ATP

00:00:00 ◼   ► - Good morning.

00:00:02 ◼   ► (laughing)

00:00:03 ◼   ► - Well done, well done.

00:00:05 ◼   ► - I had to do it, it's so, okay so we're recording

00:00:07 ◼   ► in the morning, this almost never happens.

00:00:10 ◼   ► It's very strange like to be here,

00:00:13 ◼   ► Casey, you're probably like wearing a tuxedo

00:00:15 ◼   ► 'cause I don't know what you wear during the day.

00:00:16 ◼   ► We both have our like you know fully awake,

00:00:18 ◼   ► well asterisk, you know, as much as they kind of be awake,

00:00:21 ◼   ► our dogs, you know in day time mode.

00:00:24 ◼   ► Possible outside noise from you know,

00:00:26 ◼   ► lawn mowers and stuff like that.

00:00:28 ◼   ► - It's a crazy day.

00:00:30 ◼   ► - Yeah, everything is upside down.

00:00:31 ◼   ► My voice is apparently upside down as well.

00:00:33 ◼   ► There's some viral thing going around Richmond

00:00:35 ◼   ► and probably the entirety of America right now.

00:00:37 ◼   ► But anyway.

00:00:38 ◼   ► - I think there's more, I think there's a couple maybe, yeah.

00:00:40 ◼   ► - Yeah, fair.

00:00:41 ◼   ► But yeah, so Michaela was sick last week

00:00:43 ◼   ► and had like a cough and runny nose

00:00:46 ◼   ► and we took her in and she was tested for all the things

00:00:49 ◼   ► and the doc said, "No, it's just a viral thing."

00:00:51 ◼   ► And so I think it's now my turn.

00:00:53 ◼   ► But anyway, so I sound a little weird.

00:00:55 ◼   ► It's not Marco's editing, it's not anything like that.

00:00:58 ◼   ► Or it could be, maybe it was because I was so excited

00:01:00 ◼   ► about colors and we'll talk about that in a little while.

00:01:03 ◼   ► Maybe that's why I was just screaming at the top of my lungs

00:01:06 ◼   ► in excitement last night.

00:01:07 ◼   ► - Well, I mean, continuing Apple's Halloween theme,

00:01:09 ◼   ► maybe you're just doing like a Dracula voice or something.

00:01:12 ◼   ► - That's what it is.

00:01:13 ◼   ► - What do you mean, I always sound like this?

00:01:16 ◼   ► - I have come to suck your wallet dry.

00:01:18 ◼   ► Or at least that's what Tim said to me last night.

00:01:20 ◼   ► (laughing)

00:01:22 ◼   ► But nevertheless, yeah, so everything is upside down.

00:01:24 ◼   ► ATP is not in morning show.

00:01:26 ◼   ► We all sound weird to some degree.

00:01:28 ◼   ► Actually, Marco, you sound normal.

00:01:30 ◼   ► But John typically has a morning voice.

00:01:32 ◼   ► I haven't heard too much of it yet.

00:01:33 ◼   ► I obviously sound all sorts of funky.

00:01:35 ◼   ► So we're just gonna have a whole

00:01:38 ◼   ► Halloween extravaganza this morning.

00:01:39 ◼   ► Oh, that's what I was gonna say.

00:01:40 ◼   ► I am in my official Limitless LLC dress, costume, tuxedo,

00:01:45 ◼   ► no, whatever it's called, dress code, there it is.

00:01:48 ◼   ► I have on an ATP shirt because I am that guy

00:01:51 ◼   ► that wears his own band's t-shirt almost all the time.

00:01:53 ◼   ► I feel like we talked about this recently.

00:01:54 ◼   ► - But during your own concert, you are wearing your shirt?

00:01:58 ◼   ► - Yeah, yeah, yeah, oh yeah, absolutely.

00:02:00 ◼   ► And I have on Mack Weldon, former sponsor,

00:02:03 ◼   ► I don't think future sponsor,

00:02:04 ◼   ► but certainly former sponsor Mack Weldon Sweats,

00:02:06 ◼   ► or I forget what these are officially called,

00:02:07 ◼   ► but I call them sweatpants,

00:02:08 ◼   ► and they are the most comfortable.

00:02:10 ◼   ► - The ace sweatpants, I believe.

00:02:11 ◼   ► - Yes, yes, I think that's right.

00:02:13 ◼   ► The most comfortable sweatpants that I've ever put on.

00:02:15 ◼   ► And I actually do have Mack Weldon slippers as well.

00:02:19 ◼   ► - They aren't sponsoring this episode, man.

00:02:21 ◼   ► I'm basically a walking Mack Weldon billboard all the time.

00:02:24 ◼   ► They really shouldn't still be sponsored.

00:02:26 ◼   ► - Yeah, so anyway, so I did have their slippers,

00:02:29 ◼   ► which I also recommend.

00:02:30 ◼   ► I am a rubber sole slipper kind of person.

00:02:33 ◼   ► I know that's a very controversial topic.

00:02:36 ◼   ► So I did have those on, although I've kicked them off

00:02:38 ◼   ► because the office gets very warm

00:02:40 ◼   ► when I close the door and start recording.

00:02:42 ◼   ► So that is the situation here.

00:02:44 ◼   ► Marco, you said you're a walking Mack Weldon advertisement.

00:02:47 ◼   ► - Well, mostly.

00:02:48 ◼   ► I'm also a marine layer advertisement,

00:02:51 ◼   ► another one of our former sponsors.

00:02:53 ◼   ► See, being a nerd, I basically never go clothes shopping

00:02:58 ◼   ► ever out in the world.

00:03:02 ◼   ► All of my clothes come from podcast sponsors

00:03:06 ◼   ► and Instagram ads that I've come across.

00:03:07 ◼   ► And it was funny, I recently, I was buying,

00:03:11 ◼   ► I wanted this, somebody made a really nice looking

00:03:14 ◼   ► fake suede bomber jacket.

00:03:15 ◼   ► I'm a sucker for jackets, I love jackets.

00:03:17 ◼   ► I'm so, this morning, I got to wear my beloved

00:03:20 ◼   ► Relwen Instagram ad, Relwen vertical insulator jacket,

00:03:23 ◼   ► which I think is my favorite jacket I've ever owned

00:03:25 ◼   ► 'cause it was finally cold enough to wear it.

00:03:27 ◼   ► I'm such a Relwen fanboy.

00:03:28 ◼   ► Anyway, I was getting this vegan suede jacket thing

00:03:32 ◼   ► from one of the big mainstream stores on clearance

00:03:35 ◼   ► on their website.

00:03:36 ◼   ► I was doing the find my fit thing.

00:03:37 ◼   ► Oh, should I get a smaller or medium?

00:03:39 ◼   ► It was like, all right, which of these popular brands

00:03:41 ◼   ► do you own something from to compare?

00:03:44 ◼   ► And it listed all basically the mall clothing brands,

00:03:47 ◼   ► all the big names.

00:03:48 ◼   ► And I literally had zero shirts from any of them.

00:03:52 ◼   ► (laughing)

00:03:53 ◼   ► 'Cause all my stuff is from internet brands.

00:03:56 ◼   ► - That's amazing.

00:03:57 ◼   ► - So it really told me, wow, I'm living in a very

00:04:00 ◼   ► different world than most people with my fashion choices.

00:04:04 ◼   ► - So John, it is after July in Boston.

00:04:09 ◼   ► So I assume you have on approximately,

00:04:10 ◼   ► let me do some mental math, carry the one, 15 layers on?

00:04:14 ◼   ► - It's been still pretty hot here.

00:04:16 ◼   ► We've been having the thing where the summer lasts

00:04:18 ◼   ► longer and longer.

00:04:19 ◼   ► We had like an 80 degree days last week,

00:04:21 ◼   ► so I was still in shorts and a t-shirt.

00:04:22 ◼   ► But yeah, now we're back to fall weather at least here.

00:04:26 ◼   ► So it was like, what, 39 degrees this morning.

00:04:29 ◼   ► I was thinking about it, I'm kind of dressed like

00:04:32 ◼   ► how people dress in the hospital.

00:04:34 ◼   ► Like if you're in the hospital, but you're like--

00:04:35 ◼   ► - What?

00:04:36 ◼   ► (laughing)

00:04:37 ◼   ► - You're not stuck in a bed.

00:04:38 ◼   ► Like you're a patient in a hospital,

00:04:39 ◼   ► you've been there for a while,

00:04:39 ◼   ► but you're not stuck in a bed.

00:04:41 ◼   ► - Do you have your rump covered right now

00:04:42 ◼   ► or is that flapping in the breeze

00:04:43 ◼   ► because you have a hospital gown on?

00:04:45 ◼   ► - Not like that, but just like what are the most comfortable

00:04:48 ◼   ► clothes that you wear when you don't care

00:04:50 ◼   ► if anybody sees you and that's how I dress

00:04:52 ◼   ► pretty much all the time.

00:04:53 ◼   ► That's living the dream.

00:04:55 ◼   ► Yeah, so I'm just wearing sweatpants and a sweatshirt.

00:04:59 ◼   ► - What did you wear?

00:05:00 ◼   ► Wow, did you hear my voice crack there?

00:05:01 ◼   ► Apparently you're going through puberty today as well.

00:05:03 ◼   ► What did you, what did you, what did you wear?

00:05:05 ◼   ► - Putting on a lot of costumes today.

00:05:07 ◼   ► (laughing)

00:05:09 ◼   ► - What did you wear when you went to the office?

00:05:10 ◼   ► Was your--

00:05:11 ◼   ► - I always wear jeans, I wear jeans and sneakers,

00:05:14 ◼   ► running shoes usually.

00:05:15 ◼   ► And then I would wear some kind of long sleeve shirt

00:05:19 ◼   ► just sometimes with a collar.

00:05:22 ◼   ► You've seen my long sleeves like striped rugby shirts.

00:05:25 ◼   ► I have various other kind of like waffle-y pattern shirts.

00:05:28 ◼   ► Not too dressy, but it was always long sleeve

00:05:30 ◼   ► even in the summer because the air conditioning

00:05:32 ◼   ► was freezing in the office all the time.

00:05:34 ◼   ► - That's big John Energy there.

00:05:35 ◼   ► - And it's a t-shirt underneath.

00:05:36 ◼   ► Like I'm wearing a t-shirt underneath my sweatshirt,

00:05:38 ◼   ► t-shirt underneath whatever long sleeve thing

00:05:40 ◼   ► I would wear to work.

00:05:41 ◼   ► - Yeah, this is also controversial these days.

00:05:43 ◼   ► I always wear what I would call an undershirt.

00:05:45 ◼   ► It's just a white tee under my clothes,

00:05:48 ◼   ► regardless of the time of year.

00:05:49 ◼   ► And this is apparently a very nerdy or old man thing to do.

00:05:53 ◼   ► - What?

00:05:54 ◼   ► No, it's a sweat control thing.

00:05:56 ◼   ► - That's what I thought.

00:05:57 ◼   ► - At minimum, if you wear a long sleeve thing

00:06:01 ◼   ► on top of nothing, it's gonna be sweatier in there

00:06:04 ◼   ► and grosser smelling.

00:06:05 ◼   ► - Yeah, you can't do that.

00:06:06 ◼   ► That's illegal.

00:06:07 ◼   ► I will say though, it is kind of amazing.

00:06:09 ◼   ► We're recording this on Halloween

00:06:10 ◼   ► and Apple's done this spooky event.

00:06:13 ◼   ► Can you think of anything scarier than tech podcasters

00:06:16 ◼   ► talking fashion?

00:06:17 ◼   ► (laughing)

00:06:18 ◼   ► - We're really good in the spirit.

00:06:19 ◼   ► - I mean, we're talking about clothing.

00:06:20 ◼   ► Let's, yeah, that's all that's--

00:06:22 ◼   ► - Is that that much better?

00:06:24 ◼   ► - We do cover our bodies with stuff.

00:06:27 ◼   ► Fashion, I feel like is something different.

00:06:29 ◼   ► - We should probably dig in and as per tradition

00:06:33 ◼   ► and as per the pre-flight that none of you got to hear,

00:06:36 ◼   ► we are not allowed to talk follow-up

00:06:37 ◼   ► and we will jump straight to Apple's October 30th event.

00:06:42 ◼   ► This was very quick.

00:06:43 ◼   ► It clocked in at, I think just barely over 30 minutes.

00:06:46 ◼   ► So if by ATP tradition says we should go somewhere

00:06:51 ◼   ► between two and three hours this episode.

00:06:53 ◼   ► So we'll see what happens.

00:06:54 ◼   ► But we're getting a lot of attention to the Mac

00:06:57 ◼   ► and all the good Macs 'cause we don't have to talk

00:06:59 ◼   ► about the Mac Pro, right, John?

00:07:00 ◼   ► - Ooh, sick burn.

00:07:01 ◼   ► - I mean, they did really emphasize

00:07:03 ◼   ► that they were talking about laptops,

00:07:04 ◼   ► but you know, whose fault is that?

00:07:06 ◼   ► Not ours.

00:07:08 ◼   ► - It's mine.

00:07:09 ◼   ► It's actually Marco and mine 'cause we don't have desktop.

00:07:11 ◼   ► - No, it's Apple's fault.

00:07:13 ◼   ► - Yeah, it was interesting.

00:07:15 ◼   ► Right from the start, there was Tim Cook's intro.

00:07:18 ◼   ► Well, first there was a little video

00:07:20 ◼   ► about how Macs are doing all this hard work

00:07:22 ◼   ► and making it easy on us, which was cute, fine, forgettable.

00:07:26 ◼   ► But the cool thing was they did this event at night

00:07:30 ◼   ► and it had this Halloween theme, the whole thing.

00:07:34 ◼   ► They're playing all these spooky sounds

00:07:36 ◼   ► and there's the choir chanting in the background

00:07:39 ◼   ► as they zoom through Apple Park at night.

00:07:41 ◼   ► And then Tim comes in and says, "Good evening,"

00:07:43 ◼   ► 'cause he normally--

00:07:44 ◼   ► - He should have said good evening like a vampire,

00:07:46 ◼   ► but he didn't.

00:07:46 ◼   ► - I don't think he has that in him.

00:07:48 ◼   ► - Yeah, right?

00:07:49 ◼   ► - It is not possible to make Apple Park spooky, I'm sorry.

00:07:52 ◼   ► I'm not sure.

00:07:53 ◼   ► It's kind of like, do they have Halloween in California?

00:07:56 ◼   ► All I can say is at Apple Park, not spooky.

00:07:59 ◼   ► Like, they can't, I don't know if they have fall there,

00:08:02 ◼   ► but they did amazing effects, amazing visual effects

00:08:05 ◼   ► by adding fog and it's dark

00:08:07 ◼   ► and they have fake bats flying around.

00:08:08 ◼   ► I was like, great, good.

00:08:10 ◼   ► It's not spooky.

00:08:10 ◼   ► Apple Park is not spooky.

00:08:11 ◼   ► Anyway, I give them an E for effort.

00:08:14 ◼   ► - Yeah, I mean, it's, look,

00:08:16 ◼   ► what Apple is challenged to do recently

00:08:20 ◼   ► is not seem like one of the biggest corporations

00:08:23 ◼   ► in the world run by mostly a bunch of older white men.

00:08:25 ◼   ► - The way to do that is not to spend this much money

00:08:27 ◼   ► on a half an hour presentation

00:08:29 ◼   ► where you introduce a couple of rats.

00:08:31 ◼   ► How much money did this cost?

00:08:33 ◼   ► - Yeah, but I gotta give them credit.

00:08:35 ◼   ► To whatever degree they are able to

00:08:38 ◼   ► as a giant corporation,

00:08:40 ◼   ► they seem to have some fun with this.

00:08:43 ◼   ► And yeah, of course it's still a very corporate,

00:08:45 ◼   ► basically commercial, but they did have some fun with it.

00:08:49 ◼   ► I think it worked.

00:08:50 ◼   ► I think it would be a little much

00:08:52 ◼   ► if they did this frequently, but they don't.

00:08:54 ◼   ► So it was a fun thing for this one.

00:08:56 ◼   ► - It kinda reminds me of the holiday events they have

00:08:58 ◼   ► in live service video games,

00:09:00 ◼   ► like video games that are ongoing,

00:09:02 ◼   ► sometimes you pay a subscription or whatever.

00:09:03 ◼   ► Like, they'll always have like a,

00:09:05 ◼   ► some kind of non-denominational winter holiday event

00:09:09 ◼   ► and something for Halloween or whatever.

00:09:11 ◼   ► Just because that's a fun thing to do

00:09:12 ◼   ► because you have a bunch of people who are there

00:09:14 ◼   ► and you want to mark the change of the holiday

00:09:16 ◼   ► so you take your game stuff and you're like,

00:09:18 ◼   ► "Oh, let's have pumpkins on top of our sci-fi setting

00:09:20 ◼   ► or our fantasy world or whatever."

00:09:22 ◼   ► And now it's like Apple's doing that too

00:09:23 ◼   ► because I guess their keynote events

00:09:25 ◼   ► are kind of like a live service game

00:09:26 ◼   ► where the community that's always hanging around.

00:09:28 ◼   ► So yeah, I endorse it.

00:09:29 ◼   ► I think it's fun.

00:09:30 ◼   ► It was just a little bit silly.

00:09:32 ◼   ► - Yeah, and it wasn't too heavy-handed, like you guys said.

00:09:35 ◼   ► I loved Johnny Srouji's "Welcome to My Lab."

00:09:37 ◼   ► I don't know, there's just,

00:09:38 ◼   ► something about that just made me giggle.

00:09:39 ◼   ► - He could have done that in a vampire accent.

00:09:41 ◼   ► - Also true.

00:09:42 ◼   ► But yeah, so we're here to quote,

00:09:44 ◼   ► "Celebrate a product we all love, the Mac."

00:09:46 ◼   ► Which, I don't know, I'm here for it,

00:09:48 ◼   ► but it just kind of struck me funny

00:09:50 ◼   ► that we're now saying a product that we all love, the Mac,

00:09:52 ◼   ► when we were all pretty convinced

00:09:54 ◼   ► and I feel like it was all but officially stated

00:09:56 ◼   ► that nobody cared about the Mac for a few years.

00:09:59 ◼   ► - But they've turned that around.

00:10:00 ◼   ► I mean, I feel like now at this point,

00:10:02 ◼   ► it's a little bit where every time they emphasize it,

00:10:06 ◼   ► it makes me believe them less now.

00:10:08 ◼   ► But anyway, the point is, with their actions,

00:10:09 ◼   ► they've shown that they do.

00:10:11 ◼   ► So I feel like they don't need to say it anymore.

00:10:14 ◼   ► What they've done has convinced us all.

00:10:16 ◼   ► Yes, you're doing good.

00:10:18 ◼   ► - Yep, yep, agreed.

00:10:19 ◼   ► So we go talk to Johnny and for the first time,

00:10:23 ◼   ► to the best of my recollection,

00:10:24 ◼   ► I believe Stephen Hackett had a post about this.

00:10:26 ◼   ► I'll try to link that in the show notes,

00:10:27 ◼   ► but for the first time, we got all three,

00:10:29 ◼   ► or three of the four, we would assume,

00:10:32 ◼   ► of the different flavors of M3 all at once.

00:10:35 ◼   ► We got the M3, the M3 Pro, and the M3 Max,

00:10:38 ◼   ► and all three nanometer.

00:10:40 ◼   ► Which three nanometer though, John?

00:10:41 ◼   ► How do we know?

00:10:42 ◼   ► - Yeah, so that's tricky.

00:10:43 ◼   ► So having them all at once is interesting.

00:10:47 ◼   ► It makes it for a good presentation.

00:10:48 ◼   ► Like great, they can like roll out the whole,

00:10:50 ◼   ► the quote unquote whole line out really.

00:10:51 ◼   ► Obviously, we assume there's gonna be an ultra

00:10:53 ◼   ► and maybe one more, but probably not.

00:10:55 ◼   ► It's great that they can roll them all out at the same time.

00:10:58 ◼   ► Why?

00:10:59 ◼   ► Why are they able to roll them out at the same time?

00:11:02 ◼   ► That is the mystery.

00:11:04 ◼   ► Is it because the M3 is late,

00:11:07 ◼   ► and so it came out at the same time as the other ones?

00:11:10 ◼   ► I think I said that in the last episode,

00:11:11 ◼   ► and someone sent in some follow-ups saying,

00:11:13 ◼   ► actually, if you measure the time

00:11:16 ◼   ► from between the M1 and M2,

00:11:18 ◼   ► the M3 is not behind schedule.

00:11:20 ◼   ► The plain old M3 is not behind schedule.

00:11:22 ◼   ► In fact, it had like a month or two more to go

00:11:25 ◼   ► to still be on schedule, so I apologize for that mistake.

00:11:28 ◼   ► But so what that means is basically the M3,

00:11:29 ◼   ► if you look at the spacing between the plain M1,

00:11:31 ◼   ► the plain M2, and the plain M3,

00:11:33 ◼   ► this is more or less on schedule for that chip,

00:11:36 ◼   ► if not a little bit early.

00:11:38 ◼   ► And that means the Pro and the Max are very early

00:11:41 ◼   ► because they usually come after the plain M3.

00:11:43 ◼   ► We talked about it in the last episode.

00:11:45 ◼   ► What, you know, you were in the same case,

00:11:47 ◼   ► like maybe this won't be three nanometer, right?

00:11:49 ◼   ► Which is a reasonable thing to think,

00:11:50 ◼   ► 'cause like what could, you know,

00:11:52 ◼   ► we've heard all about the M3D process

00:11:54 ◼   ► and how it's expensive,

00:11:56 ◼   ► and how like they wanna move on

00:11:58 ◼   ► as quickly as possible to M3E.

00:11:59 ◼   ► This is TSMC we're talking about,

00:12:01 ◼   ► which will be less expensive because it has better yields,

00:12:04 ◼   ► and Apple is the only company that's buying M3B.

00:12:08 ◼   ► And in the earnings call that we talked about

00:12:10 ◼   ► in the last episode, TSMC said they would be ramping up

00:12:13 ◼   ► volume production in the fourth quarter of 2023 on M3E.

00:12:16 ◼   ► And we're like, hey, it's the fourth quarter of 2023.

00:12:20 ◼   ► Maybe these chips will be on M3E.

00:12:22 ◼   ► Obviously, as predicted, Apple did not use the phrases

00:12:26 ◼   ► TSMC, M3B, or M3E anywhere in any of their presentations

00:12:30 ◼   ► or media communication about this.

00:12:32 ◼   ► I'm not sure if ask a direct question

00:12:34 ◼   ► if they would answer the question.

00:12:37 ◼   ► So we are left to speculate.

00:12:39 ◼   ► One thing they did say though is three nanometers.

00:12:41 ◼   ► So that narrows it down substantially.

00:12:43 ◼   ► It's one of those M3 somethings,

00:12:45 ◼   ► and the only two possible ones we think it could be

00:12:47 ◼   ► is M3B or M3E.

00:12:48 ◼   ► And last episode, it seems like,

00:12:51 ◼   ► hey, M3E, volume production fourth quarter,

00:12:53 ◼   ► that must be what they're doing.

00:12:55 ◼   ► But our good friend Jonathan Deets Jr.

00:12:58 ◼   ► wrote in to tell us, this is an inserted follow-up

00:13:01 ◼   ► into an event episode, unprecedented.

00:13:03 ◼   ► - Ooh, what? - But it has to be done.

00:13:04 ◼   ► - What?

00:13:05 ◼   ► I'm telling Jon, this is unacceptable.

00:13:08 ◼   ► - But it's very relevant, because that's what we were

00:13:09 ◼   ► thinking, like, what the hell?

00:13:10 ◼   ► So here's what he says.

00:13:12 ◼   ► M3B cycle times, reading from his message here,

00:13:15 ◼   ► M3B cycle times are extremely long.

00:13:17 ◼   ► The cycle time is the total amount of time it takes

00:13:19 ◼   ► to go from wafer start, when blank silicon wafer

00:13:22 ◼   ► begins its journey, to wafer out,

00:13:23 ◼   ► when a finished wafer comes off the production line.

00:13:25 ◼   ► Exposure times for EUV, that's extreme ultraviolet

00:13:28 ◼   ► lithography tools, are considerably longer

00:13:30 ◼   ► than for DUV tools.

00:13:31 ◼   ► Or is that not extreme?

00:13:33 ◼   ► D, I don't know what the D stands for anyway.

00:13:34 ◼   ► It's less extreme than E.

00:13:36 ◼   ► - Wait, hold on, real quick, didn't they mention,

00:13:38 ◼   ► I swear they mentioned something about

00:13:40 ◼   ► ultraviolet lithography very briefly.

00:13:42 ◼   ► - Yeah, yeah, no, they did.

00:13:42 ◼   ► Like, that's emphasizing that all the three

00:13:44 ◼   ► nanometer things are EUV.

00:13:46 ◼   ► But anyway, remember we talked about

00:13:47 ◼   ► how many different exposures?

00:13:48 ◼   ► Anyway, continuing from the thing here.

00:13:50 ◼   ► That means the more EUV steps a process requires,

00:13:54 ◼   ► the longer it takes.

00:13:55 ◼   ► The cycle time for M3B, the most EUV intensive

00:13:58 ◼   ► manufacturing process to date, is said to be

00:14:00 ◼   ► in excess of 100 days.

00:14:03 ◼   ► That means Apple likely had to begin ramping up

00:14:05 ◼   ► production of the A17 Pro and Q1 of this year

00:14:07 ◼   ► in order to ensure sufficient inventory

00:14:09 ◼   ► for September iPhone release.

00:14:10 ◼   ► And he continues, anything three nanometer

00:14:13 ◼   ► that ships this year is M3B.

00:14:16 ◼   ► M3E should just be reaching volume

00:14:17 ◼   ► manufacturing milestone now.

00:14:19 ◼   ► That means TSMC will start to manufacture

00:14:21 ◼   ► the first orders now, with finished wafers 90 days out.

00:14:25 ◼   ► Those wafers will still need to be diced

00:14:26 ◼   ► into individual ships, tested, sorted, packaged,

00:14:29 ◼   ► sent to the device manufacturer to be included

00:14:31 ◼   ► in the final end user product.

00:14:32 ◼   ► Then that needs to be assembled, packaged,

00:14:34 ◼   ► and shipped around the world to various

00:14:35 ◼   ► retail channels prior to the product launch.

00:14:37 ◼   ► Apple requires volume manufacturing by April to ship now.

00:14:40 ◼   ► M3E products and shipping in 2023 has always

00:14:43 ◼   ► been a non-starter since the day TSMC announced M3E.

00:14:46 ◼   ► So here's Jonathan coming down strongly saying,

00:14:48 ◼   ► these cannot be M3E because there's no way

00:14:51 ◼   ► anything would be produced in time for that.

00:14:54 ◼   ► Which means, leaving the only alternative,

00:14:56 ◼   ► that these would be M3B.

00:14:58 ◼   ► Which, as we said, is the more expensive,

00:15:00 ◼   ► but better, higher density, more layers, whatever, process.

00:15:05 ◼   ► And that is interesting because, you know,

00:15:08 ◼   ► obviously we knew they would do it for the phone.

00:15:11 ◼   ► You gotta do what you gotta do, phone's gotta be out,

00:15:13 ◼   ► you want it to be three nanometers, spend the money.

00:15:15 ◼   ► Do it on M3B.

00:15:18 ◼   ► But here we have a full suite of the main M3 family

00:15:23 ◼   ► of processors for Macs that seem like

00:15:27 ◼   ► it's likely they are on M3B.

00:15:30 ◼   ► Related to that, and possibly, you know,

00:15:33 ◼   ► hurting Jonathan's credibility,

00:15:34 ◼   ► but I think I'll include it here.

00:15:36 ◼   ► Before the event, Jonathan said,

00:15:39 ◼   ► an M3 max size die on a three nanometer process

00:15:43 ◼   ► is impossible right now due to the economics

00:15:45 ◼   ► of making it ship that size in this point

00:15:47 ◼   ► in the yield curve.

00:15:49 ◼   ► Defect densities have to settle down

00:15:50 ◼   ► before you can go that big.

00:15:51 ◼   ► If Apple were to do that, it would be unprecedented

00:15:53 ◼   ► on so many levels.

00:15:55 ◼   ► Well, they did it.

00:15:56 ◼   ► (laughing)

00:15:58 ◼   ► Because they said three nanometer,

00:16:00 ◼   ► and they introduced the Pro and the Max and the Plane,

00:16:02 ◼   ► and they're all three nanometers.

00:16:03 ◼   ► So after the event, Jonathan had this to say.

00:16:05 ◼   ► This lends credence to the rumor

00:16:07 ◼   ► that Apple struck a deal with TSMC

00:16:08 ◼   ► to only pay for known good dies

00:16:11 ◼   ► rather than paying for a fixed wafer cost.

00:16:13 ◼   ► What a deal, like, you know,

00:16:14 ◼   ► we don't wanna pay for your mistakes, just, right.

00:16:16 ◼   ► He says, I guess if you give TSMC $3.5 billion,

00:16:20 ◼   ► they'll run their leading edge fab at full capacity

00:16:22 ◼   ► for the better part of the year to deliver your order,

00:16:24 ◼   ► even if they end up scrapping half the output.

00:16:27 ◼   ► It is the most efficient way to run the fab

00:16:28 ◼   ► and the quickest way to improve yields.

00:16:30 ◼   ► Like I said before, unprecedented.

00:16:32 ◼   ► So this all seems reasonable to me,

00:16:36 ◼   ► the fact that these would be M3B.

00:16:37 ◼   ► It also fits with the rumor we were saying about,

00:16:39 ◼   ► oh, Apple's manufactured all these IMAX,

00:16:41 ◼   ► which we'll get to later, already,

00:16:43 ◼   ► and they're sitting around since June or August.

00:16:45 ◼   ► With all the lead times involved

00:16:46 ◼   ► and the fact that they said they'd begin volume production

00:16:48 ◼   ► and it takes 100 days from the time a silicon wafer goes in

00:16:52 ◼   ► to the time it comes out finished,

00:16:54 ◼   ► I am willing to believe that these are M3B,

00:16:56 ◼   ► which is kind of cool from our perspective,

00:16:59 ◼   ► because again, M3B is the best process available.

00:17:02 ◼   ► M3E will be cheaper, but slightly worse,

00:17:06 ◼   ► and then there'll be the other processes after that.

00:17:08 ◼   ► So this is pretty cool.

00:17:10 ◼   ► If either Apple is eating this cost

00:17:13 ◼   ► or Apple is forcing TSMC to eat this cost,

00:17:16 ◼   ► we may be getting better three nanometer MAC chips

00:17:18 ◼   ► than we would have otherwise.

00:17:20 ◼   ► - I think the only weird kind of side note to this

00:17:23 ◼   ► is that by all accounts from every analyst,

00:17:27 ◼   ► including Jonathan Dietz, who knows a lot about this,

00:17:29 ◼   ► a lot more than us, and all the tech analysts

00:17:32 ◼   ► who follow the chip business more closely,

00:17:33 ◼   ► by all accounts, this M3 or M3B,

00:17:36 ◼   ► as it's called by a lot of us,

00:17:38 ◼   ► this process is quote, a dead end,

00:17:40 ◼   ► meaning that it doesn't seem like TSMC intends

00:17:43 ◼   ► to produce these for a long time into the future

00:17:45 ◼   ► on this process.

00:17:47 ◼   ► And if you look at Apple's products,

00:17:49 ◼   ► the Pro and Max chips in each family of these M series chips,

00:17:54 ◼   ► they're like the iPhone Pro line.

00:17:57 ◼   ► They're produced for only the time

00:17:59 ◼   ► they are the current version.

00:18:01 ◼   ► And as soon as a new version comes out,

00:18:02 ◼   ► like as soon as the M2 Pro and Max came out,

00:18:05 ◼   ► they stopped selling things with the M1 Pro and Max in them.

00:18:09 ◼   ► So they could stop manufacturing the M1 Pro and Max,

00:18:11 ◼   ► presumably, around that time,

00:18:12 ◼   ► or you know, shortly before, or whatever.

00:18:14 ◼   ► And so with this, we've assumed,

00:18:16 ◼   ► and I was speculating last episode,

00:18:19 ◼   ► that maybe the reason they would do the M3 Pro and Max now,

00:18:23 ◼   ► but not do things earlier,

00:18:25 ◼   ► actually, I wasn't even talking about that,

00:18:26 ◼   ► I was talking about the A17 Pro,

00:18:27 ◼   ► but regardless, if TSMC is not intending

00:18:30 ◼   ► to run this process for a long time,

00:18:33 ◼   ► then Apple probably isn't gonna be selling these chips

00:18:36 ◼   ► for longer than the current generation,

00:18:39 ◼   ► which is interesting in the context of, say, the iMac,

00:18:42 ◼   ► which was on a two year cycle before.

00:18:44 ◼   ► Are they gonna keep making non-Pro M3s

00:18:47 ◼   ► for longer than they make the M3 Pro and Max?

00:18:51 ◼   ► On this process, that's apparently a dead end, maybe?

00:18:53 ◼   ► I don't know how the timelines work on that.

00:18:55 ◼   ► - Again, you keep talking about it as dead end.

00:18:57 ◼   ► Like, I know what you mean,

00:18:58 ◼   ► in saying that they're not gonna keep running this

00:18:59 ◼   ► for five, six, seven years.

00:19:00 ◼   ► Like, I feel like that's true of all their processes,

00:19:03 ◼   ► is that they eventually get to the most economical variant

00:19:06 ◼   ► of five nanometer, or seven nanometer, whatever,

00:19:09 ◼   ► and that's the one they run for years,

00:19:10 ◼   ► 'cause they have a bunch of ones that are not quite as good

00:19:12 ◼   ► that they run for short periods of time,

00:19:14 ◼   ► but eventually they find the one that is the sweet spot.

00:19:16 ◼   ► I don't know how they pick that.

00:19:17 ◼   ► Is it the cheapest?

00:19:18 ◼   ► Is it the one with the best yields?

00:19:19 ◼   ► Whatever.

00:19:20 ◼   ► Is it the highest performance?

00:19:21 ◼   ► But yeah, the very first one you do

00:19:23 ◼   ► is probably not the one you're gonna be running for years,

00:19:25 ◼   ► and I forget, I've already deleted this from the notes,

00:19:27 ◼   ► but I forget which one of these processes

00:19:29 ◼   ► were sort of compatible.

00:19:30 ◼   ► Like, the idea that if you have a chip design,

00:19:32 ◼   ► and you've designed it for N3B,

00:19:34 ◼   ► can you use it in N3E?

00:19:35 ◼   ► Can you use it in N3S or P, or all these other things?

00:19:39 ◼   ► It's not always straightforward to say,

00:19:40 ◼   ► "Oh, I'll just use your new process

00:19:42 ◼   ► "to fab my chip that I already designed."

00:19:45 ◼   ► That's not always easy to do.

00:19:47 ◼   ► So I get where you're going, it's like,

00:19:48 ◼   ► look, if they're gonna make M3s

00:19:49 ◼   ► for the next two to three years in that iMac,

00:19:53 ◼   ► does that mean TSMC's running N3B for three years,

00:19:56 ◼   ► instead of stopping running it as soon as they possibly can?

00:20:01 ◼   ► The thing I think about related to that question is,

00:20:04 ◼   ► what are the volumes?

00:20:06 ◼   ► Like, maybe they sell so few Macs compared to iPhones,

00:20:10 ◼   ► that they'll just fab all the M3 chips on N3B, they think,

00:20:14 ◼   ► within the next six months, right, they think they need.

00:20:17 ◼   ► We'll just get 'em all, put 'em in a box,

00:20:18 ◼   ► put 'em off to the side.

00:20:20 ◼   ► Same thing with the Mac chips, maybe.

00:20:21 ◼   ► Maybe I'm wrong, maybe they would never do that,

00:20:23 ◼   ► and it's like, you'd be guessing.

00:20:24 ◼   ► - I doubt it, 'cause think about the risk you'd be running

00:20:27 ◼   ► of there being a big shortage if you ran out,

00:20:29 ◼   ► and then you didn't have a new MacBook Pro

00:20:31 ◼   ► slated for six months away,

00:20:33 ◼   ► then you just couldn't make MacBook Pros?

00:20:34 ◼   ► - Yeah, no, and plus, you'd have to pay for them up front,

00:20:36 ◼   ► it is mysterious, but here's the thing.

00:20:39 ◼   ► If these really are N3B, then people are like,

00:20:42 ◼   ► "Oh, the only thing that will ever be an N3B is the A17 Pro."

00:20:45 ◼   ► If these are really N3B, that assumption was wrong.

00:20:48 ◼   ► Like, the idea is that, yes, N3B is more expensive

00:20:52 ◼   ► and crappy or whatever, but maybe Apple is saying,

00:20:54 ◼   ► "I don't care," because the three-animator process

00:20:57 ◼   ► from TSMC came out later than TSMC originally

00:21:00 ◼   ► had hoped it would, right?

00:21:01 ◼   ► And I think it came out later than Apple had hoped it would.

00:21:03 ◼   ► That was part of the rumor with the new GPU core

00:21:06 ◼   ► that's in the A17 Pro and also in these Mac chips.

00:21:08 ◼   ► That GPU core was rumored for last year,

00:21:11 ◼   ► well, not last year, really, but anyway,

00:21:13 ◼   ► that was rumored for the M2 generation.

00:21:16 ◼   ► But I think that GPU was designed for three-nanometers,

00:21:19 ◼   ► so of course, they couldn't put it on the M2,

00:21:20 ◼   ► because three-nanometer wasn't ready.

00:21:21 ◼   ► So I think that has disrupted everything here,

00:21:25 ◼   ► and maybe Apple is saying,

00:21:26 ◼   ► "Look, I know you don't wanna do it,

00:21:27 ◼   ► "and I know the financial terms of our deal

00:21:30 ◼   ► "would mean that we're kinda screwing you

00:21:31 ◼   ► "by making you do it, but you're just gonna have to keep

00:21:34 ◼   ► "making the N3B stuff for us for longer

00:21:36 ◼   ► "than you would otherwise want to, eating the costs."

00:21:39 ◼   ► And I know you basically said you can never make it

00:21:41 ◼   ► as good as N3E in terms of yields,

00:21:43 ◼   ► and every chip you make, you're gonna have to throw out

00:21:45 ◼   ► half of them or whatever, but tough luck,

00:21:47 ◼   ► because that's what we need you to do,

00:21:49 ◼   ► and we're your biggest customer,

00:21:50 ◼   ► and so there may be something like that going on.

00:21:52 ◼   ► I guess we'll find out.

00:21:54 ◼   ► There are still unknowns here.

00:21:55 ◼   ► First, we don't know that these are N3B.

00:21:57 ◼   ► We think they might be based on this feedback,

00:21:59 ◼   ► but we don't know for sure.

00:22:00 ◼   ► Surely, once people get these chips

00:22:01 ◼   ► and start chopping them open or something,

00:22:03 ◼   ► someone will be able to tell, and second is,

00:22:06 ◼   ► we'll have to wait to see how long.

00:22:07 ◼   ► Maybe they actually update the iMac to the M4.

00:22:11 ◼   ► Maybe they just sell it with the M3 briefly.

00:22:13 ◼   ► Maybe the M4 comes sooner than we think

00:22:14 ◼   ► now that three nanometer is selling down.

00:22:16 ◼   ► There's a lot of questions here,

00:22:17 ◼   ► but I think despite all the questions,

00:22:19 ◼   ► I think it seems like the outcome is good no matter what.

00:22:24 ◼   ► We don't have these chips in our hands yet,

00:22:26 ◼   ► and we'll talk about their specs in a second,

00:22:27 ◼   ► but if they're N3B, we like that

00:22:31 ◼   ► because that's the best process,

00:22:32 ◼   ► and if they're not, we like that because it means

00:22:35 ◼   ► that we got this whole family of chips all at once

00:22:38 ◼   ► when maybe we wouldn't have.

00:22:40 ◼   ► Even though I'm curious about the details here,

00:22:44 ◼   ► I think no matter what the case is,

00:22:46 ◼   ► I'm happy with all these chips coming out all at once,

00:22:49 ◼   ► and I'm happy that they're all three nanometer.

00:22:51 ◼   ► - I think the real tell will be

00:22:54 ◼   ► whether we end up seeing the M3

00:22:57 ◼   ► in products that are usually kept around

00:23:00 ◼   ► as the cheaper last year's model.

00:23:01 ◼   ► So for instance, the MacBook Air,

00:23:03 ◼   ► that would be a real tell.

00:23:05 ◼   ► Does the MacBook Air get the M3?

00:23:07 ◼   ► Maybe the MacBook Air stays on M2 and skips M3

00:23:10 ◼   ► and goes right to M4.

00:23:11 ◼   ► - Yeah, 'cause the MacBook Air has higher volume.

00:23:13 ◼   ► That's what we're talking about.

00:23:14 ◼   ► They usually roll out the plain M3,

00:23:16 ◼   ► and they usually roll that out first in the cheap computer,

00:23:17 ◼   ► but here they are rolling out the plain M3,

00:23:19 ◼   ► but they're not rolling it out on the cheap laptop first,

00:23:22 ◼   ► maybe because the cheap laptops sell too many.

00:23:24 ◼   ► - Yeah, or maybe they are longer term

00:23:28 ◼   ► gonna create some segmentation,

00:23:29 ◼   ► and maybe they're gonna always keep the MacBook Air

00:23:31 ◼   ► one behind the Pro, who knows?

00:23:33 ◼   ► I mean, there's lots of different paths they could take here

00:23:35 ◼   ► and typically, we are presented as the public

00:23:39 ◼   ► and the nerds are in the circle.

00:23:41 ◼   ► Apple presents us with a product lineup

00:23:44 ◼   ► that generally abstracts away

00:23:47 ◼   ► all of the stuff that's happening below,

00:23:49 ◼   ► behind the scenes with chip manufacturing

00:23:51 ◼   ► and supply chain, stuff like that.

00:23:52 ◼   ► We don't see any of that.

00:23:54 ◼   ► All we see is this year you got M3,

00:23:57 ◼   ► last year you had M2.

00:23:58 ◼   ► You have those nice clean, the illusion

00:24:02 ◼   ► of a much simpler arrangement behind the scenes

00:24:05 ◼   ► than what actually is going on.

00:24:07 ◼   ► Occasionally, some detail about the manufacturing situation

00:24:11 ◼   ► leaks out into the products,

00:24:12 ◼   ► where you get, oftentimes you get some weird long delay

00:24:16 ◼   ► before a product gets updated,

00:24:18 ◼   ► or something skips a generation,

00:24:20 ◼   ► and oftentimes the reasons for that

00:24:21 ◼   ► are some of those details.

00:24:23 ◼   ► I think it'll be interesting to see

00:24:24 ◼   ► as we go through the next year to 18 months,

00:24:28 ◼   ► what happens to the M3 line after it's no longer the newest?

00:24:32 ◼   ► Does it stick around?

00:24:33 ◼   ► Does certain products never get it?

00:24:36 ◼   ► Who knows?

00:24:36 ◼   ► - Does Apple revise it so it does work on an M3e?

00:24:39 ◼   ► 'Cause that's a thing they can do.

00:24:40 ◼   ► You know, it's more cost, more work, more time,

00:24:43 ◼   ► but if they really do want to keep

00:24:45 ◼   ► the plain old M3 around for a long time

00:24:46 ◼   ► and eventually put it in the MacBook Air

00:24:47 ◼   ► and they want it to be an M3e,

00:24:49 ◼   ► they can change it to make sure it works with M3e.

00:24:52 ◼   ► - We are brought to you this week by Trade Coffee,

00:24:57 ◼   ► helping you make the best coffee you can right at home

00:25:00 ◼   ► and discover all sorts of new coffees.

00:25:02 ◼   ► So the best way to make great coffee at home

00:25:05 ◼   ► is freshly roasted beans.

00:25:07 ◼   ► Coffee goes kinda stale after not that long.

00:25:10 ◼   ► So with Trade, they know fresh is best.

00:25:12 ◼   ► So every coffee sent to you has been roasted within 48 hours

00:25:15 ◼   ► and shipped directly to you by one of their partner roasters.

00:25:19 ◼   ► And they connect you to the best roasters in the country

00:25:22 ◼   ► so that you can discover all sorts of great coffee

00:25:24 ◼   ► to brew at home.

00:25:25 ◼   ► They have over 450 coffees to choose from.

00:25:28 ◼   ► So Trade guarantees you'll find something new to love.

00:25:32 ◼   ► And this is great for discovering

00:25:34 ◼   ► all sorts of new craft coffee.

00:25:36 ◼   ► These are from over 55 of the top indie roasters

00:25:39 ◼   ► in the country so that you can make

00:25:40 ◼   ► your best cup of coffee at home every day.

00:25:43 ◼   ► And they curate it to your taste.

00:25:45 ◼   ► So they use their matching algorithm

00:25:47 ◼   ► to curate exactly what you will like.

00:25:50 ◼   ► And they have these wonderful feedback mechanisms

00:25:53 ◼   ► if you want to tweak anything,

00:25:54 ◼   ► but honestly, you probably won't need to.

00:25:55 ◼   ► I haven't needed to.

00:25:56 ◼   ► And they also have created flexible subscription plans

00:25:59 ◼   ► and a diverse assortment of roasts

00:26:00 ◼   ► tailored to whatever your volume needs,

00:26:03 ◼   ► taste preferences, whatever it is.

00:26:05 ◼   ► You can of course pause it, cancel it,

00:26:07 ◼   ► change whatever you need to change anytime.

00:26:09 ◼   ► It's super easy to manage.

00:26:10 ◼   ► And it's all, all the coffee is roasted fresh to order.

00:26:14 ◼   ► I love fresh roasted coffee and Trade is an amazing way

00:26:18 ◼   ► to discover all sorts of great new beans,

00:26:21 ◼   ► great new roasters, great new roasts.

00:26:23 ◼   ► I really enjoy Trade.

00:26:25 ◼   ► So upgrade your morning routine with better coffee.

00:26:28 ◼   ► Right now, Trade is offering you a free bag of coffee

00:26:32 ◼   ► with any new subscription at drinktrade.com/atp.

00:26:36 ◼   ► That's drinktrade.com/atp for a free bag of coffee

00:26:41 ◼   ► with any subscription purchase.

00:26:42 ◼   ► Drinktrade.com/atp.

00:26:46 ◼   ► Thanks to Trade Coffee for sponsoring our show.

00:26:48 ◼   ► - So where did we leave off?

00:26:53 ◼   ► We're still in the lab with Johnny

00:26:55 ◼   ► and we're talking about how it's three nanometer.

00:26:59 ◼   ► And then something was brought up

00:27:00 ◼   ► and I genuinely don't understand what this is.

00:27:04 ◼   ► Like I understand the words that were used

00:27:06 ◼   ► in the explanation, but I don't think

00:27:07 ◼   ► I understand the application.

00:27:09 ◼   ► So Johnny starts talking about dynamic caching

00:27:11 ◼   ► and he says it's in the hardware,

00:27:13 ◼   ► it's transparent to developers.

00:27:14 ◼   ► And he starts explaining that, oh, you know,

00:27:17 ◼   ► a lot of decisions are made at compile time

00:27:20 ◼   ► and we're gonna basically do them just in time at runtime

00:27:22 ◼   ► with regard to memory allocation, specifically on the GPU.

00:27:27 ◼   ► John, do you have any idea what this is about

00:27:28 ◼   ► or Marco, either of you, can you explain it?

00:27:31 ◼   ► - In the presentation, it was very confusing.

00:27:34 ◼   ► They showed some diagrams that basically saying,

00:27:36 ◼   ► we're making more efficient use of resources,

00:27:37 ◼   ► which I think is the level you should present that

00:27:39 ◼   ► if you're gonna say anything about it at all

00:27:41 ◼   ► in this kind of presentation.

00:27:42 ◼   ► Fine, you showed me diagram,

00:27:43 ◼   ► I can look at the little boxes and say,

00:27:44 ◼   ► you're making more efficient use of resources,

00:27:46 ◼   ► thumbs up, right?

00:27:47 ◼   ► But from a technical perspective,

00:27:49 ◼   ► I didn't know what they were talking about.

00:27:51 ◼   ► - Here is a quote from Jason Snell's article

00:27:54 ◼   ► and he actually talked to Apple folks.

00:27:56 ◼   ► So I'm hoping he has a better take on it,

00:27:58 ◼   ► but I still have questions.

00:27:59 ◼   ► So we'll put a link in the show notes to his article,

00:28:01 ◼   ► quoting from Jason.

00:28:02 ◼   ► "Memory is usually allocated to different threads

00:28:05 ◼   ► at compile time,

00:28:07 ◼   ► meaning that some threads allocate a larger amount of memory

00:28:09 ◼   ► in order to handle peak need,

00:28:10 ◼   ► while other threads might choose a smaller amount of memory,

00:28:12 ◼   ► but risk a bottleneck.

00:28:14 ◼   ► The M3 graphics system dynamically allocates

00:28:16 ◼   ► the memory per thread in a way

00:28:17 ◼   ► that's completely transparent to software developers.

00:28:20 ◼   ► Apps don't need to be rewritten to take advantage

00:28:21 ◼   ► of the new system,

00:28:22 ◼   ► which Apple says makes some huge gains

00:28:23 ◼   ► by bringing a lot of memory efficiency out of the system.

00:28:26 ◼   ► Memory that was previously reserved for a specific thread

00:28:28 ◼   ► can be given to a different thread instead.

00:28:30 ◼   ► A thread that's in a bottleneck can be given more space.

00:28:33 ◼   ► It's all to the goal of increasing overall throughput."

00:28:36 ◼   ► So here's the question that I have

00:28:38 ◼   ► about this whole big thing.

00:28:40 ◼   ► Apple's architecture, as they emphasize many times,

00:28:43 ◼   ► is a unified memory architecture

00:28:44 ◼   ► where there is not a separate pool of VRAM and regular RAM.

00:28:48 ◼   ► There's just a pool of RAM and the GPU has access to it

00:28:51 ◼   ► and the CPUs have access to it.

00:28:53 ◼   ► This idea of allocating memory to threads at compile time

00:28:58 ◼   ► based on their peak need in the GPU,

00:29:01 ◼   ► there's too many words there that I don't understand.

00:29:04 ◼   ► - Right? Okay, so it's not just me.

00:29:05 ◼   ► - Yeah, so here are some possibilities.

00:29:08 ◼   ► Again, hopefully someone with knowledge

00:29:09 ◼   ► of Apple's GPU architecture will write in and tell us.

00:29:12 ◼   ► One, even though there's a unified pool of memory,

00:29:15 ◼   ► maybe the GPU, just like the CPU, has a cache hierarchy.

00:29:18 ◼   ► Or maybe it is the CPU's cache hierarchy

00:29:21 ◼   ► of level one cache, level two cache, so on and so forth.

00:29:23 ◼   ► And maybe those chunks of small pools of local cache memory

00:29:28 ◼   ► that are local to the GPU or local to the GPU cores,

00:29:32 ◼   ► maybe that's what's being statically allocated

00:29:35 ◼   ► at compile time, where when you compile something

00:29:37 ◼   ► for some code that's gonna run on the GPU,

00:29:40 ◼   ► it says, "Okay, for this thread

00:29:42 ◼   ► "where you're gonna be running,

00:29:43 ◼   ► "you're gonna need this much memory

00:29:44 ◼   ► "and what memory are they talking about?"

00:29:46 ◼   ► Some kind of tiny, very local pool of memory

00:29:50 ◼   ► inside the GPU core.

00:29:51 ◼   ► That is a possibility, 'cause GPU cores do that,

00:29:54 ◼   ► there is cache hierarchy does exist or whatever.

00:29:57 ◼   ► The second possibility is they're talking about

00:29:59 ◼   ► chunks of main memory, and somehow things that run

00:30:02 ◼   ► on the GPU, again, are hard-coded at compile time

00:30:06 ◼   ► to preallocate chunks of main memory

00:30:08 ◼   ► that they sort of have exclusive access to for their threads

00:30:11 ◼   ► so they don't thrash with that memory getting pulled.

00:30:13 ◼   ► Maybe they wire it so it doesn't get swapped out or whatever

00:30:16 ◼   ► and that's what they're using more efficiently?

00:30:18 ◼   ► I don't know, I don't know the answers to those questions.

00:30:21 ◼   ► This also could, with anything that Apple says about GPUs,

00:30:24 ◼   ► it's very possible that what they're saying

00:30:26 ◼   ► is either something that has existed in the,

00:30:29 ◼   ► quote-unquote, PC GPU space for ages,

00:30:31 ◼   ► but Apple just has a different name for it,

00:30:32 ◼   ► or they're doing something new to accomplish a task

00:30:35 ◼   ► that PC GPUs have been doing for a long time.

00:30:37 ◼   ► It's very rare that Apple does anything with their GPUs

00:30:40 ◼   ► that is unprecedented in the GPU,

00:30:42 ◼   ► where they're not the GPU leader,

00:30:43 ◼   ► but they do do interesting things

00:30:45 ◼   ► because their trade-offs are different.

00:30:46 ◼   ► So maybe this thing that they're doing is novel

00:30:50 ◼   ► in that the PC GPU world just uses brute force

00:30:52 ◼   ► and they don't care about this,

00:30:53 ◼   ► but Apple does care about more efficient use of resources.

00:30:55 ◼   ► So I understand conceptually that it's basically saying,

00:30:57 ◼   ► "Hey, your program may be written to do this,

00:31:00 ◼   ► "but that is an inefficient use of resources,"

00:31:02 ◼   ► what those resources are, we're not sure.

00:31:04 ◼   ► And so dynamically, without your program knowing,

00:31:07 ◼   ► we will shuffle things around

00:31:08 ◼   ► to make more efficient use of resources, and that's good.

00:31:11 ◼   ► So again, even though we don't understand this,

00:31:13 ◼   ► kind of like the N3E, N3B thing,

00:31:15 ◼   ► even though we don't know the details,

00:31:17 ◼   ► it all sounds good, right?

00:31:18 ◼   ► More efficient use of resources is better.

00:31:22 ◼   ► We can't really, I was thinking about this

00:31:25 ◼   ► when we were watching the presentation.

00:31:27 ◼   ► We're gonna talk all about this presentation

00:31:29 ◼   ► that Apple gave, and Apple did provide

00:31:30 ◼   ► a bunch of performance charts,

00:31:32 ◼   ► but when I was thinking about what we were gonna discuss,

00:31:33 ◼   ► I'm like, "Well, of course we can't talk about performance."

00:31:36 ◼   ► And the reason I say, "Of course we can't talk

00:31:37 ◼   ► "about performance," is the only thing we have to go on

00:31:39 ◼   ► are Apple's charts that have like, they're so abstract

00:31:43 ◼   ► that they might as well, they're not Bezos charts,

00:31:45 ◼   ► 'cause there is like, there are labels on the axes, kind of,

00:31:48 ◼   ► but they're real close.

00:31:49 ◼   ► I feel like we can't talk about performance

00:31:51 ◼   ► until people actually get these and start testing them

00:31:53 ◼   ► and benchmarking them.

00:31:55 ◼   ► Because Apple's claims, it's not that they're useless,

00:31:59 ◼   ► it's good to know what they think in the ballpark,

00:32:01 ◼   ► but with respect to dynamic caching,

00:32:04 ◼   ► that's all well and good, cool presentation.

00:32:08 ◼   ► What does this mean for the performance of this GPU

00:32:11 ◼   ► in a given task?

00:32:12 ◼   ► Does it run games at a higher frame rate?

00:32:14 ◼   ► Does it do some rendering thing in some video app faster?

00:32:17 ◼   ► Like, I don't care that much about the details

00:32:22 ◼   ► of the tech stuff in terms of like, a presentation

00:32:25 ◼   ► meant for the general public, unless it provides something.

00:32:29 ◼   ► And they didn't say, "Hey, now that we've done

00:32:31 ◼   ► "dynamic caching, you get an hour extra battery life

00:32:33 ◼   ► "when playing Resident Evil Village on your MacBook Pro,"

00:32:36 ◼   ► or something like that, because we make more efficient use

00:32:37 ◼   ► of resources so it burns less power to play the game

00:32:41 ◼   ► at the same frame rate, or something like that.

00:32:42 ◼   ► They didn't make any claims like that, really.

00:32:45 ◼   ► So I say the jury's still out on dynamic caching.

00:32:47 ◼   ► Seems cool, I would love to see it in a WWDC presentation,

00:32:49 ◼   ► like that's the level you would expect to see

00:32:51 ◼   ► all these little block diagrams and the dynamic allocation,

00:32:54 ◼   ► but it was kind of weird to see it in this presentation.

00:32:56 ◼   ► I get that they're proud of it and it's a new feature

00:32:58 ◼   ► of their GPU, but all I wanna know is,

00:33:00 ◼   ► show me the frame rate, like what is this doing for me?

00:33:03 ◼   ► - So let's talk about the different CPUs,

00:33:07 ◼   ► themselves, we start with the M1, the CPU cores--

00:33:11 ◼   ► - M3. - I'm sorry, sorry, M3,

00:33:14 ◼   ► that's what I meant.

00:33:15 ◼   ► I was reading ahead, the CPU cores, as compared to the M1,

00:33:19 ◼   ► there we go, the performance cores are 30% faster,

00:33:22 ◼   ► the efficiency cores are up to 50% faster.

00:33:27 ◼   ► - That was a good time to mention that Apple loves

00:33:29 ◼   ► to compare this chip to the M1.

00:33:30 ◼   ► - Yeah, this entire presentation was comparing it

00:33:33 ◼   ► to the M1 mostly, and then, oh, also the M2.

00:33:37 ◼   ► Like it was a very, which actually, I love that,

00:33:40 ◼   ► because here I was, I skipped the M2, and so I'm like--

00:33:44 ◼   ► - It's a marker, it's just for you, Marker, yeah.

00:33:46 ◼   ► - No, it was for both of us, it was the ATP presentation,

00:33:48 ◼   ► because I am also sitting on an M1 Max MacBook Pro.

00:33:52 ◼   ► - Yeah, so I get by, and I've said this on Mastodon,

00:33:54 ◼   ► people were like, well, but a lot of people are,

00:33:56 ◼   ► not everyone's upgrading from the M2,

00:33:58 ◼   ► like that's why they do it, like, I don't object to them

00:34:00 ◼   ► comparing it to the M1, I object to them not comparing it

00:34:02 ◼   ► to the M2, right?

00:34:04 ◼   ► - Well, they did both, ultimately,

00:34:05 ◼   ► like they did show, for most of the performance comparisons

00:34:09 ◼   ► they made, they showed M1 and M2 comparisons.

00:34:12 ◼   ► - Yeah, they showed it on the graph,

00:34:14 ◼   ► but they didn't say it as much, or didn't emphasize it.

00:34:16 ◼   ► Sometimes they would just say the M1,

00:34:18 ◼   ► sometimes they would say both of them,

00:34:19 ◼   ► but here's the reason why comparing to the M2 is important,

00:34:22 ◼   ► even if you don't own an M2, because when Apple releases

00:34:26 ◼   ► a new line of computers, especially in the Pro line,

00:34:28 ◼   ► where they're emphasizing performance, like scary fast,

00:34:30 ◼   ► it's the theme here, even if you don't own an M2,

00:34:34 ◼   ► and you just own an M1, what you wanna know,

00:34:37 ◼   ► as a buyer, as a tech nerd buyer, anyway,

00:34:39 ◼   ► is how big of a leap is this new line of computers

00:34:44 ◼   ► over its predecessor?

00:34:45 ◼   ► Like, is this the generation I should buy, right?

00:34:47 ◼   ► 'Cause let's say the M2 came out,

00:34:49 ◼   ► and it was 1% better than the M1.

00:34:52 ◼   ► You don't care how much better it is

00:34:53 ◼   ► than like the past Intel one, of course, Apple,

00:34:55 ◼   ► compared stuff to the Intel as well.

00:34:57 ◼   ► You care, like, is this generation like a big leap

00:35:00 ◼   ► over the previous one, is this the time to buy,

00:35:02 ◼   ► or should I wait, is this a generation

00:35:04 ◼   ► that's just kind of like, ah, it's a really incremental

00:35:06 ◼   ► update, maybe you should wait?

00:35:07 ◼   ► And you're not waiting because it's not faster

00:35:09 ◼   ► than what you have, like say you still have Intel, right?

00:35:12 ◼   ► You're waiting because you think, ah, this one seems like

00:35:15 ◼   ► it's not as big a leap over its predecessor

00:35:18 ◼   ► as it could have been, so maybe I should wait.

00:35:19 ◼   ► That's why comparison of the M2 is important.

00:35:21 ◼   ► We're just saying, how well did you do, Apple,

00:35:23 ◼   ► in this new generation?

00:35:25 ◼   ► Is it the same, you know, the distance between the M1

00:35:28 ◼   ► and M2, is the distance between the M2 and the M3 the same?

00:35:31 ◼   ► Is that bigger or smaller, right?

00:35:33 ◼   ► And I think that is a very relevant piece of information

00:35:36 ◼   ► regardless of what you currently own.

00:35:38 ◼   ► Because I think most people, like, everyone doesn't upgrade

00:35:42 ◼   ► every year, right, most people have computers

00:35:44 ◼   ► that are so old that no matter what they get,

00:35:46 ◼   ► it's gonna be fantastically faster.

00:35:48 ◼   ► Like, I'm in that situation and I'm a tech nerd

00:35:51 ◼   ► because I buy these really expensive computers,

00:35:52 ◼   ► it takes a long time to save up this money, right?

00:35:54 ◼   ► But like, you're not questioning whether it's gonna be

00:35:56 ◼   ► faster than what you have, you know if you buy anything,

00:35:58 ◼   ► you know, if you buy like a low-end MacBook Air,

00:36:00 ◼   ► it'll be better than your Intel iMac or something.

00:36:02 ◼   ► What you wanna know as a tech nerd is like,

00:36:04 ◼   ► is this the one to buy?

00:36:05 ◼   ► They really knocked it out of the park this year.

00:36:07 ◼   ► And by knocking it out of the park this year,

00:36:09 ◼   ► what you mean is how much faster is it than the M2?

00:36:11 ◼   ► And that's why I wish they didn't just show the M2 on slides

00:36:16 ◼   ► but also made it the primary source of comparison.

00:36:19 ◼   ► And then I feel like you can say, they can say, you know,

00:36:22 ◼   ► we did, you know, X percent better than last year.

00:36:25 ◼   ► And then they can say, and by the way,

00:36:27 ◼   ► if you've still got an old M1,

00:36:28 ◼   ► look how much better we are.

00:36:30 ◼   ► Like, like ramp it up to that, you know,

00:36:32 ◼   ► don't not compare to the M1, do,

00:36:34 ◼   ► but that's like, that's your cherry on top.

00:36:36 ◼   ► I really wanna know how it compares to the M2.

00:36:37 ◼   ► Anyway, they did put it in the graphs most of the time.

00:36:39 ◼   ► And here's the thing, very often those graphs

00:36:41 ◼   ► were impressive versus the M2.

00:36:43 ◼   ► It's not like they're trying to hide something like,

00:36:44 ◼   ► oh, we can only compare to the M1

00:36:45 ◼   ► 'cause it's not that much better than the M2.

00:36:47 ◼   ► No, very often it was better than the M2

00:36:49 ◼   ► by a significant amount.

00:36:50 ◼   ► So weird decision.

00:36:52 ◼   ► - Continuing with the, the vanilla M3,

00:36:55 ◼   ► same multi-threaded performance as the M1

00:36:57 ◼   ► but at half the power.

00:36:58 ◼   ► It's one quarter the power of what they claim

00:37:01 ◼   ► to be an equivalent PC chip or Intel chip.

00:37:04 ◼   ► It's one fifth the power. - Whatever that means.

00:37:06 ◼   ► - Yeah, exactly.

00:37:07 ◼   ► I think they did actually have

00:37:08 ◼   ► in the bottom right hand corner what they were using.

00:37:10 ◼   ► And I think it was an Intel i7, I wanna say,

00:37:12 ◼   ► but I don't remember for sure.

00:37:14 ◼   ► It was, it's one fifth the power of the equivalent PC,

00:37:18 ◼   ► sorry, it uses one fifth the power

00:37:20 ◼   ► of the equivalent PC GPU.

00:37:22 ◼   ► It's got a 16 core neural engine,

00:37:23 ◼   ► engine which is 60% faster than M1.

00:37:26 ◼   ► And then I didn't understand for a moment

00:37:29 ◼   ► what they were talking about.

00:37:30 ◼   ► They said the media engine adds AV1.

00:37:32 ◼   ► And I was like, man, I've heard of this.

00:37:33 ◼   ► Why do I know this?

00:37:34 ◼   ► And then they continued for YouTube and Netflix

00:37:35 ◼   ► and other things like that.

00:37:36 ◼   ► Oh, okay, that makes sense.

00:37:38 ◼   ► So M3, eight CPU cores for performance for efficiency,

00:37:42 ◼   ► which is the same quantities as the M2.

00:37:45 ◼   ► 10 GPU cores or eight if you get the cheaper version,

00:37:48 ◼   ► same as on the M2.

00:37:50 ◼   ► 25 billion transistors,

00:37:52 ◼   ► which allegedly is five billion more than the M2.

00:37:56 ◼   ► And eight to 24 gigs RAM, same as the M2.

00:37:58 ◼   ► But, and Marco, future Marco,

00:38:01 ◼   ► feel free to edit in the sad trombone here.

00:38:04 ◼   ► Only one external display, wamp, wamp.

00:38:07 ◼   ► - Yeah, I don't understand that at all.

00:38:08 ◼   ► Like, I mean, it being the same makes sense.

00:38:11 ◼   ► It's the low end one.

00:38:12 ◼   ► I think they think for performance and for efficiency

00:38:14 ◼   ► is a reasonable balance.

00:38:15 ◼   ► It's a small GPU 'cause it's the small, cheap chip.

00:38:18 ◼   ► Where did the extra five billion transistors come from?

00:38:20 ◼   ► Remember that these are not the same GPU cores, right?

00:38:23 ◼   ► They're the new fancy ones with dynamic caching

00:38:26 ◼   ► and the ray tracing stuff and all sorts of things.

00:38:28 ◼   ► So, you know, I'm assuming that's where

00:38:29 ◼   ► a lot of the transistors come from.

00:38:30 ◼   ► And also the CPU cores are not the same either.

00:38:33 ◼   ► They're a little bit better, as you just noted.

00:38:34 ◼   ► Like they're, you know, they're more efficient.

00:38:37 ◼   ► They can do the same work with less power

00:38:39 ◼   ► and they have a higher top end.

00:38:41 ◼   ► When you're in there making your brand new M3 chip

00:38:45 ◼   ► and you've had two generations where people have been annoyed

00:38:48 ◼   ► by the fact that there's only one external display supported,

00:38:51 ◼   ► this would be the time to add it.

00:38:52 ◼   ► Like when the M2 came out, we excused it and said,

00:38:54 ◼   ► well, you know, yeah, one external display,

00:38:56 ◼   ► but maybe they didn't realize that people would be cranky

00:38:58 ◼   ► about it and, you know, the M2 had to be planned.

00:39:00 ◼   ► They didn't have enough time to make this change

00:39:01 ◼   ► for the M2, but maybe in the M3 generation.

00:39:03 ◼   ► And the answer is no, not in the M3 generation.

00:39:06 ◼   ► So maybe Apple thinks this is fine

00:39:07 ◼   ► and that you should just buy a better chip

00:39:10 ◼   ► and not get the plain old M3

00:39:12 ◼   ► if you need more than one external display.

00:39:14 ◼   ► And yeah, there are also workarounds

00:39:15 ◼   ► with all sorts of USB-C things and stuff like that.

00:39:17 ◼   ► It just, it seems like an odd choice.

00:39:19 ◼   ► It seems like it's gotta be driven by, you know,

00:39:23 ◼   ► the size and the power draw and the space and the cost

00:39:26 ◼   ► of supporting more external displays

00:39:28 ◼   ► in the lowest of low-end chips.

00:39:29 ◼   ► And they really want this to be able to extend down low.

00:39:32 ◼   ► Maybe they wanna make sure they can jam it

00:39:35 ◼   ► into a Vision Pro in the future, for example,

00:39:37 ◼   ► like a context where external display support

00:39:40 ◼   ► doesn't make any sense and why would you burn

00:39:42 ◼   ► the chip space and power budget on that?

00:39:45 ◼   ► So I don't know exactly what happened here,

00:39:47 ◼   ► but that is the one disappointing aspect

00:39:50 ◼   ► of the plain old M3.

00:39:51 ◼   ► - I think at this point, you know,

00:39:53 ◼   ► there is some element of a trade-off

00:39:57 ◼   ► of how you spend your limited die space on the CPU die.

00:40:01 ◼   ► There's a whole bunch of stuff that has to go in here.

00:40:03 ◼   ► Display controllers and Thunderbolt controllers

00:40:05 ◼   ► are not small, I mean, they're not massive,

00:40:08 ◼   ► but they do take up space on the die.

00:40:10 ◼   ► And so it's a trade-off.

00:40:11 ◼   ► When you're making this smaller chip made for lower-end,

00:40:14 ◼   ► smaller, cooler-running products,

00:40:16 ◼   ► there's only so much die space you have,

00:40:18 ◼   ► you gotta choose the trade-off for different things.

00:40:20 ◼   ► But I think at this point, three generations later,

00:40:24 ◼   ► this seems a lot more like intentional market segmentation.

00:40:26 ◼   ► And I don't think that's totally unreasonable.

00:40:30 ◼   ► - It's a little bit unreasonable, given the fact

00:40:32 ◼   ► that this is, as we'll say in a little bit,

00:40:34 ◼   ► it's in a MacBook Pro now.

00:40:36 ◼   ► - Yes, but that new MacBook Pro has a few asterisks on it,

00:40:40 ◼   ► which, and we'll get to that in a second,

00:40:41 ◼   ► I'm actually very happy about that product overall,

00:40:43 ◼   ► but at this point, they're very clearly saying,

00:40:46 ◼   ► if you're gonna drive more than one external display,

00:40:49 ◼   ► you're gonna need the Pro or Max chips.

00:40:51 ◼   ► And I don't think that's totally unreasonable.

00:40:55 ◼   ► I do think if you're holding out hope

00:40:57 ◼   ► that maybe the M4 will support two external displays,

00:41:00 ◼   ► I think you can probably stop holding out that hope.

00:41:02 ◼   ► - Well, so here's the thing.

00:41:04 ◼   ► I have heard some rumors that the M3 was originally planned

00:41:08 ◼   ► to support more than one display.

00:41:10 ◼   ► And obviously, that plan changed.

00:41:11 ◼   ► Plans change all the time.

00:41:12 ◼   ► But the fact that it was planned to support more than one

00:41:14 ◼   ► shows that someone thought that maybe the M3 generation

00:41:18 ◼   ► would be the time to reconsider that decision.

00:41:20 ◼   ► But then they reconsidered it back, which makes me think,

00:41:23 ◼   ► that's why I was thinking about the Vision Pro.

00:41:24 ◼   ► Like, in terms of the trade-offs or whatever,

00:41:27 ◼   ► what is their most important product

00:41:29 ◼   ► where it's really important

00:41:32 ◼   ► that the M3 be able to extend low?

00:41:34 ◼   ► I don't think it's the MacBook Air.

00:41:35 ◼   ► I think it's the Vision Pro at this point,

00:41:37 ◼   ► because you really don't,

00:41:38 ◼   ► like the power budget on the Vision Pro is harsh.

00:41:41 ◼   ► It's, even with that external battery back,

00:41:42 ◼   ► two hours of battery life is not great.

00:41:44 ◼   ► That is a great reason to say,

00:41:47 ◼   ► look, we need the M3 to go all the way down

00:41:50 ◼   ► to our face-mounted thing with a tiny battery.

00:41:53 ◼   ► So remember we planned more than one display?

00:41:55 ◼   ► No, that gets cut.

00:41:56 ◼   ► We can't afford the power budget, the space,

00:41:59 ◼   ► anything like that.

00:42:00 ◼   ► So who knows?

00:42:01 ◼   ► I mean, every year we'll keep checking.

00:42:02 ◼   ► But I mean, if they're gonna have a limitation like this,

00:42:05 ◼   ► you're right that the lowest end chip it makes sense on,

00:42:07 ◼   ► but this M3 is being asked to do a lot.

00:42:09 ◼   ► I'm assuming it will eventually be in the Vision Pro

00:42:12 ◼   ► unless they have an all-new chip for that

00:42:13 ◼   ► with a V in the name or something in the future,

00:42:16 ◼   ► which means it has to go all the way from

00:42:18 ◼   ► the head-mounted thing with a tiny battery

00:42:20 ◼   ► all the way up to the lowest end MacBook Pro.

00:42:22 ◼   ► And that is a tough task for this chip.

00:42:26 ◼   ► - Yeah, it's a bummer though.

00:42:27 ◼   ► I concur with what Marco was saying that

00:42:30 ◼   ► it certainly smells to me like market segmentation,

00:42:33 ◼   ► which stinks, like I get it, that's there, right.

00:42:36 ◼   ► But I know there's been a handful of times

00:42:39 ◼   ► that I've been talking to people

00:42:40 ◼   ► about what Mac they should get.

00:42:42 ◼   ► And even if it's a non-technical person,

00:42:46 ◼   ► I always have to ask, well, what are you plugging this into?

00:42:49 ◼   ► Is this getting plugged into HDMI?

00:42:52 ◼   ► 'Cause that may make you want a MacBook Pro.

00:42:54 ◼   ► Is this getting plugged into multiple monitors?

00:42:56 ◼   ► Well, then you definitely don't want

00:42:58 ◼   ► like a MacBook Air, for example,

00:42:59 ◼   ► even though from a compute standpoint,

00:43:01 ◼   ► it would have been more than sufficient, et cetera, et cetera.

00:43:03 ◼   ► So that's a little bit of a bummer.

00:43:05 ◼   ► - Yeah, I mean, speaking of those caveats,

00:43:07 ◼   ► thought about this one, we'll get to the line

00:43:10 ◼   ► of the computers in a second,

00:43:11 ◼   ► we're still just talking about the CPUs,

00:43:13 ◼   ► but that's kind of the way, now that,

00:43:15 ◼   ► especially the Apple's laptop line is in such good shape,

00:43:18 ◼   ► the way I'm measuring how well they're doing

00:43:21 ◼   ► is no longer saying like, does the keyboard work?

00:43:23 ◼   ► Does it have a reasonable number of ports?

00:43:24 ◼   ► Is the performance good?

00:43:25 ◼   ► Like that's all good.

00:43:26 ◼   ► - Yeah, the basics are taken care of.

00:43:28 ◼   ► - Yeah, now it's like, is there anything

00:43:30 ◼   ► I need to warn people about, to Casey's point?

00:43:32 ◼   ► Like, you know, 'cause we feel so great about these computers

00:43:35 ◼   ► like if you get one, you're getting a great computer,

00:43:37 ◼   ► don't worry about it.

00:43:38 ◼   ► Like we said the same thing about the M2s,

00:43:39 ◼   ► we said, you know, the MacBook Pros

00:43:42 ◼   ► were not in desperate need of an update.

00:43:43 ◼   ► We love that they got updated,

00:43:44 ◼   ► but they weren't desperate need of it

00:43:45 ◼   ► because the M2 ones were already great.

00:43:47 ◼   ► But this, the whole like, oh, wait, wait, wait,

00:43:50 ◼   ► are you gonna use external displays?

00:43:51 ◼   ► And like I said, you can use more than one external display

00:43:54 ◼   ► through various USB-C shenanigans

00:43:55 ◼   ► that we've talked about on past shows.

00:43:56 ◼   ► It's just a complication that if someone is asking you

00:43:59 ◼   ► about what computer to buy, you should ask them about it.

00:44:02 ◼   ► Oh, do you plan on using two external displays?

00:44:06 ◼   ► How are they connected or whatever?

00:44:07 ◼   ► And the second thing is the whole one SSD chip versus two

00:44:12 ◼   ► that has the speed of your SSD.

00:44:14 ◼   ► I don't yet know if they're still doing that

00:44:16 ◼   ► in this generation, but I would assume they are.

00:44:19 ◼   ► So that's also a bummer.

00:44:20 ◼   ► And that's the list of like,

00:44:22 ◼   ► on this amazing line of computers we all think are great,

00:44:25 ◼   ► the two things we need to ask people is

00:44:26 ◼   ► if you're getting M3, do you have more than one display?

00:44:28 ◼   ► And if you're planning on buying any of these,

00:44:31 ◼   ► if you're buying one of the models

00:44:32 ◼   ► that has the single chip SSD,

00:44:35 ◼   ► that's worth at least having a conversation about.

00:44:36 ◼   ► It doesn't mean you shouldn't buy it and it's terrible,

00:44:38 ◼   ► like maybe you don't care, but it's not like 1% slower.

00:44:42 ◼   ► It's like half the speed.

00:44:43 ◼   ► So it's worth discussing.

00:44:45 ◼   ► - Indeed.

00:44:46 ◼   ► So let's talk about what you would do

00:44:47 ◼   ► if you were to go the next step up.

00:44:49 ◼   ► Let's talk M3 Pro.

00:44:51 ◼   ► So the M3 Pro has 12 CPU cores.

00:44:53 ◼   ► That's four more than the M3.

00:44:56 ◼   ► That's six and six, so two and two more than the M3.

00:45:00 ◼   ► 18 GPU cores, which is eight more than the vanilla M3.

00:45:05 ◼   ► It's also 14 on the cheaper version,

00:45:09 ◼   ► but both are one fewer than the M2 Pro,

00:45:12 ◼   ► which was 19 and 16, which is a little bit different.

00:45:15 ◼   ► So, or I guess it's one fewer and two fewer

00:45:18 ◼   ► than the M2 Pro.

00:45:19 ◼   ► Nonetheless, 40% faster than the M1 Pro.

00:45:23 ◼   ► It's 37 billion transistors,

00:45:26 ◼   ► three billion fewer than the M2 Pro,

00:45:29 ◼   ► which had an even 40 billion,

00:45:30 ◼   ► which had eight perf cores, four efficiency cores,

00:45:33 ◼   ► and up to 19 GPU cores.

00:45:35 ◼   ► Max RAM, 36 gigs, up from 32 on the M2 Pro.

00:45:39 ◼   ► - This is a weird chip.

00:45:41 ◼   ► Why does it have six efficiency cores?

00:45:43 ◼   ► None of the other M series chips, M3 series chips,

00:45:47 ◼   ► have six efficiency cores.

00:45:48 ◼   ► The little one has what, four?

00:45:50 ◼   ► - Four. - And the big one,

00:45:51 ◼   ► the max, has four.

00:45:53 ◼   ► The M3 Pro has six.

00:45:55 ◼   ► And again, fewer transistors than its predecessor.

00:45:59 ◼   ► Part of that fewer transistors is probably due to the fact

00:46:02 ◼   ► that it has fewer power cores, right?

00:46:05 ◼   ► And then more efficiency cores.

00:46:07 ◼   ► I would have thought that with this arrangement

00:46:10 ◼   ► that Apple would have made, you know, had an explanation.

00:46:12 ◼   ► Here's why we did it this way.

00:46:14 ◼   ► 'Cause I can think of reasons, like, you know,

00:46:15 ◼   ► the efficiency cores, they emphasize,

00:46:17 ◼   ► the efficiency cores keep getting better and better.

00:46:18 ◼   ► If you're thinking of these as crappy little cores,

00:46:21 ◼   ► they're not.

00:46:21 ◼   ► Like, every year, the efficiency cores

00:46:23 ◼   ► get better and better.

00:46:25 ◼   ► They haven't done the comparison yet,

00:46:26 ◼   ► but I imagine there's gonna be a point

00:46:27 ◼   ► where the efficiency cores are more powerful

00:46:28 ◼   ► than the power cores in some past M line chip.

00:46:31 ◼   ► So maybe having six efficiency cores and six performance cores

00:46:34 ◼   ► is a better trade-off for this middle thing,

00:46:37 ◼   ► you know what I mean?

00:46:37 ◼   ► 'Cause you get more parallelism,

00:46:39 ◼   ► 'cause now you've got 12 cores.

00:46:41 ◼   ► I don't know, Apple didn't give an explanation,

00:46:43 ◼   ► but it is notable.

00:46:44 ◼   ► And it's also notable when we get to the max,

00:46:46 ◼   ► is that, I meant to look this up,

00:46:48 ◼   ► 'cause my memory's getting so bad,

00:46:49 ◼   ► but in the earlier M lines, wasn't the fact

00:46:52 ◼   ► that the Pro was basically just the max

00:46:54 ◼   ► with GPU cores cut off?

00:46:55 ◼   ► Which generations did that?

00:46:56 ◼   ► Was it just the M1, or was it M1 and M2?

00:46:59 ◼   ► - I think it was both.

00:47:00 ◼   ► And that's what's interesting here,

00:47:01 ◼   ► is the M3 Pro and the M3 Max

00:47:05 ◼   ► are basically nothing alike from their die design.

00:47:09 ◼   ► It's totally different.

00:47:10 ◼   ► - Yeah, one is not the other with things cut off.

00:47:13 ◼   ► They're separate chips.

00:47:15 ◼   ► - If anything, the M3 Pro looks a lot more like the M3

00:47:19 ◼   ► than the M3 Max.

00:47:21 ◼   ► It's closer to that.

00:47:22 ◼   ► - Yeah, and the uniformity in that it's got

00:47:24 ◼   ► four performance, four efficiency, M3,

00:47:26 ◼   ► and six and six on the M3 Pro,

00:47:28 ◼   ► so there's a symmetry there.

00:47:30 ◼   ► I'll have to look at the die shot

00:47:31 ◼   ► to see if there's any family resemblance.

00:47:33 ◼   ► But this, again, I like this.

00:47:35 ◼   ► I understood, oh, isn't it cool

00:47:38 ◼   ► they can just sell the Pro and the Max

00:47:40 ◼   ► and they just chop off the cores or whatever?

00:47:41 ◼   ► But I like three separate chips,

00:47:43 ◼   ► where they say we've got our low on one

00:47:45 ◼   ► that has to do these jobs,

00:47:46 ◼   ► we've got the middle one that most people will buy,

00:47:47 ◼   ► and we've got the big one.

00:47:49 ◼   ► And I encourage Apple to do exactly this,

00:47:52 ◼   ► which is make a separate chip for all three of those.

00:47:54 ◼   ► We're not asking a lot.

00:47:55 ◼   ► Obviously, you share cores and stuff like that.

00:47:58 ◼   ► And it seems that they've done that.

00:47:59 ◼   ► My only question is, why is six and six

00:48:00 ◼   ► the right choice for this middle one?

00:48:03 ◼   ► Again, we'll find out in the benchmarks.

00:48:04 ◼   ► Maybe it shows, oh, this is a great trade-off

00:48:05 ◼   ► because it's the one most people buy,

00:48:08 ◼   ► and it has better battery life, for example,

00:48:11 ◼   ► because it has more efficiency cores.

00:48:13 ◼   ► And the efficiency cores are more powerful now,

00:48:14 ◼   ► and you rarely use the performance.

00:48:16 ◼   ► Whatever the explanation is,

00:48:17 ◼   ► I'm willing to believe that this is a good arrangement,

00:48:19 ◼   ► but I love the fact that it is its own chip now.

00:48:22 ◼   ► - I think when you look at the relative sizes

00:48:26 ◼   ► and capabilities of these chips now,

00:48:29 ◼   ► again, the M3 Pro is a lot closer to the M3

00:48:32 ◼   ► than it is to the M3 Max.

00:48:33 ◼   ► I think what we might see now is more products

00:48:37 ◼   ► being offered with M3 and M3 Pro options.

00:48:41 ◼   ► So for instance, the next generation of Mac Mini,

00:48:43 ◼   ► if it gets this, then maybe you'll see that.

00:48:46 ◼   ► Maybe, again, maybe you'll see the M3 Pro appear

00:48:48 ◼   ► in the next 15-inch MacBook Air as an option.

00:48:52 ◼   ► - Yeah, that'd be cool.

00:48:52 ◼   ► - Because it's really not that much bigger than the M3.

00:48:55 ◼   ► And again, many of the capabilities look like they scaled up

00:48:59 ◼   ► the M3 for the M3 Pro, and then the M3 Max

00:49:01 ◼   ► is this totally different design.

00:49:03 ◼   ► We might start seeing these lines get shifted around

00:49:05 ◼   ► as to what products get which of these chip options.

00:49:08 ◼   ► The M3 Pro sure looks like it was designed

00:49:11 ◼   ► to be in more products than the M1 and M2 Pros were.

00:49:14 ◼   ► - Yeah, and like I said, I think it does seem like,

00:49:17 ◼   ► on paper, a good middle chip.

00:49:20 ◼   ► Because I think it is differentiated enough from the M3,

00:49:24 ◼   ► if only from supporting more than one display.

00:49:26 ◼   ► But 12 cores versus eight is a big difference.

00:49:29 ◼   ► But it is not a giant monster chip like the Max.

00:49:32 ◼   ► And I really do think this is the one

00:49:34 ◼   ► that most people will buy, because the M3

00:49:37 ◼   ► is clearly the bottom of the bottom line,

00:49:39 ◼   ► and it has a bunch of compromises or whatever.

00:49:41 ◼   ► And the M3 Max is really expensive and big.

00:49:43 ◼   ► So if you're getting a MacBook Pro,

00:49:45 ◼   ► when I say most people, I mean most people

00:49:46 ◼   ► who are buying MacBook Pros.

00:49:47 ◼   ► When you're getting a MacBook Pro, I think people will,

00:49:49 ◼   ► just from price anchoring, grab towards the middle option,

00:49:52 ◼   ► and this is the middle option.

00:49:53 ◼   ► I hope it is good, I hope this middle option is good.

00:49:54 ◼   ► We'll see, again, we'll see when people

00:49:56 ◼   ► start benchmarking this and checking at the thermals

00:49:58 ◼   ► and all this stuff like that.

00:49:59 ◼   ► I hope this is a good compromise,

00:50:00 ◼   ► but I like the fact that it is its own thing.

00:50:03 ◼   ► And also, the Max RAM being 36.

00:50:07 ◼   ► We'll get to RAM in a little bit

00:50:08 ◼   ► when we talk about the Max chip.

00:50:10 ◼   ► But some weird amounts here, for reasons that,

00:50:13 ◼   ► again, we'll explain, but instead of 1632, you've got 1836.

00:50:18 ◼   ► - Right, 'cause there are multiples of three,

00:50:20 ◼   ► 'cause there's three sets of memory controllers.

00:50:22 ◼   ► - Yeah, yeah, but anyway, that's two more in each measure.

00:50:26 ◼   ► I like that, we so rarely get this.

00:50:28 ◼   ► It's always like, oh, well, you can only go up by doubling.

00:50:30 ◼   ► It's either gonna be 16, 32, 64, and it's like,

00:50:32 ◼   ► well, this year, on this machine,

00:50:36 ◼   ► we're not moving up from 16 to 32,

00:50:38 ◼   ► so you're just gonna be stuck on 16

00:50:39 ◼   ► until you get enough activation energy to make the leap to 32.

00:50:43 ◼   ► I love going from 32 to 36, that's great.

00:50:45 ◼   ► (laughing)

00:50:46 ◼   ► I like the idea of buying a newer computer

00:50:49 ◼   ► and it having more stuff than the previous one,

00:50:51 ◼   ► even if it's not double.

00:50:52 ◼   ► So I know, again, there's reasons for that

00:50:54 ◼   ► in terms of the memory chips or whatever,

00:50:56 ◼   ► but I give that a thumbs up even if it was just an accident.

00:50:59 ◼   ► - All right, and then let's talk about the M3 Max, baby.

00:51:03 ◼   ► 16 cores, 12 performance, which is four more

00:51:07 ◼   ► than the M2 Max, and four efficiency.

00:51:09 ◼   ► So fewer efficiency cores than the M3 Pro,

00:51:12 ◼   ► but obviously more performance cores.

00:51:15 ◼   ► 40 GPU cores as compared to 18, or 30 instead of 14

00:51:20 ◼   ► for the cheaper one.

00:51:22 ◼   ► So the M2 Max was 38 and 30, now we're 40 and 30.

00:51:27 ◼   ► Up to 128 gigabytes of RAM, the previous limit was 96.

00:51:32 ◼   ► And here's where the computer initially got sold

00:51:37 ◼   ► to one Casey Liss, up to 80% faster than the M1 Max.

00:51:42 ◼   ► My goodness, 92 billion transistors,

00:51:45 ◼   ► 25 billion more than the M2 Max,

00:51:47 ◼   ► which is 67 billion. - That's a lot.

00:51:49 ◼   ► That is a lot of Bs, a lot of Bs.

00:51:51 ◼   ► - That's a big increase year over year.

00:51:54 ◼   ► This Max is not just like an incremental increase

00:51:58 ◼   ► over the previous one, it is,

00:52:00 ◼   ► like this is separating itself from the rest of the line.

00:52:02 ◼   ► Like four more performance cores,

00:52:04 ◼   ► 25 billion more transistors.

00:52:06 ◼   ► The GPU cores, even though there's only two more of them,

00:52:08 ◼   ► again, they're different GPU cores.

00:52:10 ◼   ► 128 gigs of RAM is great because if you squint at it,

00:52:13 ◼   ► you can say the Ultra's gonna have 256 as the Max,

00:52:16 ◼   ► which is, you know, wonderful.

00:52:18 ◼   ► I think the minimum RAM is 36.

00:52:21 ◼   ► I mean, I guess now is the time to jump

00:52:22 ◼   ► into the RAM things for the Max because there's,

00:52:25 ◼   ► well, we'll do the RAM things for all the things in a second.

00:52:28 ◼   ► But yeah, let's just talk about the Max.

00:52:30 ◼   ► I heartily approve of this chip.

00:52:33 ◼   ► If it is not a giant power and heap destroying monster,

00:52:36 ◼   ► which I suspect it is not, boy, this is great.

00:52:39 ◼   ► And this really does make people

00:52:41 ◼   ► with M1s and M2s look twice, I think.

00:52:44 ◼   ► - Oh yeah, that's what got me too.

00:52:46 ◼   ► (laughs)

00:52:47 ◼   ► - We'll get to that, but yeah.

00:52:49 ◼   ► This is-- - It's incredibly impressive.

00:52:50 ◼   ► - Yeah, especially to go,

00:52:53 ◼   ► you know, they had a process shrink here

00:52:55 ◼   ► from five nanometers to three,

00:52:57 ◼   ► and so you would expect that you're able

00:52:59 ◼   ► to cram in more stuff with the same general size

00:53:03 ◼   ► class of chip.

00:53:04 ◼   ► And to go from eight performance cores to 12,

00:53:08 ◼   ► that's a huge increase.

00:53:10 ◼   ► That's amazing.

00:53:11 ◼   ► 50% more performance cores,

00:53:13 ◼   ► and to have each of the cores be more powerful

00:53:16 ◼   ► than the previous ones by something like 15%,

00:53:20 ◼   ► that's really good.

00:53:21 ◼   ► That's a nice, you know,

00:53:23 ◼   ► when we sit around as nerds and we watch the process sizes

00:53:27 ◼   ► from the fabs and we're like, oh man,

00:53:29 ◼   ► we can't wait until the next process shrink

00:53:31 ◼   ► 'cause it'll be a big jump.

00:53:32 ◼   ► And in this case, I think it was.

00:53:35 ◼   ► I think this shows you that this was a big jump.

00:53:37 ◼   ► This was worth, like if you had M1

00:53:40 ◼   ► or if you didn't even get that,

00:53:41 ◼   ► and if you kind of skipped M2,

00:53:43 ◼   ► 'cause you're like, well, M3 might be a process shrink.

00:53:46 ◼   ► This is the kind of jump you were waiting for.

00:53:49 ◼   ► And that's fantastic.

00:53:51 ◼   ► - Well, they didn't give the size of the die in millimeters,

00:53:54 ◼   ► did they though?

00:53:55 ◼   ► 'Cause that's the question.

00:53:56 ◼   ► Have they just simply made a bigger chip,

00:53:58 ◼   ► which again, I still approve of,

00:53:59 ◼   ► but I'm not ready to put a Laurel wreath

00:54:04 ◼   ► around the three nanometer process quite yet.

00:54:06 ◼   ► 'Cause if they just simply made the thing bigger

00:54:09 ◼   ► and the density is similar as the five nanometer process,

00:54:11 ◼   ► that's not impressive.

00:54:12 ◼   ► And again, we have to see the power draw figures.

00:54:14 ◼   ► I endorse adding more power.

00:54:17 ◼   ► I'm not entirely sure how much of that more power we got,

00:54:21 ◼   ► not for free, but we got as a benefit

00:54:24 ◼   ► of the new three nanometer process.

00:54:26 ◼   ► - I think time will tell and benchmarks will tell

00:54:28 ◼   ► and real life use will tell.

00:54:29 ◼   ► But you can, they did give you some idea

00:54:32 ◼   ► in the presentation with their mostly label-free graphs.

00:54:35 ◼   ► It appears that the M3 line of chips

00:54:39 ◼   ► will push higher at peak power than the previous ones.

00:54:43 ◼   ► It's a similar move that the M2 did compared to the M1.

00:54:45 ◼   ► It is able to peak higher, but they also claim,

00:54:49 ◼   ► it does seem like it's going to be more efficient

00:54:51 ◼   ► in the same way that the M2 is technically more efficient

00:54:53 ◼   ► than the M1, even though it peaks higher.

00:54:55 ◼   ► But also, they claim longer battery life

00:54:58 ◼   ► on the 16-inch now.

00:55:00 ◼   ► So the 16-inch max versus max,

00:55:01 ◼   ► they're basically claiming one additional hour

00:55:03 ◼   ► of battery life compared to the previous ones.

00:55:05 ◼   ► - Yeah, I would think that the battery life claims,

00:55:07 ◼   ► we'll get to them when we get to the machines,

00:55:08 ◼   ► but I would think a lot of them have to do with the fact

00:55:11 ◼   ► that pretty much all of Apple's battery tests

00:55:14 ◼   ► are not like, let's max everything out.

00:55:16 ◼   ► - Oh, not even close.

00:55:17 ◼   ► - Because the efficiency cores are more efficient and better

00:55:21 ◼   ► and because overall efficiency is better,

00:55:23 ◼   ► I would imagine that yeah, the battery life is gonna go up

00:55:25 ◼   ► because you're mostly hanging out on the efficiency cores

00:55:28 ◼   ► and they're better, but we never see the,

00:55:31 ◼   ► okay, what's the everything maxed out battery life?

00:55:34 ◼   ► And I would imagine as we go from M1 to M2 to M3,

00:55:36 ◼   ► the everything maxed out battery life is going to decrease,

00:55:39 ◼   ► even though the work you accomplish

00:55:41 ◼   ► while burning your entire battery should be increasing.

00:55:44 ◼   ► - That seems likely, I think that's right.

00:55:47 ◼   ► But when you compare, from my position as an M1 owner--

00:55:51 ◼   ► - Who keeps his laptop plugged in.

00:55:53 ◼   ► - Yes, well, I do bring it when I travel somewhere.

00:55:56 ◼   ► I've always thought the M1 Pro and Max

00:55:59 ◼   ► didn't have enough E-cores, the efficiency cores.

00:56:01 ◼   ► We only have two E-cores and then we have eight

00:56:04 ◼   ► Pro Core, performance cores.

00:56:06 ◼   ► And when you look at modern Mac OS,

00:56:09 ◼   ► certain background type of tasks,

00:56:11 ◼   ► things like indexing and stuff,

00:56:13 ◼   ► only ever take place on the efficiency cores.

00:56:16 ◼   ► That's the way the OS schedules them.

00:56:17 ◼   ► And so when I look and I see my iStat menu's menu graph

00:56:20 ◼   ► in my menu bar here, those first two cores

00:56:23 ◼   ► are almost always well above 50% use.

00:56:26 ◼   ► Sometimes they're peaked at 100% both for a while,

00:56:29 ◼   ► if it's like indexing something or whatever.

00:56:31 ◼   ► But the efficiency cores get well used on modern Mac OS.

00:56:36 ◼   ► And so to go from two in the M1 line

00:56:38 ◼   ► to four in the M2 line, that's a nice jump.

00:56:41 ◼   ► And that's the kind of thing,

00:56:44 ◼   ► when you're looking at overall battery life usage,

00:56:46 ◼   ► my understanding, what I've heard from people

00:56:48 ◼   ► who know more than I do about this,

00:56:49 ◼   ► is that maxing out the efficiency cores

00:56:52 ◼   ► is basically free in terms of power budget

00:56:55 ◼   ► and battery life.

00:56:56 ◼   ► You really don't notice them.

00:56:58 ◼   ► They're that efficient, they're that low powered

00:57:00 ◼   ► that you can basically max them out

00:57:01 ◼   ► and not really worry about it.

00:57:03 ◼   ► And so all of this adds up to this overall

00:57:05 ◼   ► battery life picture of these chips,

00:57:07 ◼   ► where yes, you're right, that if you actually do

00:57:11 ◼   ► push them hard, of course they're not gonna get

00:57:13 ◼   ► 22 hours of battery life.

00:57:15 ◼   ► That's your movie watching kind of battery life.

00:57:18 ◼   ► But as time goes on, more and more tasks

00:57:22 ◼   ► that you actually do are running on those efficiency cores.

00:57:26 ◼   ► Or a larger portion of what is happening

00:57:28 ◼   ► is running on the efficiency cores.

00:57:29 ◼   ► And as the performance cores get better,

00:57:32 ◼   ► they do finish their work in less time.

00:57:34 ◼   ► So I think there is a lot of merit to the fact

00:57:37 ◼   ► that yes, these actually are getting probably

00:57:39 ◼   ► slightly better or the same battery life

00:57:42 ◼   ► as the predecessors, even though the chip

00:57:44 ◼   ► might be running bigger or hotter or whatever.

00:57:46 ◼   ► - Related to the M3 Max, obviously there's no

00:57:49 ◼   ► ultra announcement today, because the Max Studio

00:57:51 ◼   ► was just updated, and it's probably not ready.

00:57:54 ◼   ► Looking at the Max, it's tempting.

00:57:56 ◼   ► I'm excited about the Max, 'cause it's a great chip

00:57:59 ◼   ► for the laptops, it's more powerful.

00:58:00 ◼   ► Love to see the separation from the middle tier,

00:58:02 ◼   ► love to see going harder on the high end.

00:58:05 ◼   ► - And you love laptops.

00:58:06 ◼   ► - Yeah, if they do the Ultra as two M3 Maxes,

00:58:10 ◼   ► that means the Ultra will be that much more powerful.

00:58:12 ◼   ► Again, 256 gigs maximum RAM, all this GPU,

00:58:15 ◼   ► so many cores, looks great.

00:58:17 ◼   ► I gotta find a link to this in the show notes,

00:58:19 ◼   ► but I watched a video recently of Johnny Ceruggi

00:58:22 ◼   ► giving a talk at his old university.

00:58:24 ◼   ► I'm not sure how recent it was, but anyway.

00:58:26 ◼   ► Obviously he's not gonna say anything about anything.

00:58:30 ◼   ► But if you watch these Apple executives

00:58:32 ◼   ► during presentations, they're very careful.

00:58:35 ◼   ► But usually you can glean something from them.

00:58:38 ◼   ► And of all the questions about what is Apple gonna do

00:58:41 ◼   ► about X or what do you have planned for Y or whatever,

00:58:43 ◼   ► he was willing to say one thing, kind of in the way,

00:58:47 ◼   ► the same reason we all knew that Apple's gonna make a phone.

00:58:49 ◼   ► Once Steve Jobs started saying the phone

00:58:52 ◼   ► is very interesting or whatever,

00:58:54 ◼   ► it's kind of like, I don't know if y'all are using

00:58:55 ◼   ► Tim Cook phrase, but Tim Cook would talk about AR/VR

00:58:57 ◼   ► in the same way, sort of an open secret.

00:58:59 ◼   ► You haven't said Apple is gonna do anything.

00:59:01 ◼   ► Once executives say an area is interesting in some way,

00:59:06 ◼   ► you listen.

00:59:07 ◼   ► So what he said about future silicon stuff or whatever,

00:59:10 ◼   ► 'cause obviously that's his business,

00:59:11 ◼   ► his part of the business,

00:59:13 ◼   ► was that he thought that packaging,

00:59:16 ◼   ► I'm not using the right phrase,

00:59:17 ◼   ► but packaging was an area of interest.

00:59:18 ◼   ► Packaging for the future might be an interesting area

00:59:21 ◼   ► of innovation or something to that effect.

00:59:22 ◼   ► And by packaging, I mean you take the silicon chips

00:59:25 ◼   ► that you get from the wafer and you stick them

00:59:28 ◼   ► inside a container with all of the other stuff

00:59:30 ◼   ► that they need and in the Apple's case,

00:59:33 ◼   ► you put the RAM chips in there and everything.

00:59:34 ◼   ► You make a little package out of it.

00:59:35 ◼   ► That's your SOC.

00:59:36 ◼   ► He said that packaging might be interesting.

00:59:42 ◼   ► I'm not sure if the silicon interposer between the Maxes

00:59:45 ◼   ► that makes the Ultra counts as packaging

00:59:48 ◼   ► because I'm not entirely clear on, you know,

00:59:50 ◼   ► does that nomenclature apply to the silicon interposer?

00:59:53 ◼   ► Is that actually part of fabbing?

00:59:54 ◼   ► Like what do you consider that?

00:59:55 ◼   ► But we do know, and we've talked in the past,

00:59:57 ◼   ► that many other chip manufacturers have used

00:59:59 ◼   ► various packaging innovations to put more stuff

01:00:03 ◼   ► in a single package.

01:00:04 ◼   ► The chiplets things, the various,

01:00:06 ◼   ► we've talked about this in the past episode years ago,

01:00:07 ◼   ► like the various words for the different technologies

01:00:10 ◼   ► of arranging a bunch of different chips

01:00:12 ◼   ► inside a single package and how you connect them

01:00:14 ◼   ► to each other and what the cost trade-offs are

01:00:16 ◼   ► and what the efficiency trade-offs and all of that.

01:00:19 ◼   ► And because I'm desperate to hear good news

01:00:20 ◼   ► about Desktop Max, I heard him say that

01:00:23 ◼   ► and saw this monster M3 Max and thought,

01:00:25 ◼   ► maybe in the M3 generation or the M4 generation,

01:00:28 ◼   ► 'cause the timelines of these are so long,

01:00:30 ◼   ► they won't do the end-to-end Maxes as an Ultra

01:00:34 ◼   ► with the interposer.

01:00:35 ◼   ► Maybe they've come up with a new way to package

01:00:39 ◼   ► multiple units of stuff that they've made

01:00:42 ◼   ► inside a larger chip that would allow them to,

01:00:45 ◼   ► here it comes, put four of them in.

01:00:48 ◼   ► (laughing)

01:00:49 ◼   ► I would love that.

01:00:51 ◼   ► I'm basing that on nothing other than the fact

01:00:53 ◼   ► that Johnny Serugio said the word packaging

01:00:55 ◼   ► in a talk that he gave where he otherwise

01:00:57 ◼   ► revealed nothing about Apple's future plans.

01:01:00 ◼   ► But when I look at this M3 Max, at least I think,

01:01:04 ◼   ► okay, if you make an Ultra out of this,

01:01:05 ◼   ► it will be a good Ultra, good thumbs up.

01:01:08 ◼   ► By Ultra, I mean you put the two Maxes together.

01:01:10 ◼   ► But if this really is a generation

01:01:12 ◼   ► where there'll be packaging changes,

01:01:13 ◼   ► and based on the fact that the M3 Pro is its own thing

01:01:17 ◼   ► and not its own thing, but I guess it is more differentiated

01:01:21 ◼   ► from the Max than it used to be.

01:01:23 ◼   ► To Marco's point, it's less differentiated

01:01:24 ◼   ► from the plain M3 than it used to be, maybe,

01:01:27 ◼   ► but it's more differentiated from the Max.

01:01:28 ◼   ► So maybe Apple is putting some separation

01:01:32 ◼   ► between the middle and the upper upper high end.

01:01:36 ◼   ► So stay tuned to see if there are any packaging innovations

01:01:39 ◼   ► that happen when we finally go beyond M3 Max,

01:01:42 ◼   ► or if this is just another Ultra interposer year.

01:01:45 ◼   ► - Did you wanna talk memory stuff now,

01:01:48 ◼   ► or do you wanna save that for later?

01:01:49 ◼   ► - Yeah, let's talk about RAM, because it is actually,

01:01:51 ◼   ► people have questions about the RAM of the machines,

01:01:53 ◼   ► and we'll talk about the machines in a moment.

01:01:55 ◼   ► But in the end, the SOC, the M3, M3 Pro, M3 Max,

01:02:00 ◼   ► dictates a lot about the memory configurations

01:02:03 ◼   ► of the computers Apple sells.

01:02:04 ◼   ► So here, this is from Cameron McKay,

01:02:07 ◼   ► another inserted follow-up in an event episode.

01:02:09 ◼   ► He says-- - Unprecedented.

01:02:10 ◼   ► - Yes, "I'm not a chip designer,

01:02:12 ◼   ► "but I work in a related field.

01:02:13 ◼   ► "Here's how I make sense of the strange RAM configurations

01:02:16 ◼   ► "on the M3 series.

01:02:17 ◼   ► "This is based on looking at the memory controller blocks

01:02:19 ◼   ► "around the perimeter of the M3 series die shots

01:02:21 ◼   ► "and Apple's tech specs."

01:02:23 ◼   ► And Apple does put, like in their presentation,

01:02:26 ◼   ► they put what we assume are real die shots,

01:02:29 ◼   ► where you can actually see the silicon chip,

01:02:32 ◼   ► photoshopped a little bit, it's beautified,

01:02:34 ◼   ► as it would be in the wafer,

01:02:36 ◼   ► where you can see the different functional blocks

01:02:38 ◼   ► by how the little transistors are arranged.

01:02:41 ◼   ► Cameron continues, "The M3, two RAM chips

01:02:45 ◼   ► "with 100 gigabytes per second of memory bandwidth,

01:02:47 ◼   ► "and you can divide up those two RAM chips this way.

01:02:49 ◼   ► "If you've got two four-gigabyte chips,

01:02:51 ◼   ► "that's an eight gigabytes of RAM.

01:02:52 ◼   ► "If you've got two eights, that's 16 gigs,

01:02:54 ◼   ► "and two twelves, that's 24."

01:02:56 ◼   ► So that explains the memory configurations

01:02:58 ◼   ► of the plain old M3.

01:02:59 ◼   ► You can get eight, 16, and 24.

01:03:01 ◼   ► We'll talk about the eight later

01:03:02 ◼   ► when we talk about machines.

01:03:03 ◼   ► But anyway, there's only two of them,

01:03:04 ◼   ► and those are the sizes they come in,

01:03:07 ◼   ► and that's what you get.

01:03:08 ◼   ► M3 Pro, three RAM chips,

01:03:11 ◼   ► 150 gigabytes per second memory bandwidth,

01:03:13 ◼   ► because there's three memory controllers,

01:03:16 ◼   ► 50 gigabytes per second each.

01:03:18 ◼   ► If they're six-gigabyte chips,

01:03:21 ◼   ► that three times six is a total of 18,

01:03:22 ◼   ► and if they're 12-gigabyte chips,

01:03:24 ◼   ► three times 12 is 36.

01:03:25 ◼   ► That's where you get 18 and 36 from.

01:03:27 ◼   ► That's where we got our little bonus.

01:03:28 ◼   ► Instead of 16 and 32, we get 18 and 36.

01:03:30 ◼   ► It's because it's three times six, so three times 12.

01:03:33 ◼   ► The M3 Max with the 14-core CPU has three RAM chips,

01:03:38 ◼   ► 300 gigabytes per second of memory bandwidth,

01:03:40 ◼   ► and it's either three times 12 for 36

01:03:41 ◼   ► or three times 32 for 96.

01:03:44 ◼   ► And then finally, the M3 Max with the 16-core CPU

01:03:48 ◼   ► has four RAM chips, one more than the 14-core one does.

01:03:52 ◼   ► So it's 400 gigabytes per second memory bandwidth,

01:03:54 ◼   ► and you get four times 12, which is 48,

01:03:56 ◼   ► four times 16, which is 64,

01:03:58 ◼   ► and four times 32, which is 128.

01:04:00 ◼   ► Camera continues, "The M3 Max is a bit odd.

01:04:03 ◼   ► "The memory controller blocks in the die shot

01:04:05 ◼   ► "are twice as big as on the other M3 series chips.

01:04:07 ◼   ► "I assume that means twice the bandwidth

01:04:09 ◼   ► "to each memory chip.

01:04:10 ◼   ► "Maybe the memory chip packages

01:04:11 ◼   ► "each contain two memory dies?

01:04:13 ◼   ► "Apple says memory bandwidth

01:04:14 ◼   ► "is up to 400 gigabytes per second.

01:04:15 ◼   ► "I'm assuming this means that the top-end M3 Max

01:04:17 ◼   ► "is 400 gigabytes per second,

01:04:19 ◼   ► "and the lower 14-core version is 300 megabytes per second.

01:04:22 ◼   ► "The available RAM configurations

01:04:23 ◼   ► "support this interpretation.

01:04:25 ◼   ► "It makes sense that Apple would disable

01:04:26 ◼   ► "one memory controller in the lower-end M3 Max.

01:04:28 ◼   ► "Look how much die area those memory controllers take up."

01:04:31 ◼   ► So this is another example of Apple

01:04:34 ◼   ► really pushing the top end,

01:04:36 ◼   ► like to the point where the big Max

01:04:38 ◼   ► is better in a substantial way than the not-so-big Max,

01:04:41 ◼   ► because the big Max has four RAM chips,

01:04:44 ◼   ► 400 gigabytes per second memory, four controllers.

01:04:47 ◼   ► But if you just saw the right-hand side

01:04:50 ◼   ► of this little thing, it says,

01:04:51 ◼   ► so the configurations are 36 and 96 for this Max,

01:04:55 ◼   ► but 48, 64, and 128 for this one, what the hell?

01:04:59 ◼   ► This is why.

01:04:59 ◼   ► It's how many RAM chips and what size they come in,

01:05:02 ◼   ► and that ends up bifurcating,

01:05:06 ◼   ► especially the computers that come with the M3 Max,

01:05:08 ◼   ► into these weird RAM tiers that don't make any sense

01:05:11 ◼   ► when you're using a configurator.

01:05:12 ◼   ► In fact, Apple has a little message I feel for them,

01:05:14 ◼   ► 'cause I was making similar messages

01:05:15 ◼   ► on our membership thing,

01:05:16 ◼   ► of like when you click a different CPU,

01:05:19 ◼   ► it changes the RAM for you because it has to,

01:05:22 ◼   ► and then it pops up a little apology message,

01:05:23 ◼   ► and it's like, "Sorry, based on the CPU you picked,

01:05:26 ◼   ► "we've changed you to 48 gigs of RAM,"

01:05:28 ◼   ► which is a number you're probably not used to seeing,

01:05:29 ◼   ► but it makes sense if you look at the chips on the SoC.

01:05:33 ◼   ► - Yeah, and in fact, user F. Hausler,

01:05:36 ◼   ► earlier in the recording, posted a mastodon,

01:05:39 ◼   ► an annotated diagram of the chip dies

01:05:42 ◼   ► to F's best guesses of what these different sections are.

01:05:46 ◼   ► And it looks right to me, or at least plausible,

01:05:48 ◼   ► or at least I'm sure most of it's probably right.

01:05:51 ◼   ► So it's helpful to look at this to see

01:05:53 ◼   ► how these chip designs differ,

01:05:55 ◼   ► what they're using their die space for,

01:05:58 ◼   ► kind of the trade-offs that they make, so it's interesting.

01:06:02 ◼   ► (upbeat music)

01:06:03 ◼   ► We are sponsored this week by Collide.

01:06:05 ◼   ► If you work in security or IT,

01:06:07 ◼   ► and your company has Okta, this message is for you.

01:06:10 ◼   ► Have you noticed that for the past few years,

01:06:12 ◼   ► the majority of data breaches and hacks you read about

01:06:14 ◼   ► have something in common?

01:06:16 ◼   ► It's employees.

01:06:17 ◼   ► Hackers absolutely love exploiting

01:06:19 ◼   ► vulnerable employee devices and credentials,

01:06:22 ◼   ► but it doesn't have to be this way.

01:06:24 ◼   ► Imagine a world where only secure devices

01:06:27 ◼   ► can access your cloud apps.

01:06:29 ◼   ► In this world, phished credentials are useless to hackers,

01:06:32 ◼   ► and you can manage every OS, even Linux,

01:06:35 ◼   ► from a single dashboard.

01:06:37 ◼   ► Best of all, you can get employees

01:06:38 ◼   ► to fix their own device security issues

01:06:40 ◼   ► without creating more work for IT.

01:06:43 ◼   ► The good news is, you don't have to imagine this world.

01:06:46 ◼   ► You can just start using Collide.

01:06:48 ◼   ► Collide is a device trust solution for companies with Okta,

01:06:52 ◼   ► and it ensures that if a device isn't trusted and secure,

01:06:54 ◼   ► it can't log into your cloud apps.

01:06:56 ◼   ► Visit collide.com/ATP to watch a demo and see how it works.

01:07:01 ◼   ► That's Collide, spelled K-O-L-I-D-E, .com/ATP.

01:07:06 ◼   ► Collide.com/ATP.

01:07:09 ◼   ► See a demo today.

01:07:11 ◼   ► Thank you so much to Collide for sponsoring our show.

01:07:14 ◼   ► - We have to talk about the MacBook Pro.

01:07:20 ◼   ► It appears to be the same with some asterisks

01:07:25 ◼   ► and daggers and double daggers.

01:07:26 ◼   ► So there is no more 13-inch touch bar MacBook Pro.

01:07:31 ◼   ► Ding dong, the witch is dead.

01:07:34 ◼   ► - Thank goodness.

01:07:36 ◼   ► We've been talking for a while.

01:07:40 ◼   ► One of the most scary things we've ever heard from Apple

01:07:43 ◼   ► wasn't in their scary fast event.

01:07:44 ◼   ► It was, I believe, last WWDC when they said that

01:07:49 ◼   ► that was their second best-selling laptop.

01:07:51 ◼   ► Remember that?

01:07:53 ◼   ► And everyone was like, oh God, really?

01:07:57 ◼   ► That's, and it makes sense.

01:07:59 ◼   ► As we talked about before, there's a lot of individuals

01:08:02 ◼   ► who go into an Apple store saying,

01:08:03 ◼   ► I want something called a MacBook Pro,

01:08:05 ◼   ► and that's the cheapest one by a lot,

01:08:06 ◼   ► and so they pick that.

01:08:07 ◼   ► Also a lot of corporate and volume buyers,

01:08:09 ◼   ► they say, oh yeah, we'll give our employees a MacBook Pro,

01:08:11 ◼   ► and then they buy them that one,

01:08:13 ◼   ► 'cause again, it's the cheapest laptop

01:08:16 ◼   ► that's named MacBook Pro in the lineup.

01:08:18 ◼   ► And so a lot of sales of that were basically people buying it

01:08:22 ◼   ► because it was the cheapest thing called Pro,

01:08:24 ◼   ► not because it was necessarily the one they liked the most.

01:08:27 ◼   ► And so it seemed kind of a tragedy to us nerds

01:08:30 ◼   ► that all of the modern conveniences

01:08:33 ◼   ► of the actual new MacBook Pros and even the MacBook Airs

01:08:38 ◼   ► were not applying to this product,

01:08:41 ◼   ► and yet they kept selling so many.

01:08:43 ◼   ► So it seemed like one of those kind of tragedies

01:08:47 ◼   ► in the lineup, similar to the old non-retina iMac

01:08:50 ◼   ► with the spinning hard drive.

01:08:51 ◼   ► They're like, man, when are they gonna get rid of this thing?

01:08:53 ◼   ► They got rid of this thing.

01:08:55 ◼   ► - Yeah, and there's an explanation for this,

01:08:57 ◼   ► which is not very satisfying, but I've come to believe,

01:09:01 ◼   ► we joke about it all the time,

01:09:01 ◼   ► and I've come to believe it more and more.

01:09:02 ◼   ► In fact, I think I said in Slack earlier today,

01:09:05 ◼   ► this is the real Tim Cook doctrine,

01:09:07 ◼   ► or the real Cook doctrine, which is not own and control

01:09:09 ◼   ► the primary technologies behind your products or whatever.

01:09:12 ◼   ► It's actually keep selling old products

01:09:14 ◼   ► that you already manufacture to maintain margins

01:09:16 ◼   ► on your product line, because we've heard

01:09:18 ◼   ► from actual sources inside the company

01:09:20 ◼   ► that it's not just a thing we think on the outside,

01:09:23 ◼   ► but that decisions are actually made inside Apple

01:09:26 ◼   ► based on the overall margins of the product line.

01:09:29 ◼   ► So why has this laptop been around

01:09:31 ◼   ► for so long with the touch bar or whatever?

01:09:32 ◼   ► Why didn't it get the M2?

01:09:34 ◼   ► Why do we have to wait until the M3?

01:09:36 ◼   ► It's not because the people making these computers

01:09:38 ◼   ► don't know how to make good computers,

01:09:40 ◼   ► or are trying to stick it to you or whatever.

01:09:42 ◼   ► The straightforward thing to do would be,

01:09:45 ◼   ► okay, well the touch bars are going away,

01:09:46 ◼   ► so let's upgrade everybody to non-touch bar.

01:09:49 ◼   ► I bet the people who control these product lines,

01:09:54 ◼   ► that's their first plan.

01:09:55 ◼   ► Like, yeah, all right, well, we've decided

01:09:57 ◼   ► we're leaving behind touch bar, we got this new design,

01:09:58 ◼   ► let's just do it.

01:10:00 ◼   ► And apparently, inside the company,

01:10:02 ◼   ► there is actual pushback from people

01:10:04 ◼   ► who care about the numbers saying,

01:10:07 ◼   ► if you did that, the overall margins

01:10:09 ◼   ► on the Apple laptop product line,

01:10:11 ◼   ► or the MacBook Pro product line,

01:10:13 ◼   ► would be unacceptably low.

01:10:15 ◼   ► So you have to keep selling the touch bar one

01:10:17 ◼   ► for another year to maintain our margins.

01:10:22 ◼   ► We don't wanna hear that on the outside,

01:10:23 ◼   ► but that's just the fact of running a company sometimes.

01:10:26 ◼   ► You know, again, maybe that's not the way

01:10:28 ◼   ► we think it should be run, but we see from the outside

01:10:31 ◼   ► that that is how it happens, and apparently,

01:10:34 ◼   ► it's not just an accident, or like,

01:10:36 ◼   ► they're not thinking about it, or like,

01:10:38 ◼   ► whoopsie, they forgot to update that one.

01:10:40 ◼   ► It's someone inside Apple saying,

01:10:43 ◼   ► you gotta keep selling the crappy one for another year.

01:10:45 ◼   ► And this is the type of thing where it's like,

01:10:47 ◼   ► I know Apple's big thing is we make long-term decisions,

01:10:50 ◼   ► not short-term ones, but I do feel like

01:10:53 ◼   ► that type of decision, it doesn't,

01:10:56 ◼   ► it hurts Apple's reputation as a company

01:10:59 ◼   ► a tiny, tiny, tiny bit.

01:11:02 ◼   ► Because if you buy one of these not knowing

01:11:06 ◼   ► the weird limitations and the fact that the features

01:11:09 ◼   ► have been updated on the other laptops

01:11:11 ◼   ► and you might be happier with them,

01:11:13 ◼   ► maybe you're slightly less satisfied,

01:11:15 ◼   ► or maybe even if you just listen to a tech podcast

01:11:17 ◼   ► and everyone is saying that's the bad one

01:11:18 ◼   ► and you bought the bad one, you feel bad about it

01:11:20 ◼   ► or whatever, that type of situation,

01:11:22 ◼   ► it's not a big deal, it was still an okay computer,

01:11:24 ◼   ► it's fine, right, but the fact that they sold so many of them

01:11:27 ◼   ► they're not selling their best work.

01:11:29 ◼   ► They know they can do better.

01:11:31 ◼   ► They have done better, but they're still selling

01:11:33 ◼   ► this lesser computer that is less satisfying to customers

01:11:36 ◼   ► and less useful overall.

01:11:37 ◼   ► To maintain the margins on the product line

01:11:40 ◼   ► of the MacBook Pro, that is not long-term

01:11:44 ◼   ► brand-preserving thinking in my opinion.

01:11:46 ◼   ► Maybe it is the right decision if you care

01:11:50 ◼   ► about the long-term financial health of the company,

01:11:52 ◼   ► but as a tech nerd, I care less about that to the degree,

01:11:57 ◼   ► as long as they're not going out of business

01:11:58 ◼   ► and they're doing well, I don't care about

01:12:00 ◼   ► the tiny percentages, I don't have a bunch of Apple RSUs

01:12:03 ◼   ► that are investing or something, I'm not investing.

01:12:04 ◼   ► And honestly, I don't think Tim Cook cares.

01:12:07 ◼   ► He's a bazillionaire, he doesn't need any more money,

01:12:08 ◼   ► but I feel like Tim Cook, the real Tim Cook doctrine is

01:12:12 ◼   ► don't let margins fall, do what's right

01:12:15 ◼   ► for the company financially.

01:12:16 ◼   ► Not because he's greedy and wants money, he doesn't care,

01:12:19 ◼   ► but I think as a CEO of a company,

01:12:21 ◼   ► as a person whose expertise was in operations or whatever,

01:12:25 ◼   ► I think he sees that as his responsibility

01:12:28 ◼   ► to make sure the company does well financially

01:12:31 ◼   ► in addition to all the other things.

01:12:32 ◼   ► And he's really good at that.

01:12:35 ◼   ► And so he does things to pursue that goal

01:12:38 ◼   ► that I personally think are the wrong decisions

01:12:40 ◼   ► for the Apple brand in the long-term,

01:12:43 ◼   ► even if they're the right decisions

01:12:44 ◼   ► for the Apple stock price in the long-term.

01:12:47 ◼   ► - Yeah, I don't think it's as ruinous to the brand

01:12:50 ◼   ► as you seem to think so.

01:12:51 ◼   ► - I said a tiny, tiny, tiny, tiny bit,

01:12:55 ◼   ► but non-zero, measurable, real.

01:12:57 ◼   ► - Yeah, it's not ruinous so much as corrosive.

01:12:59 ◼   ► - Yeah, you know, that's a better word for it.

01:13:01 ◼   ► - And not even corrosive, it's just like,

01:13:02 ◼   ► it's a less optimal decision.

01:13:04 ◼   ► If your decision was I want our customers to be the happiest,

01:13:06 ◼   ► you wouldn't keep selling that computer for a year, right?

01:13:09 ◼   ► I don't think it's a big deal, but it is a thing.

01:13:12 ◼   ► It is a trade-off.

01:13:13 ◼   ► - Well, if they wanted their customers to be happiest,

01:13:14 ◼   ► they wouldn't have anything that starts

01:13:16 ◼   ► with 256 gigs of disk space either.

01:13:18 ◼   ► - Oh yeah, we'll get there.

01:13:19 ◼   ► (laughing)

01:13:20 ◼   ► - All right, so moving right along.

01:13:21 ◼   ► So no more touch bar, the touch bar is dead.

01:13:24 ◼   ► - Oh, thank God.

01:13:25 ◼   ► - But what was the other, the dagger, if you will,

01:13:29 ◼   ► on that laptop?

01:13:30 ◼   ► It only has two non-Thunderbolt ports.

01:13:33 ◼   ► - Yes.

01:13:34 ◼   ► - Non-Thunderbolt for USB-C ports, right?

01:13:36 ◼   ► - Yes, so this new kind of lower end 14 inch

01:13:40 ◼   ► is basically a replacement for that touch bar Mac

01:13:44 ◼   ► price-wise, sort of.

01:13:45 ◼   ► It kind of replaces the high end of the price.

01:13:48 ◼   ► Starts at 1600 bucks, the outgoing touch bar one,

01:13:52 ◼   ► I think was 1,314 or 1,500 as two configs in that ballpark.

01:13:56 ◼   ► So this is a little more money.

01:13:58 ◼   ► But first of all, I think the need for that

01:14:01 ◼   ► in the price slot is greatly reduced by the existence

01:14:05 ◼   ► of the 15 inch MacBook Air.

01:14:06 ◼   ► And frankly, the 13 inch MacBook Air.

01:14:09 ◼   ► The MacBook Air is, I think, will solve almost

01:14:12 ◼   ► everyone's needs who have this kind of need,

01:14:14 ◼   ► with very few exceptions.

01:14:15 ◼   ► But anyway, what's interesting about this new low end

01:14:18 ◼   ► 14 inch Pro, yes, it does only have two USB-C

01:14:22 ◼   ► Thunderbolt ports, only the two left ones.

01:14:25 ◼   ► There's no more one on the right for that model.

01:14:27 ◼   ► And that's because the M3 does not support three

01:14:31 ◼   ► USB-C Thunderbolt ports, it only supports two.

01:14:33 ◼   ► So there is a reason for that in the silicon.

01:14:35 ◼   ► Again, die space, market segmentation, these are trade-offs.

01:14:38 ◼   ► That's why they did this.

01:14:40 ◼   ► They didn't just delete the port to save 10 bucks

01:14:42 ◼   ► on the cost of the port or whatever.

01:14:44 ◼   ► They deleted the port because the entire silicon

01:14:47 ◼   ► chip does not support that many of them.

01:14:49 ◼   ► That's the same reason why MacBook Airs only have two ports,

01:14:51 ◼   ► 'cause they're using the same chip trade-offs.

01:14:53 ◼   ► And I think that's fine.

01:14:56 ◼   ► Again, this is a low end product.

01:14:57 ◼   ► It's kind of a mid range product at this point,

01:15:01 ◼   ► because what's interesting also,

01:15:03 ◼   ► when they were starting the presentation and they said,

01:15:06 ◼   ► we have all three of these chips, and they said

01:15:08 ◼   ► they're all gonna go into MacBook Pro,

01:15:09 ◼   ► I thought for a second, oh crap,

01:15:11 ◼   ► they updated the Touch Bar one.

01:15:13 ◼   ► 'Cause where else would the regular--

01:15:14 ◼   ► - You thought they were gonna put an M3

01:15:15 ◼   ► in the Touch Bar one, that would have been terrible.

01:15:17 ◼   ► - 'Cause where else would they put the non-pro and max chip?

01:15:21 ◼   ► I figured, well, it's not gonna go in the 14 and 16,

01:15:23 ◼   ► so it must go, anyway, so over time,

01:15:26 ◼   ► as they were introducing this low end 14,

01:15:28 ◼   ► then I was thinking, wow, okay, so they added

01:15:31 ◼   ► a low end config to the 14 to fill that void.

01:15:33 ◼   ► That's great news, but surely they're gonna do things

01:15:36 ◼   ► like take away some of the nice pro screen stuff.

01:15:40 ◼   ► Maybe it's not HDR or it doesn't have 120 hertz or whatever.

01:15:44 ◼   ► And they didn't do that.

01:15:45 ◼   ► Then I thought, oh, maybe they'll take away

01:15:47 ◼   ► like some of the speakers or the microphones or whatever,

01:15:50 ◼   ► make the webcam worse.

01:15:51 ◼   ► They didn't do any of those things.