00:00:00 ◼ ► I feel like I should acknowledge that if something weird happens, it could be my fault because I have a brand new computer that I'm recording on.
00:00:08 ◼ ► Marco, it could be your fault because you have a brand new computer that you're recording on, and it could actually even be John's fault because his computer's so old, busted, ancient, and slow that it could be his fault after all. We don't know. Nobody knows.
00:00:26 ◼ ► But why is he already worried about your computers? There's nothing weird about your computers that, I mean, yes, they are new, but like, do you expect there to be problems? They're not too different from the computers you're replacing them.
00:00:36 ◼ ► We're also jumping ahead, and I need to put the kibosh on this until we get there, but uh-
00:00:51 ◼ ► By a week! It's like I'm the older twin because I'm five minutes older. Anyway, but no, what I got on the call this evening, first thing Marco said to me was, "Oh, you're on the wrong mic." Which was true, for the record.
00:01:03 ◼ ► I mean, that never happened with the old computers, though. You're right, this is a big change.
00:01:28 ◼ ► And let me remind you, if you are an ATP member, and I assure you, almost all of the cool kids are. There's only a few cool kids left that aren't.
00:01:42 ◼ ► I think actually by joining you become cool, and we're not going to do any further research on that.
00:01:54 ◼ ► This is one of those times that you get 15% off anything you buy from the ATP store in this time limited sale.
00:02:02 ◼ ► You can go into your member dashboard or whatever we're calling it to get your bespoke discount code that you use at Cotton Bureau.
00:02:26 ◼ ► The headlining stuff that you need to know is that there's a new shirt where we try to do the logo out of pixels,
00:02:37 ◼ ► And then we have a space black joke shirt where the logo is not quite black, just like the new MacBooks.
00:02:46 ◼ ► We have the ATP 6 color shirt where the shirts come in lots of different colors, but the logo is just white on all of them.
00:02:52 ◼ ► That's the main colorful one. Then we have the plain ATP one, the hoodie, the polo, a new mug and pint glasses.
00:02:58 ◼ ► And the only thing I want to add that we didn't mention last week is that the mugs and the pint glasses, unlike the other things,
00:03:03 ◼ ► like we have to guess how many we're going to sell these and we order them up front in a certain amount,
00:03:08 ◼ ► which means that they may run out. So if you want a mug or if you want a pint glass, don't wait, because they may run out.
00:03:15 ◼ ► And if they run out, that's it. We can't get any more in time for anything close to the holidays.
00:03:34 ◼ ► Either way, we already know the power of the iPad. It shatters windshields. It destroys mugs.
00:03:41 ◼ ► In all fairness, the mug damage on the drop incident was only like a little chip off the side. The mug was mostly intact.
00:03:54 ◼ ► Oh yeah, for sure. Well, not wielded by Casey, though. We don't know what the velocities involved here.
00:03:59 ◼ ► Indeed. But yeah, it's just got a patina now, the mug. But yes, we are saying this in part for our own selfish purposes because we want you to buy and buy now.
00:04:10 ◼ ► I don't remember the quantities, and honestly I don't know if we'd tell you anyway, but we have limited quantities of the mug and the glass.
00:04:25 ◼ ► Again, I worry about people who want this, but it's a perfect gift for someone to get you.
00:04:29 ◼ ► If you're hard to shop for and people don't know what to get you and you're afraid they're going to get you socks or something, send them this link.
00:04:35 ◼ ► Do not be shy. You can't beat around the bush. You can't lay hands. You have to say, "Look, this sale ends November 26th, and I really want one of these things."
00:04:45 ◼ ► Say, "I know this stuff looks dumb and expensive, but I like this podcast, and this would be a good gift for me."
00:04:50 ◼ ► Tell somebody who actually wants to get you a gift, tell them that and send them the link before the sale ends.
00:05:07 ◼ ► And I was like, "Oh, I meant to do that." Because we did add sweatshirts to a lot of them.
00:05:14 ◼ ► It's the same product. The product will be called ATP Space Black, and that comes in short sleeve t-shirt.
00:05:23 ◼ ► Not every variety is in everything, but think, "Oh, I don't want a t-shirt, so I don't want to click on that."
00:05:27 ◼ ► Click through on the design that you like and see if it's offered in sweatshirt, long sleeve t-shirt, tank top, whatever you want.
00:05:33 ◼ ► And we also added youth and kid sizes. We can't do baby sizes for these, because whatever the printing process is,
00:05:42 ◼ ► The on-demand ones, I think they can do on babies, because that's a different printing process where it's just like,
00:05:46 ◼ ► I don't know anything about how it works, but I imagine in my head that it's more like an inkjet printer that can print anything,
00:05:51 ◼ ► whereas these are like, I don't know, some kind of pattern that they run ink through or something.
00:05:56 ◼ ► Anyway, the point is, no inference for these, but we do have youth and kid sizes, which are smaller than adult sizes,
00:06:02 ◼ ► and we do have long sleeve t-shirts on several of them and sweatshirts on several of them.
00:06:09 ◼ ► Yep, and as a final note on that, the youth sizes were a request from my beloved wife Erin, who has said,
00:06:16 ◼ ► and I'm pretty damn sure she said it jokingly, but she said, you know, when the three of you are just swimming Scrooge McDuck-style
00:06:22 ◼ ► in all of your merch proceeds, because everyone has gotten youth-sized tees for their children or friends' children or what have you,
00:06:30 ◼ ► you know, I should thank Erin. So if you would like her to be able to rub something in my face, probably until the end of time,
00:06:37 ◼ ► go ahead and buy your friends' kids, your kids, your random kids, throw them in the schools like a t-shirt cannon at a stadium.
00:06:49 ◼ ► I believe this is the first time we've offered anything in youth sizes, so check it out, atp.fm/store.
00:06:54 ◼ ► We also have something else to talk about. We have a new member special, and I don't know, Marco, would you like to take a stab at describing this,
00:07:04 ◼ ► So, you know, as we've discussed in the show a little bit here and there, in recent months, John has been working on the CMS,
00:07:11 ◼ ► and the CMS is something that I wrote in PHP originally, and so John has been learning, well, reluctantly, plowing through PHP.
00:07:22 ◼ ► And we thought, what is better content than having John talk about his learning process of learning not only PHP,
00:07:31 ◼ ► but learning my totally undocumented framework that is a custom framework that no one else uses, basically,
00:07:39 ◼ ► and also, you know, the whole codebase to the CMS, which, you know, let's be honest, it was not, like, it was never my top priority.
00:07:48 ◼ ► So, John had to wade through that, and we made him talk about it and recorded it, and I think it's pretty good.
00:07:57 ◼ ► Yeah, the title of it is the more aspirational goal, which is the title, this format title for our specials I'm rapidly regretting,
00:08:07 ◼ ► but now it's kind of becoming a joke, which is something colon and then something else, right?
00:08:12 ◼ ► So, we do ATP top four for the ones that are, like, top four format that make sense, ATP eats for the ones where we eat gross foods,
00:08:17 ◼ ► ATP movie club or whatever, we watch movies. This one is ATP dev colon other people's code, and I hope to make it more broadly applicable to,
00:08:33 ◼ ► And obviously, the concrete example is I'm looking at other people's code, it's more gross CMS stuff or whatever,
00:08:40 ◼ ► I tried to lay off complaining about PHP too much because, like, yeah, whatever, we get it, it's gross.
00:08:44 ◼ ► We had many shows where we talked about that, but the more applicable thing is, like, what is it like, not just for programming, but any other thing,
00:08:52 ◼ ► what is it like when you don't get to define everything from the ground up, and what is it like when you do?
00:08:57 ◼ ► Like, with Marco's perspective, he did get to write this thing from the ground up, but that means it's a weird thing that only he knows
00:09:01 ◼ ► and that he didn't document that he's not interested in anymore, but it still exists and it still runs.
00:09:08 ◼ ► As we said on this member special, the next one, the one for December, will probably not be tech-related. We'll see what that ends up being.
00:09:15 ◼ ► Yep, and remember, atp.fm/join, you can join for as little as a month, you can join for a year, you can, you know, grab all the member specials,
00:09:24 ◼ ► get your discount, and then cancel, and because we respect you, that's easy to do, but because you love us, you won't do that.
00:09:34 ◼ ► Happy Christmas, Hanukkah, birthday, whatever, to me. I'm just going to stick with it, and that's what we recommend.
00:09:53 ◼ ► As someone that uses his Mac completely hands-free, Face ID on the studio display and MacBooks would be a game changer.
00:09:59 ◼ ► I am paralyzed from the shoulders down. Having Face ID in place when Apple's accessibility feature, voice control, came about was truly a life-changing moment.
00:10:06 ◼ ► It gave me complete independence, being able to operate my phone and iPad by voice completely by myself.
00:10:12 ◼ ► It would be nice to have the same independence with my MacBook and Macs connected to the studio display."
00:10:16 ◼ ► You know, I've been trying, or I've been telling myself I need to try, to work on being more aware of the non-vanilla case for everything.
00:10:24 ◼ ► That's oftentimes thinking about accessibility, but not exclusively, and this was just a complete oversight on my part, and I thought Todd made the point very succinctly,
00:10:32 ◼ ► "Yeah, we should think about that, and that would be real." I mean, it would be nice for all of us, as with most accessibility things, but it would be doubly nice for those of us who have that kind of need.
00:10:43 ◼ ► And especially the studio display, because you don't have problems with, like, "Oh, it's not thick enough." You know what I mean?
00:10:47 ◼ ► That's a place where I feel like you can solve these problems, and apparently they're not afraid to add hundreds of dollars for accessories like the stand, so I just throw a face at you in there.
00:10:56 ◼ ► Yeah, something tells me that they have the margin in that product to add the one additional sensor, because they already have a camera!
00:11:04 ◼ ► And you're right, they have room at the wazoo in that product, so I'm pretty sure they could add the IR blaster thing, and that would be fine.
00:11:21 ◼ ► It's like, okay, now almost every iPhone model, including almost every one of the less expensive ones, has Face ID now.
00:11:29 ◼ ► And it's probably a very short matter of time before they all have them. Once they do the next iPhone SE, it might be all Face ID at that point.
00:11:36 ◼ ► So, it's no longer a cost thing. You look at the studio display, and there's a few misses in that product, so maybe it just didn't make it in in time or whatever.
00:11:48 ◼ ► But hopefully when they update their monitors, which they do very frequently, I'm sure it's just a small matter of time.
00:11:54 ◼ ► Once a generation, they update the monitors. And we mean human generations, not hardware generations.
00:12:00 ◼ ► I don't want to think about it. Alright, Jamie West writes, "Oily noses are not just useful for testing for fingerprints."
00:12:07 ◼ ► I can't find the original source now, but at some point in the past, BBC TV technicians used vision mixers which had faders that would sometimes get a bit sticky, causing the crossfades to visibly not be smooth.
00:12:18 ◼ ► As everything was basically still done live back then, you couldn't exactly stop to perform some quick maintenance.
00:12:22 ◼ ► But often a quick rub on the side of the operator's nose and an application of nose grease to the fader would be enough to get it unstuck and moving smoothly again.
00:12:33 ◼ ► This sounds a lot like something that a British person would make up to tell Americans to make them see how gullible they are.
00:12:41 ◼ ► How do you get the grease down into the part where the friction surfaces meet? Like, you could put grease on the top of it, but that doesn't help you. Anyway, I hope this isn't true.
00:12:49 ◼ ► Alright, and then we have microwave follow-up because how could we not? This is the Accidental Appliance podcast.
00:12:56 ◼ ► Nate writes The Modern Knob. That was the whole tweet, or toot, I guess. And it's a photograph of a LG microwave.
00:13:07 ◼ ► It has a display in the upper right-hand corner. It has what appears to be a horizontal slider labeled "time" with a minus on one side, a plus on the other.
00:13:19 ◼ ► I saw this and I thought of car controls and I was thinking that's like a touch slider. You know how they do those things? Instead of having a volume knob in a car, they'll have this little tiny region with some gloss black plastic where you swipe your finger to change the volume.
00:13:31 ◼ ► It's like there was a contest. You've seen these things. There's a contest for how to make the worst kind of progress bar, I think, with a dialog box. Have you seen those?
00:13:41 ◼ ► This is like that, but in real life. Someone's like, "How can we make the worst, most annoying interface to a microwave?" And it's like, "How about just a knob? No, too simple. How about a plus and a minus button?"
00:13:53 ◼ ► And I really hope you can swipe between them because that would just make it so much worse. This is terrible.
00:13:57 ◼ ► Yeah, it's so bad. And yet, I feel like, for me, I've found one worse. And Keith Bradenham writes, "I love our microwave for its rejection of any complexity. Just four buttons. Microwave, defrost, start, and stop/cancel." I'm not kidding, folks. We will put a link in the show notes to a photograph of this microwave.
00:14:15 ◼ ► It's a display. Microwave, defrost, stop/start, and cancel. Or, excuse me, stop, cancel, and start.
00:14:23 ◼ ► I don't know. But anyway, Keith continues, "You repeatedly press the microwave or defrost buttons to set the cooking time. It has an LED display, but I love how it doesn't even bother to include a clock." And sure enough, in this photo...
00:14:35 ◼ ► Yeah, honestly, I wish fewer appliances would have clocks on them because that's fewer things that you have to reset every time they get out of sync or daily saving time happens.
00:14:43 ◼ ► Or they would just set themselves based on the weird AM radio thing. I mean, everything has internet access now. But yeah, microwave interfaces are the opposite of what Twitter apps were. They're a UI playground for people to do terrible things.
00:14:55 ◼ ► I will say, though, the feedback suggests that knob-based microwaves are actually extremely common everywhere except the US.
00:15:04 ◼ ► Yeah, but did you see the European ones that they sent pictures of? They look a lot like the Kitchen Timer, like mechanical ticky one, and that's not really what you want. Like, I know that's also a knob, but it's the kind of thing where to get it back to zero, you can push it and it goes, and it's just not great.
00:15:20 ◼ ► Alright, now we move into the scanning photos corner. Ryan writes, "I'm the crazy one who decided to scan my family photos, crop them, and apply dates. I wrote a post on how I went about it, and we'll put a link in the show notes. It took a couple of years, but it's been incredible being able to go back and find photos by dates, get face detection photos, etc."
00:15:38 ◼ ► Then friend of the show Todd Vaziri writes, "One of our pandemic projects was to do something about the thousands of prints and film negatives that we had in shoeboxes in our basement, and finally get them digitized. We ultimately used a service called Scan Cafe and were extremely happy with the results.
00:15:53 ◼ ► The print scans were far better than anything we could have accomplished at home, and the negative scans were spectacular. However, one thing to keep in mind before anyone embarks on a project like this is to plan and organize before you ship off your prints and negatives, and be prepared for tons of work after you get the scans.
00:16:07 ◼ ► Create logical batches of prints and negatives in chronological order as best you can, because the service will honor your naming conventions. Once you get your scans, you could simply import them all into Apple Photos and be done.
00:16:16 ◼ ► But taking that extra step of assigning them dates at a minimum, and at maximum with geotags and captions, will take hours and hours and hours of work. Be prepared.
00:16:25 ◼ ► In all, we organized scan categories and dated over 3,500 photos that span over three decades. I'm so glad we took the time to do so, because now we have an incredible, organized digital photo library that are backed up in triplicate, and we never have to worry about losing our physical prints or negatives ever again."
00:16:40 ◼ ► "I forgot to mention this with my scanning thing, although it doesn't take, depending on how much work you put into it, it doesn't take that much time, certainly not as much time as scanning with my stupid slow scanner, but the problem that you will encounter is, when was this taken?
00:16:52 ◼ ► If you're lucky, you have that obnoxious, burned in, like little seven segment display date on the picture, and if you're extra lucky, that date is right, because speaking of setting your clocks on things, not everyone was good about setting the clocks on their like instant cameras or whatever, not instant cameras, but their film cameras, so very often that date can be wrong.
00:17:09 ◼ ► But the older the pictures get, like before the advent of that kind of date stamping, like in my childhood, those things didn't really exist, like what year was this? How old do I look here? Am I seven? Am I eight? Am I four?
00:17:21 ◼ ► Like if you're lucky, you can see, okay, this is when we lived in that house, in this town, and then you get into stuff from like my parents' childhood. They don't know when it is. I tell my parents, when was this taken?
00:17:32 ◼ ► They're like, I don't know, sometime in this decade range? Oh, it's brutal, like, and the sooner you do it, like the more relatives you have who are alive, who could perhaps date them. Like I have super old pictures that I got from like, my uncle scanned a lot of stuff from his father, right?
00:17:46 ◼ ► So they're super old black and white pictures, and I'm lucky if I can get a decade. I do surveys with relatives like, which decade do we think this is? 40s or 50s?
00:17:55 ◼ ► Forget about geotagging with location. And they're like, oh, by the way, who's in this picture? There's Uncle Bob. I have no idea who these two people are. Does anybody know?
00:18:04 ◼ ► It gets harder the longer you wait. And even for your own pictures, even for your own pictures, like I can tell we were in college, but what year were we in college, right?
00:18:12 ◼ ► It's really difficult to do. And so what you kind of have to do, and I really, really wish photos had this feature, although it's not on my very long list of things that I want them to add immediately, but the concept and Boolean logic of don't cares,
00:18:24 ◼ ► where I want to be able to put in a date and I can say, look, all I got is the year. This is 1962. I got nothing else. But you can't do that in photos.
00:18:33 ◼ ► Right. And so you have to come to the convention where I just do January 1st at like 12 a.m. But what if something legitimately happened on January 1st?
00:18:45 ◼ ► Yeah. And then you'll find one. Then you'll find one that you have a date for. But now it's all messed up because you know that one came between these ones, but you shove them all up into January. And even though you don't know the months for those, you know, this one was somewhere in the middle of them.
00:18:56 ◼ ► It's very difficult. Even if you just know it's the 60s, you can't put like if you put 1960 later, you're going to remember that. Oh, that wasn't really 1960. I just put 1960 because I didn't know which year in the 60s it was.
00:19:06 ◼ ► So photo apps aren't good at that. So do the best you can. That's why photo albums where someone wrote in like, you know, grandmotherly cursive right next to them with dates and stuff. That is a godsend.
00:19:16 ◼ ► But do the best you can and realize that that is going to be even if you get them scanned, that's going to be the majority of your work, especially if you're not organized, like Todd says, when you send them off, because they'll send them back to you.
00:19:28 ◼ ► And sometimes there is no discernible order. And sometimes the order is just whatever you sent them in. And who knows if that order is right.
00:19:39 ◼ ► Nathaniel Velez writes, "On the topic of things Apple still needs to improve on its laptops, at the top of my list is the durability of the plastic used for the keyboards.
00:19:47 ◼ ► They become shiny with just a few days of use because of your fingertips being able to sand and polish the plastic.
00:19:57 ◼ ► But since getting a slightly fancier mechanical keyboard that doesn't have this issue, I learned that if Apple used different materials, we would not have to live with laptop keyboards that look disgusting.
00:20:05 ◼ ► I'm always curious why people don't complain about this more. I was happy to see at least one review from CNET that mentioned the shiny, disgusting keyboard problem.
00:20:16 ◼ ► I definitely noticed that my finger grease gets on the keyboard and I don't ever feel like I've worn a keyboard down to the point that the paint on it is being lost.
00:20:28 ◼ ► Maybe it's because, although I do not change keyboards and computers like Marco changes his keyboards and computers, which is like most change their underwear.
00:20:38 ◼ ► Well, I have many copies of the same one, which appears to now be fully fully for real discontinued. They're all gone from Amazon.
00:20:48 ◼ ► Me with my cheese graters and you with your keyboards are just running through the back catalog now.
00:20:53 ◼ ► At some point in the next 15 years, I might run out of them and have to switch keyboards.
00:20:58 ◼ ► Anyways, I find that the finger grease thing is a thing. For me, every month or so, and there's no real schedule to it, it's just when I notice my keyboard is getting gross.
00:21:09 ◼ ► I'll take one of the Mr. Clean Magic Erasers. I think there is a proprietary eponym, the Xerox name.
00:21:16 ◼ ► Yes, what you're looking for is melamine foam. That's the actual thing. It's like a very, very, very fine abrasive.
00:21:22 ◼ ► You've got to be careful because some of those are also infused with some gunk. Not gunk, but like gunk.
00:21:32 ◼ ► Yeah, if you're buying a Magic Eraser, make sure it's like the plain, like unscented, unflavored.
00:21:39 ◼ ► Yeah, because one time I accidentally bought a little box of the ones that use them and they foam up with soapy something.
00:21:47 ◼ ► And I don't know if I would even want that grit to go there. I know we don't have butterfly keyboards anymore, but that stuff does kind of like flake a little bit when you're using it.
00:21:55 ◼ ► But anyway, Casey, why are you taking a Magic Eraser? What are you trying to remove? Like the paint?
00:22:00 ◼ ► The paint, yeah. No, to get the keyboard or the finger grease off the keyboard. I've been doing this for years and years across many, many, many different laptops.
00:22:10 ◼ ► It's not abrasive. It's just, it works well. I don't know. It's just what I've found that works the best. I'm sure that's not the only way to do it. Maybe even just a damp cloth would be enough.
00:22:22 ◼ ► Yeah, I would suggest you maybe not do that and instead use the very slightest of dampness on the thing.
00:22:30 ◼ ► Because you just want to like carry away. The finger grease will come off pretty easily. You just got to get a cleanish thing there to... Anyway.
00:22:37 ◼ ► Don't... Doing the keyboard... This annoys me. We've talked about this before. It annoys me that iPads, I guess it's annoying too, but especially Macs, are so not waterproof like the phones are.
00:22:48 ◼ ► And for reasons that make sense. Like there are reasons, but maybe we'll get this if we actually get to this topic.
00:22:53 ◼ ► Boy, it would be great if they did that better. Because it makes cleaning your keyboard on your MacBook, and your MacBook especially, so fraught.
00:23:00 ◼ ► If you're like, "Oh, I have a vaguely damp cloth and I'm just going to use this to do the keys." And you wad it up a little bit and you press it and a little drop of water comes out and it slips between the key and the thing.
00:23:10 ◼ ► You're like, "Noooo!" You do not want water to go inside that thing at all. You know, you can clean it when it's upside down and it's annoying.
00:23:16 ◼ ► You got to buy those apps that disable the keyboard so the thing doesn't turn back on. It's too fraught.
00:23:21 ◼ ► But on the topic, Nathaniel's question about why don't people complain about the keyboard being gross. I think we've talked about this before.
00:23:27 ◼ ► Some people have fingers that excrete something that destroys keyboards. And it is not a personal failing. It does not show that they are haunted by demons. It is just a thing.
00:23:39 ◼ ► There is no value judgement here. But it's a thing that happens. We all know people, if you've worked in offices or whatever, they'll have a standard issue keyboard for the whole company or everyone will have a MacBook Pro.
00:23:50 ◼ ► And one person will wear the letters off their keys within two weeks. Another person will use a keyboard for seven years and be typing madly on it all the time and it will be fine.
00:23:58 ◼ ► Setting aside finger grease, which is an aesthetic thing of how greasy it looks, it's a separate issue. I'm saying the people who have like... I don't actually know what it is. Is it acidic? Is it they have calluses on their hands?
00:24:08 ◼ ► I don't know what it is, but something about their hands just really wears down the keycaps.
00:24:14 ◼ ► In terms of using different materials, I think matte versus glossy makes a difference. Kind of like it does for the fingerprints showing up on the glossy mirrored Apple logo on the back of the laptops versus the anodized matte bumpy thing or whatever.
00:24:29 ◼ ► So that is a factor. So part of that is the Johnny Ive hangover of like, "Boy, the keyboards look nice when no one has ever touched them and they're glossy black." But then, of course, you touch it.
00:24:37 ◼ ► I think the thing with piano black interiors on cars, which they're still making this year, they're still making them. Please stop, people. It scratches so easily, it looks horrendous.
00:24:50 ◼ ► Do you believe they still do that? Like every car review for the past five years has some section where they say, "Oh," and they use piano black on the interior and they roll their eyes with a knowing look to the audience.
00:25:02 ◼ ► Of course, we all know this is a bad idea, but the car makers don't think it's a bad idea probably because it looks good in the showroom and they don't care what it looks like after they sell it to you. Anyway.
00:25:15 ◼ ► Yeah, it's very similar to the jet black iPhone 7 finish. Like, yeah, it scratches instantly, but even with some scratches on it, it still looks really nice.
00:25:24 ◼ ► But also, the dust shows up on it a lot, especially in a car injury if you ever carry pets or anything or if you ever use a tissue and blow your nose and the tissue dust goes out of the... It's not a good idea.
00:25:34 ◼ ► So yeah, I think Apple could make a keyboard appearance that is more resilient to fingers touching it, kind of in the same way that they have with the anodized coating on the new black ones.
00:25:47 ◼ ► Obviously, in some areas, it's going to be difficult, like trackpad, which you touch a lot, and iPhones and iPads. They're touch screens. We're constantly touching the screens.
00:25:55 ◼ ► If you really ever look at an iPhone or an iPad screen, especially when it's off, they're covered with grease, but that's why Apple has the oleophobic coating and they work on that.
00:26:03 ◼ ► It's an area that Apple should continue to advance on. I feel like on the keyboards, I don't know if they're better or worse.
00:26:10 ◼ ► We should get someone to write in and say, "I am a person whose fingers wear the keys." By the way, I think they're not painted on there. That's part of the thing.
00:26:17 ◼ ► We need Mike Hurley on the show to tell us all about the keyboard technology of dying in, putting in different colors of plastic and molding them together. There's all these techniques.
00:26:26 ◼ ► Well, they're backlit, so they're probably not painted, I would assume. I'm the one who said painted a minute ago.
00:26:30 ◼ ► They're not painted. There are various techniques where you're not actually touching the letter part, so you shouldn't really ever wear it off, but for laptop keys, it may be more difficult because they're so thin.
00:26:41 ◼ ► Anyway, I agree. This is an area that Apple should improve, but I think in general, they don't look that gross to me when I see people's laptops in real life.
00:26:50 ◼ ► I think the problem of some people destroying keyboards with their fingers is it's not unsolvable, but it's going to be more difficult, especially since I personally don't understand what is the difference, but there is a difference.
00:27:01 ◼ ► You can categorize people into, "My fingers destroy keyboards, and my fingers just wear them in the normal amount."
00:27:12 ◼ ► Yeah, exactly. It is. I don't know what the difference is. It could be as simple as calluses. It could be different pH level to the secretions from their fingertips. It is what it is.
00:27:25 ◼ ► Apple should design products to withstand the use that they expect, and they could do better on the keyboards.
00:27:32 ◼ ► But I think Apple's keyboards, now that they've gotten rid of the butterfly thing, are not below average for the industry, and I personally don't think they look too disgusting even when they're kind of greasy.
00:27:41 ◼ ► Alright, moving along. We got some feedback from Gokhan Avkarugullari who writes, "If you're curious about what the new A17 Pro and M3 GPU and dynamic caching is all about, and the brand new and exceptional GPU dev tooling and shader best practices, Apple has released a bunch of new tech talks today as this was written, and one in particular, 111375"
00:28:05 ◼ ► I'll put a link in the show notes, "is all about the new GPUs," which, what did they call them in the video? It was like Apple GPU 6 or something like that. I forget exactly what they called it.
00:28:16 ◼ ► Yeah, it was very surprising for me as someone who doesn't ever pay attention to GPUs, to be completely honest with you.
00:28:21 ◼ ► But they did talk about and explain in a much better way, which I'm not going to try to recount, what this dynamic caching is all about, why it matters, and why it's helpful.
00:28:29 ◼ ► And even as someone who knows almost nothing about GPUs, I was able to grok it, and it was fascinating. And the whole video was like half an hour, so it's worth watching.
00:28:36 ◼ ► Apple Family 9 GPUs is what they're called. So this is Family 9. And yeah, what we said last week was accurate, that the dynamic caching is about cache of local memory on the GPU.
00:28:47 ◼ ► Obviously the video goes into more detail. The takeaway I got from the video is that doing this dynamic caching thing is not just about making more efficient use of resources, or rather that's not the only point.
00:29:01 ◼ ► It's like, okay, I can see maybe you save power or like, you know, you can have a smaller GPU, again, which saves you power, but you can do the same job with it.
00:29:09 ◼ ► But there's actually, I mean, this makes sense, but there's a performance story to it as well. They talk about it in the video. One of the things they're always trying to optimize in GPUs is what Apple and I think the rest of the industry calls "thread occupancy," which is how many things are in flight inside the GPU at the same time, how many things that it's doing.
00:29:27 ◼ ► And if you make inefficient use of these local caches, if one of them comes in and says, "I need to reserve, you know, a 10 by 100 block of data here, and I'm only going to use the 10 thing once, and the rest of the time I'm just using ones, but I need to reserve the whole block just for that 110 in there,"
00:29:47 ◼ ► then there's not enough room for another job to come in because, like, sorry, that shared resource that you wanted a chunk of, there's not enough left for you, so you got to wait until I'm done.
00:29:55 ◼ ► Having more better thread occupancy means you could have more of those threads in the GPU at the same time, which means, in gamer-speak, better frame rates, because something that needs to run in the GPU isn't waiting, going, "I've got to wait for these other jobs. One of these other jobs needs to finish before I can get on the GPU and do my thing."
00:30:13 ◼ ► So it is actually a performance story there, and so that's good to see and is relevant to pretty much anything that a GPU does. It's also relevant to laptops for, like, you know, getting your job done faster so you can power down the GPU and not be using it or whatever.
00:30:27 ◼ ► And you can do more work with less power because, you know, the GPU is not running as long, but it should also help give you better frame rates, so thumbs up.
00:30:35 ◼ ► Jonathan Reagan Kelly writes, and this, I think, was born from that video we just mentioned, "An interesting detail. Apple not only avoids statically allocating space for threads, they also unify the memory for different types of state within the shader core and dynamically allocate space among them.
00:30:51 ◼ ► They are also pooled with the tile cache and other cache and buffer memory. Many GPUs do have unified memory for a 'thread group', scratch pad memory, and L1 cache, but only allow the allocation between these two roles to be configured statically per kernel.
00:31:05 ◼ ► While it may seem surprising to unify 'registers' and 'caches', it's important to know that the GPU register storage is so large, up to 32 megs on a recent Nvidia GPU, that it is generally built from SRAM, not literal register file circuits, so it really is the same kind of thing as all the other caches and scratch pads or buffers.
00:31:23 ◼ ► As someone who has worked on and around GPU architectures for a long time, this is indeed a promising idea. The devil's in the details. What is the cost of the complexity this adds to the hardware? How good are the dynamic allocation algorithms in practice? How bad are the penalties when some state isn't cached?
00:31:37 ◼ ► But this is cool to see, and is actually a big change from how most other modern GPUs work, which could provide real efficiency wins.
00:31:44 ◼ ► This is actually a good pairing with Apple's unified memory architecture, and it's of the same category of, I always use this analogy, which may resonate with some of our listeners, but hopefully you can understand it, of back in the bad old days of Linux, when you would install Linux, it would ask you to sort of partition your drive to say, "How much do you want to dedicate to the slash partition? How much to the user partition?"
00:32:04 ◼ ► And you'd have to sort of say, "Okay, well, the operating system is this big, and I'll need to leave a little bit of extra room in case I want to install stuff in slash, but most of the stuff is going to be in user, including the user home directories and anything that's in user local, so how big do I want that to be, and how big do I want to do for a swap partition based on how much memory I have?"
00:32:21 ◼ ► And you have to make all these decisions kind of upfront of deciding how big these buckets will be, and the consequences of picking one wrong are annoying, because you're like, "Oh, I've got to repartition, or maybe I have to erase the entire disk," and of course, with a GPU or a CPU, if you're going to have multiple buckets of stuff, and you have to pick ahead of time how big those buckets are going to be, you don't want to make the buckets too small, because that can be a performance bottleneck, and everyone says, "Oh, they made the whatever bucket too small in the CPU, it really hurts performance,"
00:32:46 ◼ ► but if you make them too big, you're wasting power and transistors and die space on resources that aren't going to use, and that stuff could have been used elsewhere on the chip where it would have been useful, so it's hard to pick those things, and same thing with VRAM and regular RAM and having to have a big pool of VRAM and a big pool of regular RAM and transfer between them and how big are the pools.
00:33:07 ◼ ► Unified memory solves that in Apple's architecture, because it's just one big pool of hopefully very fast memory, and on the GPU, same deal. Why have three different buckets of memory needed by the GPU?
00:33:19 ◼ ► Because in any given job that the GPU is running, it may need a lot from bucket one, a little bit from bucket two, and nothing from bucket three, and if you get a whole bunch of jobs that are like that, bucket one's going to fill up and bucket two and three are going to waste, whereas if you have a unified pool, you say, "I don't make you pick ahead of time,"
00:33:35 ◼ ► when you design your chip, how big these buckets are, there's just one big bucket, and the reason it works is because, as Jonathan points out, the registers aren't like CPU registers, how you learn about them in a CPU class, or well, x86, you got A, X, B, X, C, X, like these little registers that are the size of an int or whatever that are super fast and they're right there in the CPU.
00:33:54 ◼ ► They're not like that in GPUs, it's all made of SRAM, which is what the L1 and L2 caches are, and it's big, there's like 32 megabytes of it, so Apple making one big pool of that definitely fits with their unified memory architecture and totally makes sense, and it seems like their implementation is pretty good, granted this is their first run at it, but after three weeks of looking at dynamic caching, I give it a thumbs up.
00:34:15 ◼ ► All right, M3 SSD speed versus M3 Pro SSD speed. Jon, can you take us through this, please?
00:34:24 ◼ ► Yeah, last week we were saying, "Hey, in previous generations, Apple would sometimes ship the base model with half the number of SSD packages on the logic board that would cause the SSD to be half the speed," and it's like, "Oh, if you got to 256 gigabits, it's half the speed," and then if you got to 512, which is a shame, and then with this generation,
00:34:44 ◼ ► they didn't do that, like we looked at the M3 ones, and even if you get the base model M3, granted the base model M3 MacBook Pro has 512, but either way, if you get the base model, it doesn't have half the number of chips, and so we did speed tests on it, and lo and behold, it's twice as fast as the bad old base model M2 was, as 256 was, right?
00:35:03 ◼ ► But then we compared it to a one terabyte thing, and it's like, "But the one terabyte is twice as fast again, so what's going on here?" And the thing we didn't notice last week is the one terabyte one was an M3 Pro.
00:35:15 ◼ ► It wasn't just a plain old M3 with one terabyte, it was an M3 Pro with one terabyte, so kind of like with the RAM situation of like, "What's the deal with these RAM sizes? How do they make any sense?"
00:35:24 ◼ ► They make sense in terms of how many chips there are, so the M3, the M3 SoC, supports a maximum of two NAND packages, two little flash memory rectangular things on the logic board.
00:35:36 ◼ ► The M3 Pro supports a maximum of four NAND packages, and if you look at an M3 Pro teardown, you might not see it, but there's like two on top of the logic board, and then two underneath on the underside, so you won't see another two things on the, if you're just looking at one side of it, the other two on the other side.
00:35:51 ◼ ► And then the M3 Max supports a maximum of eight NAND packages. Doesn't mean they're all filled in, like if you get the M3 Max one and you get a smaller SSD, I don't know what the size threshold is, but if you get a smaller one, all eight spots on the logic board won't be filled.
00:36:05 ◼ ► Maybe only four of them will be, but the upshot of all this is that every NAND package you add adds more maximum bandwidth than these artificial benchmarks, because you know, if you're writing to it, you can be writing to all of them at once, right?
00:36:17 ◼ ► And so that's why you get twice the speed if you have two chips, because now you're writing to them, both of them once, you know, they're both the same speed as the one was, but now you get double.
00:36:28 ◼ ► And that is borne out by Marco's monster M3 Max MacBook Pro with an eight terabyte SSD, which has all eight of the NAND packages filled in.
00:36:38 ◼ ► By the way, each NAND package has an Apple custom controller die plus four to I think 16 NAND dies all inside that little package.
00:36:50 ◼ ► So when Marco did a brief speed test with that Blackmagic speed test app that everybody uses, he was getting 8,200 megabytes per second in write speed and 5,600 megabytes per second in read speed.
00:37:00 ◼ ► So finally, if you pay whatever on godly some Apple charges for this eight terabyte SSD, you too can have a faster SSD than the $75 one terabyte stick that's in my PS5.
00:37:18 ◼ ► Why was the M3 Pro twice as fast as the M3? Because it has four NAND packages and the M3 only supports two and the M3 Max supports eight.
00:37:26 ◼ ► Who knows how many of the ultra will support? Probably only eight, but we'll see when that comes out.
00:37:31 ◼ ► Anonymous writes, "The single package quote unquote bad SSD configuration was debated a lot and it was ultimately decided it was the right trade off.
00:37:40 ◼ ► The performance analysis team, which is not the performance marketing team, this one is part of engineering, determined that there was a point of sharply diminishing returns in SSD performance in real world workloads versus benchmarks.
00:37:51 ◼ ► They were surprised at first too and the SSD and NAND teams were skeptical. So there was a lot of discussion about this, but outside of the file copy to or from a faster Thunderbolt external SSD, they just didn't see significant impact in real world workflows.
00:38:09 ◼ ► Well, anyway, this is, this is an explanation of the bad ones on the M2, right? So why did we do, why did, why did Apple do that with the M2 ones?
00:38:16 ◼ ► There was internal debate about it and this is what they came up with. Like, okay, well yes, it's half speed when you run a benchmark, but like who's running benchmarks? Like where does it impact real thing?
00:38:24 ◼ ► And so the, uh, the little part here at the end here is outside of file copy to, from a fast external SSD, I think when people are waiting on disk IO, it's probably because they're doing a big copy.
00:38:36 ◼ ► So that's a pretty big exception to say, Oh, it doesn't affect real world scenarios except for when you're really waiting for a gigantic copy. Now, obviously copies to itself on the internal drive don't count because APFS has those constant time, you know, instant clone copy and write magic things or whatever.
00:38:50 ◼ ► But if you're trying to pull something from a very fast SSD, it's connected with Thunderbolt or copy something to one, that's when you're going to hit this bottleneck.
00:38:59 ◼ ► So the bottleneck was there, but if you're doing pretty much anything other than that, what this anonymous source from inside Apple is saying is in our testing, if you're doing like a real world task that is not a big file copy to an external drive, you know, again, if you, the diminishing returns.
00:39:16 ◼ ► And I still think it wasn't a good choice. And whether Apple revisited that decision, which is why we don't have the problem on the M3 or whether we don't have the problem that empty merely because they bumped the base config to five 12.
00:39:27 ◼ ► Either way, the problem doesn't exist anymore. So it's kind of a moot point, but this just goes to show when Apple does something like this, they do debate it internally.
00:39:35 ◼ ► They may not come to the decision that we on the outside agree with, but they, they always do have their reasons.
00:39:43 ◼ ► C.G. writes, I thought of this graph when Marco was talking about how cheap SSDs have gotten.
00:39:52 ◼ ► So this is from the Reddit community data hoarder, and they show a graph of SSD prices as compared to regular spinning platter hard disks.
00:40:01 ◼ ► And back in the ye olden times of 2013, a spinning platter hard drive was about $36 a terabyte and a SSD was $625 a terabyte.
00:40:12 ◼ ► However, here in 2023, we're down to, according to this chart, about $35 a terabyte for an SSD and $13 a terabyte for platters.
00:40:23 ◼ ► And if you chart these lines out, somewhere around 2030-ish, SSDs could feasibly be even cheaper than spinning platters.
00:40:33 ◼ ► I'll believe it when I see it, but if you follow the trend line, that's what the graph would indicate.
00:40:37 ◼ ► And that's fascinating, and certainly does not match what Apple is charging people like Marco for their SSD size increases.
00:40:45 ◼ ► The SSD line is pretty steep. We'll put a link in the notes so you can look at what the graph is.
00:40:50 ◼ ► The reason they're going to meet is because the SSD line is going down faster than the hard drive line.
00:40:53 ◼ ► I'm amazed at how linear both of them are. They're very, well I guess this is a logarithmic graph, I think it is.
00:40:58 ◼ ► But anyway, having them cross in 2030, tune in in 2030 to see if that has actually happened, but boy won't that be a glorious moment.
00:41:06 ◼ ► Even just the ratios between them. Back 10 years ago, the SSD was almost 20 times more expensive per byte than the hard drive.
00:41:15 ◼ ► Now it's like 3 times. That's still a large multiple if you're looking at a huge amount of storage.
00:41:21 ◼ ► But there's also a huge number of advantages for SSDs, and so that actually might, now that it's "only" 3 times the price per gig or whatever,
00:41:36 ◼ ► As long as you're buying anybody except Apple's SSDs, everyone else's SSDs on the market are getting remarkably affordable.
00:41:45 ◼ ► Indeed. David Schaub writes, "It's hard to believe that consumer Mac laptops have had the same base RAM since 2017,
00:41:54 ◼ ► and the base RAM has only increased one time since 2011." So in the last 12 years, we've gotten one bump.
00:42:03 ◼ ► And there is a logarithmic chart that we will put in the show notes that displays this. It's sad times, y'all. Not good times.
00:42:10 ◼ ► Yeah, if you look at this graph, kind of the point of this graph is like, the graph kind of divided down the middle, the left half and the right half of the graph.
00:42:19 ◼ ► And the left half of the graph seems like we had pretty regular bumps. The slope looks pretty good. It's going up, up, up.
00:42:25 ◼ ► And then in the right half of the graph, it's like a knee in the curve, and it just levels off, and it becomes like asymptotically approaching 8 gigs, or 8.1 gigs or whatever.
00:42:35 ◼ ► And David Schaub continues with a future one where he color codes the graph. He says, "I don't think it's quite this simple."
00:42:41 ◼ ► But the graph has the left side colored in with green. It says "Steve Jobs" because it goes from 1999 to 2011 when Jobs died.
00:42:48 ◼ ► And then the right side is Tim Cook. And in the Tim Cook era, there has been one bump in base RAM on consumer laptops.
00:43:01 ◼ ► Now, does that mean that Tim Cook, unlike Steve Jobs, keeps products around longer? We know he does that.
00:43:08 ◼ ► But there are other things that may be factors here. Maybe it is the case that this is just a natural curve of RAM prices in general.
00:43:16 ◼ ► And the graph of RAM prices also takes a curve like this. I think that's really true. Someone sent us a thing of RAM prices,
00:43:24 ◼ ► but one thing about the RAM price chart is it is not as smooth as the SSD and hard drive chart that we had previously.
00:43:29 ◼ ► There is actually a lot of volatility in RAM prices, and as the person who sent us the chart pointed out,
00:43:35 ◼ ► Apple doesn't sort of buy the RAM prices at instant time. They have to sort of lock in an order of however huge quantity,
00:43:48 ◼ ► "Hey, we just ordered 10 billion sticks of this kind of RAM back when they were sticks at this price."
00:43:54 ◼ ► And that's great because now it's gone up since then. But sometimes they lock in the price at a higher price, and that's bad for them.
00:43:59 ◼ ► But either way, what this graph shows, it doesn't really allow you to assign blame or say someone was doing something wrong or whatever,
00:44:06 ◼ ► but it does show you that the feeling you have that RAM has not been going up as quickly in the past decade or so,
00:44:16 ◼ ► So hopefully we'll get that next bump soon. And by the way, the one bump that we had since 2011 was from 4 to 8.
00:44:27 ◼ ► We are brought to you this week by Notion. Notion combines your notes, docs, and projects all together in one beautiful space.
00:44:41 ◼ ► Notion AI can now give you instant answers to your questions using information from across your wiki, projects, docs, and meeting notes.
00:44:53 ◼ ► For instance, what about NextQuarters roadmap? Or you can find that marketing campaign proposal you're looking for.
00:44:59 ◼ ► You can dig up a long-lost link, all in seconds. And this is making use of your entire database of knowledge to make sure the answers are accurate and helpful,
00:45:08 ◼ ► and all comes back to you in seconds. And also, Notion AI is designed to protect your information and your privacy.
00:45:15 ◼ ► No AI models are trained with your info. Data is always encrypted, and answers will never use information from pages you don't have access to.
00:45:22 ◼ ► So it all is really secure and trustworthy. So with Notion AI, it just takes your information to the next level.
00:45:28 ◼ ► Try Notion AI for free when you go to notion.com/ATP. That's all lowercase letters, notion.com/ATP, to try the powerful, easy-to-use Notion AI today.
00:45:43 ◼ ► And when you use our link, you'll be supporting our show, notion.com/ATP, all lowercase.
00:45:57 ◼ ► Marco, you and I both have received some new treats, and I feel like we should talk about them. So Marco, tell me what you got.
00:46:05 ◼ ► And I don't know if you want to round robin it, if you just want to start talking, and I'll interrupt when necessary, but tell me about what you bought.
00:46:16 ◼ ► I did get the black, as mentioned last episode when I had it sitting next to me on my desk, but I had not yet actually moved into it yet.
00:46:24 ◼ ► It is great. I've been doing a lot of overcast work the last couple of days, a lot of having to do clean builds to work around bugs in Xcode.
00:46:33 ◼ ► So I've been doing a lot of clean builds during regular development that wouldn't be necessary in the Objective-C days, but they're certainly necessary now.
00:46:45 ◼ ► A clean build of overcast went from about 19 seconds to about 11 seconds, so it's almost twice as fast at doing this surprisingly common and increasingly common task that makes me stop and wait.
00:47:01 ◼ ► The build time reduction alone would make this worthwhile to me at this point in my development life, because, again, just how many times per day I'm having to do not only incremental builds, but especially clean builds, which are so time-consuming.
00:47:14 ◼ ► As I push more and more of my code base into Swift and SwiftUI, it just increases how much of that, how much stress is being put on the development process.
00:47:30 ◼ ► Because, for those of you who don't know, Swift and SwiftUI are extremely complex languages at the compile side.
00:47:36 ◼ ► Part of the reason why is that they are doing huge amounts of complexity at compilation time to save you from yourself, basically.
00:47:45 ◼ ► To save you from different classes of bugs, to enforce different features and protections within the Swift type system, and the Swift language itself is also extremely complicated.
00:47:54 ◼ ► And, of course, SwiftUI really pushes everything to the limits with the way it abuses the language to achieve what it achieves.
00:48:02 ◼ ► Even if you're not using previews, which I actually am not using SwiftUI Live previews that much in my work so far.
00:48:13 ◼ ► Whereas, if I compile Objective-C code today, it builds in like half a second and it's just done.
00:48:20 ◼ ► And if I make a mistake in Objective-C code, it highlights it instantly. It's so fast because Objective-C and C that it's based on are just much smaller and simpler languages to compile on modern hardware.
00:48:34 ◼ ► But as I'm pushing my code base into the more modern languages, it is costing me dearly in computing resources at development time.
00:48:42 ◼ ► Of course, that in the long run should prove worthwhile in the sense that not only does it get my code into where it needs to be for the future,
00:48:50 ◼ ► but also it should prevent and avoid a lot of bugs along the way that cost developer time, which is much more precious than computing time.
00:48:59 ◼ ► But anyway, all this is to say, the more Swift and SwiftUI I'm using and moving towards, the greater my computing needs are and the slower it is to build the app.
00:49:11 ◼ ► If I don't keep doing these upgrades every few years. So that's the biggest justification I have for the upgrade.
00:49:18 ◼ ► And that alone, cutting the build time almost in half after only two years? There's been a lot of talk about, "Oh, the M3 is only a little bit faster than the M2," which is actually not always the case.
00:49:32 ◼ ► Not always the case? Never the case. I've seen that same sentiment and I do not understand it, which is why I put these graphs.
00:49:42 ◼ ► But when I was talking before about, "Oh, why should they compare to the M2?" Because you want to know if this is the year to buy.
00:49:48 ◼ ► The thing that got Marco excited to buy this when he was not excited about the M2 is, I mean, he just talked about it.
00:49:58 ◼ ► If that's not enough, forget about synthetic benchmarks or whatever. That's him doing real world stuff.
00:50:03 ◼ ► But if you care to look at the synthetic benchmarks, this MaxTech video has a whole bunch of them.
00:50:09 ◼ ► Single core, you'll see the M2 line is barely past the M1 line. The M3 line is significantly past.
00:50:16 ◼ ► Multi-core, the M3 is like double the M2 max. The M3 max is like double the M2 max. It's not even close.
00:50:28 ◼ ► The Cinebench score, same thing. In the logic test, it's like three times as fast as the M2 max.
00:50:35 ◼ ► In the logic test, it's faster than the M1 Ultra. Is this the one to buy? Is this a good year?
00:50:40 ◼ ► Yeah, the M3 is a good year. If you're buying the highest of the high end, like Marco is, the M3 is a good year to buy.
00:50:47 ◼ ► So I think Marco was very wise, and Casey, very wise to skip the M2. Not because the M2 was bad. The M2 was great.
00:50:52 ◼ ► It was the best that you could get. But if you already have an M1, an M2 was a smaller upgrade.
00:50:57 ◼ ► The M3, depending on what you're doing, could be a monstrous upgrade. The M3 max, by the way, I'm talking about.
00:51:11 ◼ ► Like it's so bountiful with upgrades for us that seem to have basically no downsides except SSD costs.
00:51:18 ◼ ► But it wasn't that long ago that back in the tail end of the Intel days, not even just the tail end, like the whole rear third of the Intel days,
00:51:42 ◼ ► For most of the last decade or more, we'd be lucky to get maybe 15% of a gain over that same time span in sheer CPU performance during that time.
00:51:55 ◼ ► Again, 15% between generations maybe. And even then, it was usually achieved only with some real thermal aggression.
00:52:05 ◼ ► To be clear, for the single core, that's kind of what the M3 max gets. It's like 17% faster in single core than the M2 max.
00:52:12 ◼ ► But again, if you look at the difference between the M1 and the M2 max, it was not 15%.
00:52:17 ◼ ► So 17% single core over a computer that came out in January and now we're getting it in November is still pretty good.
00:52:26 ◼ ► But the big wins come from like, "Okay, but what about the multi-core and to a lesser extent some of the GPU stuff?"
00:52:32 ◼ ► You might think of like, "Oh, compiler. Isn't that a single core thing?" No, there's plenty of things to be compiled.
00:52:37 ◼ ► And so you can use those cores because each core is compiling a different file and many files make up an application.
00:52:44 ◼ ► That's why Marco is not getting 15% faster compiles because it is a multi-core type of thing.
00:52:50 ◼ ► And we do pay a price for it. The prices of the M3 max can use at peak 50 watts-ish and the M2 max is using maybe 35, 37 watts.
00:52:59 ◼ ► There's a little bit more power being used, a little bit more heat being generated, but nothing that I think the cooling system can't handle.
00:53:05 ◼ ► Yeah, that is one area that it's a little too soon to say. I mean, I've never heard the fan so far in the week that I've had this.
00:53:11 ◼ ► It's a little too soon to say whether that will be a noticeable difference here because I noticed the same thing.
00:53:16 ◼ ► I believe it was AnandTech who had like the power consumption graphs of the various max chips over time.
00:53:22 ◼ ► And they are ramping up the power use at peak. And that is, I am, I got to say I was not happy to see that in those graphs because I'm a little worried now.
00:53:32 ◼ ► You can't get blood from a stone. Like if you want it to be faster and you want to have more performance cores doing all that extra work, they have to use more power.
00:53:38 ◼ ► But like it's, we're not sure about the efficiency, but that was the point that was made by someone whose name I forget in a past episode was that the M2, despite using more heat than the M1, was still more efficient because it would be done with the same amount of work faster.
00:53:52 ◼ ► I hope so. And probably. But certainly it remains to be seen, like, you know, the M1 Max was so glorious because that computer just never spun the fan up.
00:54:04 ◼ ► And I still am so proud of the M1 Max MacBook Pro and so baffled by the M1 Max Studio because like the Max Studio was frankly embarrassed by the MacBook Pro.
00:54:19 ◼ ► That, you know, it had that, that first gen Max Studio had that weird thermal design where the fan was audible at all times, whereas the MacBook Pro had the same processor as the base Max Studio and was never audible in a laptop.
00:54:32 ◼ ► The problem with the Max Studio was that even at max load, the fans were the same speed as when it was idle.
00:54:38 ◼ ► And they were, the cooling system was so over, so like, so over-pervisioned, but then it never got quiet enough for my taste anyway.
00:54:48 ◼ ► Right. You know, so you'd run a benchmark and it'd be like, how, how fast are the fans going compared to when it was idle? Same speed.
00:54:58 ◼ ► Anyway, they fixed that in the M2 Max Studio generation. So if you're worried about a Max Studio, I'm assuming that also won't be, that it'll be fine in the M3 generation as well.
00:55:06 ◼ ► Yeah, I would, I would assume so as well. But anyway, I mean, I am, I'm extremely happy with this so far.
00:55:17 ◼ ► And you know, all the, all like the YouTubers, they're still complaining like, well, it's not that dark.
00:55:25 ◼ ► Honestly, it's pretty dark. Like I, maybe if you are in a video studio with spotlights pointing at it, maybe it's not that dark.
00:55:33 ◼ ► It's pretty dark. I'm very happy with it. I, you know, I don't think I will go black for all my computers in the future, but I like this one being black as just like a different fresh new take on things.
00:55:44 ◼ ► And yeah, I was working on it earlier today, like in laptop mode, on a couch, in the sun, using the built in speakers to play fish, which are surprisingly decent. Like it was great.
00:55:55 ◼ ► I was getting a lot of work done. Once again, I am very, very glad that I got the 16 inch just because I need as much screen space as possible.