mergeconflict298 === [00:00:00] James: Everyone before we get started this week, we need your help. Episode 300 is coming up and we are gonna be live streaming. That's right. You heard us right. Live streaming on March 20. Right around 5:00 PM. Pacific we'll put out tweets and all this stuff for you, but we're asking for something, Frank, aren't we? What are we asking for? [00:00:28] Frank: Yeah. We want to make a clip show, a clip show of you. Yes. You've heard our voices far too much. We wanna hear your voices. So we're hoping that, uh, people will, uh, do a little recording audio or video, or even both, both at the same time. That'd be clever. Mm-hmm um, maybe. Comments questions, anything? Just say hi, tell us how the weather is. Anything you got. We just wanna have a fun, interactive clip show where I don't know. We'll still talk about technology cuz how can we not talk about technology and we'll answer questions and listen to comments and not reply. Or reply depending [00:01:06] James: on our mood. Yeah. Yeah, I think so. I think that'll be accurate if you have lightning topic questions as well that we send this in, like Frank said, because it is 300, so it'll be that, but fair enough. Anything we, we're gonna go crazy on this and whatever we record, it's gonna be in the podcast, but you can join us live. We will put links to the YouTubes. Or you can find us live on our Twitter. And of course we'll put it in the show notes as well. Um, and you can then hit this, the, the kind of like save thing and you can like put it on there and you get there, but send us an email, go to merge conflict, FM, there's a contact form. You can send us a link to a video clip that you put on the internet somewhere. It doesn't, there's a website called YouTube. You can put private videos unlisted up there and you need send those. You put 'em in a one drive. You put 'em in a Dropbox, Frank. They could, they could put it literally anywhere on the internet. And you can send that to us or write in something, or just go on Twitter. You could also tweet, you could tweet us a video. You could also do that. You could literally do anything. We're gonna play back. Those videos live like Frank says it's gonna be super fun. I can't wait. It's gonna be so much hard work on me, ha to get this done, but I can't wait. [00:02:10] Frank: I was gonna say, we accept all formats because James is gonna have to deal with the file format. So please, no, I'm not gonna say I'm not gonna say that please. Uh, MP4 seven 20 P uh, 27, uh, FPS. No, I'm just kidding. Send whatever James can handle it. He's a professional. [00:02:26] James: Send it in. It is due by the 28th. Okay. So if your list listening to this on the 21st, you got a week. If you're listening to this anytime after. Do the math, you can figure it out that FM. Yeah. Do it right. Just do it right now. Literally hip stop on the podcast. You, you can listen to this podcast whenever you only have seven days to, to do this. So get in there [00:02:47] Frank: iPhones, have the voice recorder app, or can you just. Record video with an iPhone? Probably. Yeah, a camera app camera. Oh, fantastic. There's the whole camera app built into the phone. Perfect. [00:02:59] James: You get it. Tick, TikTok it to us. You can snap who you can face. What you can get all that stuff. Get it over to us. We would love that. Um, I [00:03:06] Frank: hope all right, this is gonna work. This is such an experiment, but I'm excited. It's gonna be fun. 300. If this does [00:03:11] James: work, we have another two years in opposite 400 and we can do it all again. [00:03:17] Frank: I'll hold you to that. [00:03:18] James: All right. Let's get into the podcast, Frank, but before we do it, let's thank our amazing . Now [00:03:24] Frank: I'm just kidding. Let's get, I love your, come on, do it. [00:03:27] James: I'm just kidding. No, no. All right. We are talking conspiracies, Frank ship conspiracies. The potato ships were they actually originally tennis ball container. And now they're Pringles. Nobody knows. [00:03:42] Frank: Oh. Oh. I, I didn't prepare the right notes here, man. Uh, I was doing supply chain, not, not potato supply chain. Um, are chips going up in price? Did I miss something? Is the zombie apocalypse happening again? [00:03:58] James: No, no, no, you're good. Um, I, you, one of my favorite comedians of all time, Mitch Headberg he did this bit about like tennis balls. Yeah. Tennis, tennis balls, and potatoes, and all this other anyways, go, go find Mitch. Headberg R I P we love you. Um, lots of funny stuff there, but. Supply chain. Is there issues in the supply chain? What's going on, Frank? This conspiracy, what are you? No, no killing [00:04:21] Frank: over here. This is, this is James's term. So quote, quote, conspiracy theory over here. Uh, this is some follow up on, uh, M M one ultra talk that we had. The best chip out there that I think I've decided not to purchase, but, um, uh, so I was, I had posed a question to you of why do you think they glued together two chips instead of, you know, just making a bigger chip, you know, just make it bigger. Right. And do you remember my long drawn out? Explanation. Uh, [00:04:56] James: well, I, I think it was something around technology and making things yeah. Smaller and faster mm-hmm and [00:05:04] Frank: stuff. Yeah. Well, I, that that's close enough. Thanks for listening to the episode. Um, I was making an argument and this is just. Uh, from my last experience with chips that, um, it's a clock thing. So you, you, you have to distribute the clock signal. You have to keep it in sync. So there are physical limitations of size. Well, well, sir, I have heard a better conspiracy conspiracy theory than that. And. It's not actually a conspiracy theory. It's it's economics. Um, there are, uh, economies of scale, economies of production and the dye sizes, the actual physical imaging devices they're using the hardware devices that they're using to construct these chips have kind of a sweet point in sales. And I believe what. I don't remember all the dye names. They all come in funny size names, but the rough, simple conspiracy theory, simplest answer to all this is of course it's just money. It's cheaper to make these two size things than it is to make a bigger sized thing. So perhaps not so much a technology and magical clock thing as much as, as is the boring finalities of the economics of the world. [00:06:22] James: That makes logical sense. I mean, we've seen this before, um, in, in many, many of things, uh, you know, you can attempt to create the fastest, strongest, most crazy chip, but then your prices are gonna be extremely, extremely high. This is sort of a problem. We were talking our Peton, uh, that's the shout, our patrons, but we love the patrons. We're talking about retro game councils, handheld other things like that. And there's a struggle you think in the Nintendo switch. It chose a chip set originally, and it's five years old now, um, at this point it chose a chip set that was even older than that. Right? Cause it didn't want to create a and do a top of the line chip set that was gonna drive the price up to, you know, $500 and no one was gonna buy it. However, Frank, this isn't a $500 thing they could have. Gone bananas with it and probably charged us a bunch. No, am I wrong? [00:07:15] Frank: What's the problem. It's even more complicated than that. Um, the way you make a chip these days yeah, let's go all the way into this conspiracy theory. Ready? Okay. Stop taking. Get your shovels out. Um, yeah, chips. They're complicated. We're putting a lot of transistors on there and we, we do it with flashing lights and grid patterns. It's weird and diffusion patterns and quantum stuff. It's science. We barely understand, but somehow it still works out, but we're not perfect. No human is perfect. And what you have when you have something like the. Well, let let's start with like the M one max, which is the biggest single chip thing that apple is making. It comes in variations. How many GPU cores do you want on that [00:08:08] James: chip? Yes. [00:08:11] Frank: The key to this puzzle is they're not actually different chips. Those two, one of 'em has working cores. The other does not . So what they do is they make the chip. Some of it works, some of it, doesn't what you cut the wires on the parts that don't work and you sell it at a cheaper price and saying, this is the 32 core version, not the full 48. And so even with that, that's a whole nother reason. Like it takes forever to just get that process down and set just for that one chip, because even with that beautiful design, a lot of the chips are gonna come out, not working. So now make that bigger. You know, it's, it's not two times bigger to be four times bigger. It's its area. And therefore your failure rates are four times. It's bad. and so it's just harder and harder to make high quality chips that are large. [00:09:08] James: Got it. So, so you're saying that there's not two chips, there's not two M one chips. You're saying there's one M one chip, but like tweaked, like I think I've heard this before in. Like things would be like disabled in like software or you've bought you buy like the light version of something, then it. They just kind of snip that thing or they under clock, you know what I mean? Like under clock, it basically [00:09:32] Frank: that's the classic one. Um, so you would manufacture these chips and you would run the clock as fast as you can until the chip didn't work back that off 10% and sell the chip at that speed. This is the four Giger version. What's the difference between the four gig and the two giga Hurtt version. One just happened to be in a better place, on a cleaner spot, on a piece of glass. That's all. Just better. It just got manufactured better. Uh, that art is still practiced of course, but we've kind of gotten that kind of stuff. Quality controlled pretty well. It's more like one transistor out of a billion doesn't work. And now you have to shut down that whole core because that whole core won't work. It's an important transistor, something like that. Uh, Yeah, it's interesting. But I would say that there are three M one chips right now. Uh, the original M one, the M one pro and the M one max. And then you take two M one maxes and make an M one ultra. And within each of those three, you can get ones with cores and not cores. And that's the, um, that's the part of the chip that's working and the part that's not working basically. [00:10:43] James: So you're saying the idea here is that it wasn't, that they couldn't make a bigger, better a Boulder single chip , but that's. Already, they had the manufacturing process. They had secured the supply chain, if you will. for the M one maxes, uh, and the pros and that instead of going down and having a whole new supply chain, they took a sliver of a new supply chain. Cause they still to do you some, some interweaving of them putting the two together. [00:11:18] Frank: Oh, absolutely. Absolutely. And, um, you gotta get two working ones. So if you wanna get the, I forget what it is. Is it the 64 GPS or 1 28 GPUs, but basically you have to get two perfect chips instead of the two, not quite so perfect chips. So it's even harder. It's still harder, but it turns out that's. It's still easier. and cheaper. Let's not forget the economic part. It's still cheaper than, uh, trying to build the bigger one and get the quality level quite so as high as it needs to be for them to make a profit. Got it. I love it. I, you know, it, it's, it's fun to see this stuff. Uh, I guess Intel releases this kind of information. Also. It, it's fun to see it from apple and to try to make sense of their crazy pricing. It's it's. Also interesting because they have, um, memory attached to the chip in kind of the same way. The two chips are talking to each other in the ultra, and that's yet another dye that they have, have to have yet another packaging interconnect with. And it's fun. It's fun to see all this. And then it's interest to see how supply chain . It's not just for cars. Unfortunately we aren't in the weirdest technological. Conundrum I've ever seen of what cheap chips are hard to make for some reason. Uh, it's neat to see Apple's way of getting around it. Um, the supply chain is difficult and it looks like they made some good bets. I still think we're gonna get an M two. It's not like they're never gonna make another chip again. I think they're just gonna be smart and make all their chips with interconnects and things like that in the future. So they can play game again and again. [00:13:03] James: Yeah, the, I mean, this, this does make some logistical sense too, because as these processors get. So, you know, powerful, you know, and so ultra, uh, if you will, the, the need to attempt to one up yourself with an M two chip , um, how do you do that? Right. Does well, and this is the funny part that we didn't even really talk about on, on the podcast last time, last week, was that what would an M two chip even be? I, because they have. Th four, well, they have three M one chips. So what they would've had to come out with an M one or an M two, an M two pro and M two max. You know what I mean? [00:13:43] Frank: That's literally what the rumor mill is saying right now. mm-hmm so what, okay. Mac rumors, sorry, everyone. Mac rumors are notoriously bad, but we're we're in the rumor area right now. My apologies up front rumor mail says, however that we're, we're aiming for about 40. Course, uh, 40 distinct CPU units. Mm-hmm , you know, some will be high efficiency, some will be high performance, but 40 is kind of the magic number we're trying to get up to. Will that be the M two or not? I don't know, but rumor mill definitely says apple was true to the word they're pretty much done with the, uh, No, I guess they didn't state it on stage. I was saying, I was gonna say that they were done up upgrading the M one S but chances are, they are gonna move on to the M two S uh, next year. [00:14:33] James: Yeah, the it's really funky with their high end stuff though. Right. Because right. If they started with an M one, then they did pro and max, they did ultra. And then that was over two years, pretty much year and a half, two years. So the real question becomes if they do an M two, is that the M I don't now is like the intro level or because you know how normally this works by the way. Well, it really depends. Right? Mm-hmm you, you have. You know, a new series of Intel chips and you have the I five I six or sorry, I three, I five, I seven I nine and, and those are the things. And, and if you look, probably enough at the, oh, it's [00:15:13] Frank: big grid. It's very complicated [00:15:15] James: if you, yeah, it is a big, it's even bigger than that. Intel grid. Yeah. Sorry if you boil it down to 3, 5, 7, 9. You could, you could map those to M one pro max ultra, funnily enough, in some odd way of saying here, the, the, the four lines. Now what Intel does is they release all of those at the same time. Mm-hmm and the question really becomes. Is apple supply chain, not there for them to do an M um, you know, an M two or Ooh, rumor mill will an M two. Oh my goodness. Oh my goodness. I just cracked the case are okay. Does an M two M two pro M two max M two. Ultra all released at the same time, but Frank here it is an M two is the base an M two pro is two M two S and M two. Max is four M two S and an M two ultra is eight M two S the [00:16:10] Frank: greatest getting big. I, I even got lost in yours, but I, I, I like it. I, I want, I want the, the, did you get up to 40 cores and how high did you get up onto that one? [00:16:19] James: Well, Okay. So if you, so this is, you would have to start if an M two was five cores, an M two pro would be 10 cores, an M two max would be 20 cores and an M two ultra would be 40 cores. [00:16:36] Frank: I like it. Uh, so I will say that rumor mill definitely says that there will be M two M two pro M two max, perhaps an M two ultra one thing I don't like that the rumor mill is saying, um, okay, sorry. we're doing rumor mill. We, we gotta have fun with it. Uh, iMac 27 inch, not dead. So is rumor mill. It will be back in M two form. Oddly enough. Uh, they're saying it's only gonna get up to the M two max and the ultra will always be reserved for the studio. Wouldn't that be a weird turn of events or I'm, I'm sure like a Mac pro also. Yeah, I I'm curious about that. That would make me super sad because I've really enjoyed buying ridiculously powerful IMAX in the last few years. So I would personally like to see an M two ultra, but anyway, so they will, it does seem that they are. Gonna have almost a yearly series or maybe two years, three years. And have the, uh, pro max ultra line of all these chips. Will they release 'em simultaneously? I don't know. That's that's logistically more difficult. I think like they must have built up some experience, uh, um, M one max before they had the confidence to do the M one ultra, you know, I just feel like there has to be something there, but maybe they're just good enough at chips now where they, they're not phased by any of that. [00:18:03] James: Yeah, I believe is that what they'll end up doing? Not to what I just said, but I think what they'll end up doing is have an M two and then the M two pro will be two M two S. So M two and M two pro, which is two M ones, sorry, two M two S . And then they'll be an M two max. And then. An M two ultra, which is two M two maxes. Right. So instead of creating right now, they have three chips. I think they'll, if you think of the supply chain issue that they're running into. Yeah. They can, they can bring that down even further by one or two, one or two, right? Yeah. I think they they'll have to do that. Cuz then if they want to, they can introduce the M two ultra or deluxe super turbo edition. And um, then we're just now really we're just at street fighter names at this point. Yeah. [00:18:51] Frank: What goes beyond ultra super Dulux, Zenith or back to the Zeni brand. That'd be amazing if they bought it. Um, yeah. Well, anyway, it. It's all still kind of expensive I decided not to buy one of the machines mostly because I waited in evening because I like to sleep on my decisions and then they were sold out so everyone likes the new M one op. Or at least Max's every, everyone wants a studio and by sold out there, you know, apple doesn't sell out. They just give you a couple months into the future. I decided FOMO. Wasn't a good enough reason. I'm pretty happy with my setup right now. And I still agree with everything I said on the podcast. Um, but I think I'm gonna, I'm gonna enjoy iMac 27 inch Intel for one more. Year-ish probably. [00:19:45] James: Nice. Do you think that we'll get into the room mill a little bit here too, because we've seen the chips start to be integrated into tablets and our phones. Do we? I think that, yeah. Do you think the same logic is going to start to apply to the phones like the ultra did, like, do you think we're actually ever gonna see an, uh, A max or a pro or an ultra, I mean, phone phone, do we need that? [00:20:16] Frank: Oh gosh, we don't need iPhone 14 ultra [00:20:19] James: because, you know, because Samsung does this, right. They have like a, they have a, they have a product line and then they have this ultra deluxe phone that comes out. It's like, it's, it's it sort of, of their tinkering with the next generation of stuff. And they just like jam all the stuff in there. [00:20:35] Frank: Yeah. It's yeah, it's your, what BMW seven series or whatever they used to do that on. Um, you, you had me shocked, honestly, because the M one is still so good. I'm, I'm a little surprised when I, I think I made this joke last time, even when I. Ask the, uh, iPad, what its GPU is. And it says apple M one. And I'm like, that's not a GPU, that's a chip, but it's, it's just so weird that that hardware is running it. And that hardware kind of runs fine. That said, um, I think. The hardware is far more advanced than the operating system. At this point. I love iPad OS I love iOS, but, um, we were joking. I was joking. Wouldn't it be nice if there was just like a second partition on this thing that contained Mac OS. So when I plugged it into a monitor, my iPad just became a Mac. Cause I would love to buy, buy that device. I don't think apple will ever sell it, but gosh, I would love that device. So will we put. And M one max and an iPad, you know, James, we probably will, and I'll probably buy one and I'll be like, gosh, we really need to make I iOS better because so much power is being unused here. I don't know. I, we are at a weird stage. These M ones just knocked it so far out of the pie arc that it seems like the software has to catch up. [00:22:01] James: Yeah, I'm really, really curious what they'll do because I was looking at the Samsung website and they have the S 22 S 22 plus, and the S 22 ultra. And that leaves apple in not getting more money. And, you know, I mean, okay. I mean, that's, that's truly what it is. Right. And apple likes money. It's a good thing to get other companies like money too, not just apple, but mm-hmm, , you know, when you look at an iPhone 13 pro. It starts a thousand dollars, right? Yeah. Yeah. Now what if they could have a phone that started at $1,400? That seems pretty good. [00:22:36] Frank: Uh, I see what you did there. Um, mm-hmm but there's still a processor upgrade you can do on that phone even right on the thousand dollars one. I think there's still a processor upgrade there. Well, yeah. Anyway, um, I actually do want to add to what I was saying, because, uh, I'm gonna play devil's advocate with myself here. I just converted a neural network from Python over to the iPad on my Twitch stream. Come watch my TWI stream and it, it works great. But it's slow. really slow, very slow. Now, unfortunately, I'm comparing it to an RTX 30, 90, which is really not a nice comparison to do with an iPad, but it's even slower than it should be. That's probably a good chunk of that is my fault. I have to figure out how to make my code more efficient and things like that. But the same time. I can write really inefficient Python code and a Linux box with a 30 90 can run it, you know, 50 times faster than what I'm getting on the iPad right now. So not saying that like, again, it's a software issue, you know, the power is there. I'll also comment on, there is a little bit of tricking, more, more conspiracy, the theory stuff here. So apple had a cool slide where they said that the M one ultra had the performance of an RTX 30, 90, which is exactly what I was just talking about. But. Star star star asterisk. That's about video and coding performance. And it has says nothing about video game performance or neural network performance, which are the two things you do with the, the video card. Uh, I'm not, I'm not a big video in coder. Maybe you are, you, you live the YouTube lifestyle. Are, are you excited? 90 levels of video and coding speed. [00:24:21] James: Yeah. I think if my, if the software I use supported it at all, my software Camtasia does everything on the CPU, so. Sure. [00:24:29] Frank: So that's great. It, it is a general purpose computing unit might as well. Yes. Use it. Yeah. But [00:24:34] James: right there, ignore that GPU. It's not doing anything. Just let it hang out. It's it's let it, let it cool down. It didn't do anything. Hey, why should do anything at night? Now? Things like DaVinci resolve, for example, will take advantage of that stuff. And that being said. You know, if you are video editing and publishing stuff, you think of like channel nine, right? Mm-hmm , they're publishing doing tons of encoding all day. Imagine, you know, imagine you take like a conference, like do net com or build, you gotta pump out these videos and the faster you can do it, the faster you can upload it, the faster you can process them across the board. It's it's from start to finish ish. Because even once you upload it to YouTube, it's gotta process again. It's gotta it's these videos are being like 5 billion and then Twitter [00:25:15] Frank: rejects it for no reason. Yeah. Yeah. Exactly. So, so I'm just saying it's just a slightly, okay. I get it. It's a use case and video editor, people do do max. They get it. I get [00:25:27] James: it. Blah, blah, but not, but not [00:25:29] Frank: for most people. Right? It's not the one you do the chart about, you know, you do, you do, how fast can it play? The. Whatever games, kids play. I'm trying to name a modern game and I can't even name a modern game. Um, call it duty 3000. I don't know. about Fortnite, Fortnite, Fortnite. The kids play Fortnite, right? Yeah, maybe. Yeah. Show me 120. Frames per second on fortnight at, you know, 4k or in my case, show me it train picks to picks in 30 minutes, you know, [00:26:01] James: but, but why wouldn't it? Why wouldn't it do, uh, I mean, why, why wouldn't it [00:26:07] Frank: do do this though? Well, there are some, uh, there are a few fundamental things, so, uh, Video and coding itself is usually dedicated hardware. So it's a little bit of a cheat cause you're, then you're not talking about the standard kind of GPU threads, the way you normally program a GPU, the way you program a GPU is you write a little kernel function in a C like language. And that thing gets run on 10,000 threads all at once. You know, it's, it's a massive, um, Multi threading architecture. That's what a GPU is. Video en coding is like a little magical piece of hardware embedded in all of that. That is able to encode video frames very quickly. That's why I think it's a little bit of a cheat, but we can also talk about, uh, different. Class is a hardware. I brought this up on Twitter, the 30, 90, um, a, a desktop class. And I'm talking about desktop class GPU. Things are a little bit different on the mobile class ones. The desktop class ones always do 32 bit floating point operations, you know, That's it's our favorite. It's always been a little frustrating to me because sometimes I wanna do 64 bit. But as you might remember from your shading shader language days, you can't do 64 bit math they hate it. No, 64 bit math. So everything's 32 bit math. Uh, but one thing that the M ones inherit from, uh, basically starting out as mobile pro processors is they're able to do 16 bit floating point math, and they actually can get quite a bit of performance. Bump over 32 bit, because you're literally doing half as much work, you know? So it, it, it, would've been nice to see a real comparison from apple to see like even cheat a little say like, Hey, we ran this, uh, neural network at 16 bit instead of 32 bit, but check it out. It actually did beat. The 30, 90, or it was competitive with the 30 90, you know, I, I don't, I'm not even saying I want an apple, but apples. I just want something a little better than en coding a video frame because yeah, I guess there's people out there that code video frames, but I just don't care [00:28:23] James: no. Do you think that they in like, In this conspiracy theory that you had Frank of apple, you know, holding back in some way, do you think that they tried to? Yeah. [00:28:43] Frank: Okay. Well, okay. Sorry. In the conspiracy theory, I, I have no facts, but they must have, because that would've been a huge thing to brag about is this that it's running a standard neural network at the speed of a 30 90 that we, how, [00:28:57] James: how can this okay. How can this tiny chip bit tiny little thing? How can it have the same performance as this huge wain of GPU that, how is it possible, Frank? [00:29:13] Frank: Well, step one, take your GPU apart and notice how many chips it actually is. It's a big CPU chip and then another big chip to process the PCI X 16 pipe, and then a bunch of other little Ram chips. Mm, the M one does not need the PCI chip. It's already built into it. It doesn't need the little Ram chips they're built on to the, well, it's a separate dye, but they're in the same package. So they're all right next, close to each other. And then, uh, what, what else is a giant graphics card is a giant heat sink. Well, you know what? The max studio is, it's a giant heat sink so you can think of the max studio as a GPU. Just sitting there that also has, you know, um, what an SD card slot [00:30:03] James: it's got some Thunderball [00:30:05] Frank: fors Frank, so do GPUs [00:30:09] James: I never really thought about opening up a GP. I mean, cuz they're so big, but it does make sense that because they're generic and have to fit any, they basic we have to be a small computer. Yeah. All in one. [00:30:20] Frank: So they have to build a whole computer there and that's, what's taking up all the space and they might even have, you know, I don't wanna talk, act like I know everything about GPUs. They might have multi chip ones and things like that, but all the fundamentals of what I just said were correct. Um, there's not much there. It's a computer interfacing to another computer. [00:30:42] James: Interesting. [00:30:43] Frank: A very expensive. Yeah. So what, what it needs to come down to is everyone's just gotta get their hands on these machines and we'll get some actual benchmarks doing some actual workloads. We'll see how many Bitcoins it can mine. We'll see how many neural networks it can train. We'll see how fast it can run Fortnite, and then we'll get some hard numbers. I'm still, I'm still very hopeful. I don't wanna make any of this conspiracy theory sound like, I think Apple's lying or anything. These are obviously amazing chips, but I'm excited to see what their real world performance is like. [00:31:17] James: I'm a big fan of like when the Verger and gadget do these breakdowns of, we opened a browser for . Yeah. Especially on laptops. Like we played back YouTube for 48 hours. You know what I mean? Like it's mostly what people are doing is I was gonna joke early on. We were saying like, how fast could it do? Like render XYZ. I. How fast can it open wall street, journal.com? Yeah. [00:31:44] Frank: can, can you load all these JPEGs and display them in finder? How fast can finders show a directory of JPEGs [00:31:52] James: question? Yeah. Valid question. Can the operating system it's the, the core OS take advantage, uh, of the cuz you know what I'm living and breathing every single day is like I click on the start button and things come up and I want, I want everything just to be faster. So I guess, yeah. The core operating system also has to be optimized to take advantage of these optimizations in the chip. [00:32:15] Frank: Yeah, I, I would say, but I think the beauty of the design is its simplicity. Um, like it has that unified memory architecture and the M one ultra is actually running its memory faster than my, um, Invidia RTX. I forget exactly what speed the RTX is, but, um, the M one ultra is able to get 800 gigabytes per second of bandwidth. And that's faster than G DDR, six Ram at this point. So they have fast memory, you know, UN Unix scales. They don't have to change the kernel that much. It's a very simple architecture, a. All that memory, you know, when I'm, whenever you're doing, um, GPU stuff, if you've done this, using a low level thing. You know, you have to synchronize buffers from time to time. Like if I wanna load a character texture, first, I loaded it into CPU memory. And then I have to synchronize that up to GPU memory. And then if I modify the GPU memory and want to actually save it to disc or something, I have to download it back to the CPU and then do that fidly stuff. The beautiful thing about a unified memory architecture is you don't have to do that. The GPU can access the exact same memory that the CPU can access. Nice. Simples makes sense. Yeah. Know, iOS or mobile programming, you know, it was good. I'm glad mobile programming has influenced the desktop market. [00:33:42] James: I agree. I agree with that. Uh, anything else you wanna cover on this conspiracy theory podcast? [00:33:47] Frank: Uh, I'm, I'm sure we'll, we'll have more to add to it in the future when the M four is out than we'll really know what's going on. [00:33:55] James: um, I wanna say, uh, a few things before we go. So obviously everyone heard about our amazing, uh, upcoming 300th episode. Please go to merchant with advance. Send in some stuff. We'd super appreciate it. If you're interested in supporting the, the show even further, we have a Patreon, you get bonus episodes. Every single week. We just recorded one at Superfund. We talked about retro portable, handheld councils, go check that out. You get the whole back episode, you get exclusive RSS feed, all that good stuff. Um, we love also when people just, you know, comment on the podcast, hit us up on Twitter or anything like out. We have a discord button, uh, on there as well. So I saw a community there. Um, and then, yeah, I, I just wanna also. Give a shout out to, you know, Frank live streams over on his Twitch. Uh, I live stream over on my YouTubes. I also put up videos on my YouTube as well. And I do wanna, I do wanna, we did an entire episode, I think on source generators. Do we do want OnSource generators? [00:34:49] Frank: Um, yeah, cause that's when we unfortunately use the term hack a little too much. [00:34:55] James: Okay. I just did a video on MVVM source generators and the do beauty toolkit. I saw that and. We need to do a whole episode on it, cuz okay. Source generators have officially blown my mind. I'm all in 100%. I love everything that's happening here. It's amazing. Okay. Um, and this library has changed my life and you just need to find out and, and there's this new thing. Called incremental source generators, but we'll get to that another day. If you're interested in any topic, send it into the show. We'd appreciate that. Let Frank what you are gonna do. Are you gonna buy a new Mac? Are you in the conspiracy train like he is, or what's going on over there? I know I'm not gonna buy anything cuz I don't have any money to buy any of, of these devices. Cause I just bought a new M one MacBook area year ago and it's still. Awesome. Perfectly fine. it's perfectly fine. By the way, those chips, all four of them. They're all fine. You, any of them will work for you any day? [00:35:50] Frank: That's the joke. Yeah. Get the cheap one. It's fine too. [00:35:54] James: it's completely fine. I kind of want the pro, but also the normal ones just fine. Um, I do, I really like my MacBook air is [00:36:02] Frank: great. It's it's a wonderful machine. It's very portable. It's always snappy. It loads the wall street journal very quickly. Yep. Very quickly. [00:36:11] James: That's all you need. All right. Thanks everyone. For tuning into this week's merge conflict. Hit us up at merge conflict FM, but until next time. I'm James Monte Magno [00:36:20] Frank: and I'm Frank Krueger. Thanks for listening. Peace.