 The T2 Tile project is building an indefinitely scalable computational stack. Follow our progress here on T Tuesday Updates. Our top stories this week. The frame hardware, in addition to the actual T2 tiles, we have all of this backing support material so that we can hang them on a wall in some kind of vertical position and have them not be weight, have all the weight not be hanging on the electrical connections because they can't take it. I've been working on the frame on and off for months. I believe all of that stuff is credibly done. Now you just add a one inch pipe on a stand or attach to the wall or the ceiling and we should be able to deploy the whole thing. These were the actual final pieces. Eight of these go together to hold a single power zone, which is an array of four by four tiles. In addition to that, we needed some kind of interface for the tops of the power zones to connect to the one inch steel pipe and that's what we've been working on. We did a bunch of possible designs for it with different trade offs. This is the one that I ended up settling on that was fairly short, but it kind of latched over the pipe and it had a pretty good chunk of plastic where it needed to strengthen up to keep it from flexing. Printed up a bunch of them in the Urban Gray Prusament PETG filament that I've really come to like quite a bit. The PETG seems really quite nice and did eight of them ultimately in the gray, did not have enough of the Urban Gray to do all the rest, decided I was going to use the other colors so that we could have sort of one column that was headed by one color, another column was headed by gray, another column was headed and so forth. So I've been working doing that up and there are the gray ones, the eight gray ones that each pair of them supports essentially a column of four tiles. A column of four per power zone and then all the way down as far as it goes. So we did Ultramarine Blue. They also, you know, each one has these two pieces. It's got the upper curvy thing, but then it's also got this part down here, which is the part that actually connects to the frames, which are part of the repeating tiling pattern. These things were looking pretty good, although they seem to have a little bit of a possibility of kind of flopping over more than I wanted them to. So I tried to put some little retaining clips in them. There's another picture of an eight coming off. You can kind of see the retaining clips right here. I put it under their microscope. It's this piece right here down at the edge that tries to pop over the frame piece when it comes clicking in. That turned out not really to work because this thing fails almost immediately because again, it's being done with the layers stacking up, which are not very resistant to forces to the sides. I didn't care. I just went ahead and did it. They're going to have little busted little nibs on the thing. So be it. Did Carmine Red with glitter. And that's the other bunch. So we've got Urban Gray, we've got Ultramarine Blue, and we've got Carmine Red, the Red Interface, Pipe Interfaces as well. And that's it. And so the Urban Gray ones you saw, they're still on the power zone we've got the rest of them, the blue and the red. This thing looks like a bag of gift wrapping supplies. But with that, with all of the other stuff, I think the hardware is done. There's a tiny little asterisk about how we're going to hang the wall warts, one for each power zone, small amounts of electromechanical stuff like that. But I think basically we're there. So that's my road. All right. And I wanted to spend a little time especially because I really hadn't done much work on the project myself. But the community is continuing to build the community is building stuff. And it warms my heart. It gives me hope. It gives me confidence. So I just wanted to feed a little bit of that back to the folks who are watching the videos. And so people will know. Uh, Mario Casero, hello, showed up from Sao Paulo, I guess, Brazil. And, you know, I love these things. You know, I've been, I've been around robust first and a life for more than seven years. Now, you know, this is so great. You know, you do these things. You put these things out there. You hope that people are hearing it or seeing it or some people may be getting it and then they show up and it's totally great. So, uh, Mario has been power watching his way through season one, the first year of the T Tuesday updates. He's he's gotten as far as two 29 as of a few days ago. Hi, Mario. Thanks for surfacing, uh, you know, to take that step from just sort of secretly watching videos like everybody does like I do to actually saying something. I know that's a big step, but I would very rarely do it myself. Always wanted to, you know, just kind of stay safe. But, but folks to pop up and I really appreciate that. Another one is Glenn Rusling, who put some comments over on the YouTube channel and, and he's like a hardware guy. He knows about embedded. He was saying, you know, why not use application-specific integrated circuits? You could do all kinds of incredibly great stuff for, for low power, low price and so on. Uh, I do, we had a comment, the little comments back and forth. And, and he hadn't, he's now discovered the T2 tile project and he binged through a whole bunch of them as well. Um, Glenn, thanks to, thanks for taking a look. Thanks for your comments. You know, I really hope that one day in the T3 tile or the T2.5 or something like that, we get to the point where we could think about optimizing the hardware because we would have built the stack far enough up, we would have gotten experience with it so that we would know what we want. That's what I think is the problem with most of the parallel chips. Like Glenn mentioned, a thing from a company called Green Rays and there's, you know, sort of Attila and Parallel, there's a bunch of different things that end up in, as far as I can tell, being very niche. Um, and usually it's because compared to what it's looking like we need in computational power, preach a little bit, the hardware guys haven't been willing to expend that much silicon, that much design and, and hardware down at that level. Even the, uh, Intel, Knight's Corner and Knight's Crossing, all of those things that were putting entire Pentiums and paralyzing them. They were starving the amount of SRAM, starving the instruction stack, instruction caches and so forth. So my goal in this project is to actually, you know, build all the way up to where we can demonstrate not necessarily the actual utility, but demonstration and toward utility and then say, if we wanted to have that, we could have that. Now, how would we like to optimize the hardware and how would we budget it and where will we spend our gates? Uh, and, you know, I would love it if, if, if Glenn or, you know, other embedded folks that have experience in this area, wanted to help out, wanted to contribute, wanted to be part of that kind of development. It's down the road. I have hope. And Spencer Harmon, who is definitely got the most experience actually in the MFMS code base and which I, which I say, thank you and I'm sorry. And once again, I always wish my own open source code were better quality than they are, but you, you ship the code that you've got and you try to make it better. And harm and Spencer is working on getting better JSON compatible output from the simulation. So it'd be easier to visualize. Andrew also started working with them about developing some JavaScript type visualization stuff. I don't know how far it got, but that is fantastic. Thank you. It's totally great. And Anton McHale off. So we taught, I talked several months ago or whatever it was in the previous bunch about using GPUs to speed up simulation, you know, to make cut a few corners, make some simplifications in the language that you can use to express, but not even that many simplifications. And then we were seeing, you know, for fork bomb and dragon rescue, incredible, huge arrays running fast. That stalled a little bit because it seems like Anton ran across a kind of a compiler bug in the underlying stuff that was supplied to be driving the GPU processing cards and they weren't really able to fix it because they wanted everybody to migrate away from the previous software base to this new software base called Vulkan. Anton pulled the trigger. He's taken the Vulkan plunge and, you know, saw that it's totally great. And I thought, you know, okay, so in a month or two, maybe he'll have something running. No, you know, it was like, you know, five days later, it's like running. And now, you know, here's Dreg and Rez in Vulkan GPU parallelism. So that is super great too. All of these stuff, people joining the community, contributing in any way that they can think of, thank you. It's great. It makes me feel good. All right. And finally, I committed to try to write a 50,000 word novel from scratch in the month of November. We are now a third of the way more than a third of the way through the month of November. I should have a third of 50,000 words. I have nothing like that. But man, I want an experience it's being. So I wanted to write a novel called Best Effort of 50,000 word science fiction novel. How hard could it be to sign up? I signed up. I started getting these little badges for updating my counts two days in a row, three days in a row. Got to 5,000 words, got to 1667 words. That's another word supposed to do every day. Ha, I've done that maybe like two days. I don't know what it is. I keep on updating my counts no matter how many words I've gotten or not. And so does. So now I've gotten the seven days in a row and I've gotten to 10,000 words. But there was an asterisk on those 10,000 words. So this is what it looked like a week ago. The blue dotted line was the progress rate that I needed to be developing. The black, the darker solid line is my progress actually getting starting off very brave and already falling off after five days. Now we're at 12 days and it's looking like this. I managed to limp my way to 10,000 words. And then there's this strange horizontal bit where like nothing was happening. Well what was really happening there was I was taking out a whole bunch of those words that I had dictated using the Google voice typing that we looked at last week that was really it was it was garbage. It was just it was thinking about what it be nice to have text and nice to have a story that did this. It wasn't actually a story that did anything. And that was inflating the word count significantly. I gradually tore all that stuff out. And now finally we're at sort of like 11,000 or so words. Now most of which are legitimate story words. That doesn't mean they're good. It doesn't mean they'll last into the final thing if this thing really gets done. But it means they are legitimately part of the novel. So step by step. So last week my daily word count was looking like some kind of ski jump. Now it's looking like I don't know what it's looking like. Some yeah it's looking like a couple of ski jumps or something with the most recent one. It seems like the word count counter is always a day behind. My word count was minus 36. That was yesterday's result after tearing out a bunch of stuff and not having quite as much legitimate stuff to replace it with. Last week I was going to be done done on November 16th. Sorry December 16th. Now I'll be done on December 20th. Ha. I mean one of the things that he's becoming clear to me as I actually write this. I mean the thing that's been so amazing for me is normally whenever I write a scientific paper or whatever it is or you know a script for one of these rants I have a tight space limit, a page limit, a time limit whatever it is and I'm always working backwards from that where here that it's so big I need so many words so many stories that it's really you know when I can get to the point where it's just let it run let the characters talk let them do whatever they do and it's like you know it's great you know it's the flow and so I've had a few patches of that where you know in an hour or a half an hour I did a bunch of dialogue and a bunch of little pushback and all this kind of stuff whatever it is who knows I mean it had it had purposes but it wasn't supposed to build the characters plant seeds have foreboding for later in the world in the story and so forth all seem plausible but it just flowed I felt like I was cheating you know how can this be work like that what's really been hard for me so far is doing the world building backstory stuff because I've got a chapter which is kind of interleaving with sort of stuff happening directly in real time and then little bits of backstory to start introducing the technology that is driving the best effort story for me and you know every sentence of the backstory is really you know for me it's really important because it's setting up the whole set of affordances the whole set of what might happen what might not happen in the ultimate world I seriously doubt unless I get a tremendous much more flow that I will get to 50 000 words in November and even worse at the pace that this novel is unfolding I don't believe it's going to be done in 50 000 words it starts to seem more like it's going to be some kind of trilogy or who knows but I am damning the torpedoes I'm continuing to push and we will see where we end up and my goal is to have so much stuff have you know 100 pages whatever it is of semi legitimate stuff that I will be too embarrassed not to finish it we'll see how it goes that's what my internal progress bars looked like last week this is what they look like now these going down here and here was where I was tearing out a bunch of the dictation that wasn't real were more solid situation here although who knows going forward and it just happens every day oh man it's it's quite a thing hopefully I will be back in a week to have a little bit more news thank you for being here have a good week