on the time queue, as they get packaged packets back and forth from the neighbors and they make their way through. They finish, they commit their changes get sent to the actual detail memory, and then the event window gets retired. So it can be EW initiated again, and so forth. All of this is feeling very solid and seems to be working really well. It's just the things that get released to the intertile connectors that need to talk to each other and the event windows that need to actually do the events, the transitions are just at the limited, most limited stages. But nonetheless, we got here. This is, you know, like when you're building a telescope or something, and you get first light the first time you actually look through the thing. This is a thing right here. That that's a drag. That's a dynamic regulator. It's not a real one. It's not the one written in ULAM or SPLAT. It's a hard coded one that's developed just for the native engine, because getting it converted to actually pull in the ULAM code and so forth is going to be a whole nother stage after we get the precision locking and the event window, parallel event windows, the event window pipeline all working. So here it is, our first little drag. Has it moved yet? No. I put it up in the corner because that has different consequences. We have to consider whether neighbors might be interested in what happens up there when it was in the middle we didn't have to care. After working on this thing for 10 days, I was excited to see first drag. So that's where it's at. Now, I've mentioned before that I have worries about how these things are going to look both in the light because they really aren't very bright. The displays aren't very bright and they have a very limited viewing angle. So if you're looking at a whole panel of these things and some of them are way over there and some of them are way over here, the worry is, is that you won't even be able to see it from a single position. That's not necessarily so bad if you're actually at a physical display and you can kind of wander around and see things. But if we're trying to capture this with a camera, which we are, then it's a bigger problem. So I discovered that WaveShare, which are the folks that made the original version of the displays that I have here, have one that uses an IPS panel and it's advertised. Its main thing is it has a big viewing angle, 160 degree viewing angle and that is pretty tempting. So I'm going to order one or two of these just to try it out because I think it's almost a drop in replacement for the ones I'm using. So we'll get a few, we'll see what they look like. They are three times the price or more of the ones I already have. So we'll see how we do here. But starting to think down the road, it's going to come. It's going to come pretty soon that we're going to be looking at a bunch of these and we want to see them. And in that space also, I've been looking at cameras. You know, where can we get a camera that has, you know, 8000 pixels, you know, a really big high res camera that we can record flat frames, nice, you know, snap the whole thing at once so that you can zoom in on each individual frame and get out. In principle, you might be able to recover specific say, oh, that's a drag. How do you know based on the color? I found these FLIR cameras. FLIR is the people who are now famous because they do all the thermal, the temperature probe stuff. So all these cameras that are supposed to be detecting whether people have COVID-19 because they have fevers and stuff. A lot of that is FLIR. But they also have sort of these industrial cameras for like scanning and machine learning for detecting the good potato chips from the bad potato chips on the assembly line. And so these, the Grasshopper Flea, the Oryx, I don't know, these range from like 600 to $6,000. And that's before you get a lens for them and so forth. So they're not cheap, but it seems like they might actually be the way to go. I need to learn a lot more about that. I need to find someone who can teach me about that. That would be the best. So that's the state of the native engine and looking down the road towards collecting data off of the entire grid. Nine power zones, each with 16 tiles, all going at once. That's our target. All right, and then the other major business that happened today in the last two weeks was advancing the story on the Living Computation Foundation. And a lot of that was paperwork. So in order to become a tax-exempt non-profit charitable organization, you have to file, well, you have to file one of a number of forms. We had to file 1023EZ, that was the easy one, streamlined application for it. If you were gonna have more than $50,000 annual receipts per year, annual receipts, then no, you can't do the easy one. Well, I don't think we're gonna have $50,000 a year, so we're gonna go ahead and do it. We filled this thing out. The instructions, only 23 pages for the instructions, including all of these incredibly complicated worksheets that all assume you can look at your previous years, your previous end years, or how much you made. But when you're doing it from where we are just starting, you have to kind of guess and make predictions. So that's what we did. And we got through it, we got a little advice from our lawyer about what these various things meant, and we got it done. And we paid $275 to file the form. And so that's where we're at. We haven't heard back yet. I don't know whether this takes about a month or something to get approved. We'll find out that went in May 5th, May 6th, something like that. So we'll see back once we're in actual tax exempt, official charitable foundation. That means that we can actually buy stuff and not pay federal taxes for it, which is good in state taxes usually. We have to file extra paperwork for that. But perhaps more important, people who are considering donating a chunk of money can take a tax deduction for their gift to us. So that would be great as well. That's where that is going. And so now we've got the website updated. Let's take a look at it. So here it is. It's the same thing, whoops, the same thing that we saw before. I keep forgetting this is actually real. But it actually has a PayPal donate button on it now. And one can actually click on it. And it will, I think, yeah. You can donate $5, $10, $25, whatever you want. And so I cooked up this little scheme, this little offer for the early bird nerds. This is us. And here it's described on the new contribute page. Ways to support the Living Computation Foundation. Why would you want to do that? Who knows? Well, why would you want to be watching these videos? Cause you get it. The 500, the 100, the 150, whoever you are that actually watch the videos. So the foundation's brand new. I tried to think of a lot of different ways, not just money that one could support it. And really, I feel okay. That the best thing that people can do, the easiest thing from some point of view is to talk about it. If you get it, if you get indefinite scalability, if you get robust first, best effort, talk about it. We gotta make people expect better from computing. If you're a hacker, you know, we've got folks that are working on Oolong code and Splat stuff. It's fantastic. I appreciate it so much. But now there is this possibility of financial support. And there's this careful note that we filed the application, but the application hasn't been approved. So maybe, you know, if you're gonna be a big donor, you might want to hold off for it a little bit. But if you're in a position that you want to donate five bucks or more then in exchange, we have this great perk, this great swag, you can get your own exclusive living computation foundation nerd number that each one, where is it? These exquisite finest quality numbers are made from crisp and pure it's starting from 10. And then each is individually Lexing coded for long lasting robustness. So the Lexing coded form of 10 is 210. The next one's 211, 212 and so forth. So in exchange for a $5 or more PayPal donation to the foundation, you will get your living LCF nerd number to show proudly to other LCF nerds or whatever. And that's it, you know, it's just a number. It's a number for us. So last I looked, three donations have come in. So the first three numbers are gone. All the rest are available. I'm sure the 200 series, which will last, probably last quite some time, but we'll see. If you want to join us, if you've got five bucks. Thank you, that would be cool. Nerd number is not transferable, but they're good in perpetuity for nerd cred bragging rights and nothing else. There you go, FOMO. All right, and so that's where we're at. Yeah, that's it. The next update will be in two weeks, which is gonna be the end of May. And I bet, I really bet that the native engine is gonna be cranking at, you know, intertile stuff. I'm really open for it. We'll see, stay safe, have a good two weeks. Hope to see you next time.