192: Apple's M1 Pro & Max Crushing Intel, Toyota Waking Up on EVs?, Backblaze IPO, Alibaba's Custom ARM Chips, Standard Oil, Safest Power, and Hell Let Loose
"Dreams *do* come true"
Please try to keep an open mind — not only to let new ideas in, but also to let the old ideas out!
—Larry Gonick, Cartoon History of the Universe III
🤔 So Facebook is going to change its name to "Meta" or something like that, and everybody will just keep calling it Facebook, just like with Google and Alphabet, right?
🏎 One thing I never really understood is people trying to impress others with their cars.
Other than signal that you have the money to spend on it (or maybe you don’t and just have lots of debt..?), what credit do you deserve for whatever it is that you drive? You didn’t design it or build it, you just basically picked it out of a catalogue and then argued over floor mats. Impressive.
Oh, and the whole “I’m going to accelerate really fast and make a lot of noise to impress you!”. Again, what credit do you deserve? Oh wow, you pressed down on a pedal with your foot. Very impressive.
I’d be more impressed if these dudes (because it’s dudes) sprinted down the street to prove they’re in great shape 🏃♂️ or hopped on a bike and did this…
Maybe it’s just me ¯\_(ツ)_/¯
🚿 The very first person to figure out how to make vanilla extract from these unassuming flowers must’ve felt pretty proud of themselves. I know I would’ve. What a wonderful smell and flavor.
Although its primary flavor compound is vanillin, pure vanilla extract contains several hundred additional flavor compounds, which are responsible for its complex, deep flavor. By contrast, artificial vanilla flavor is solely made up of artificially-derived vanillin, which is frequently made from a by-product of the wood pulp industry.
💻 Maybe I just don’t go out much, but it’s the first time I see one of these “vending machines” where you can borrow a laptop or a tablet in a public library. Feels straight out of the Robocop universe.
📁 I decided to stop using Dropbox. I’ve been using it since 2008, something like that.
I just want it to sync files to the cloud and across devices, and once in a while share a file. That’s it.
But over the years — probably because of the growth pressures of being a public company and because their core product isn’t the hardest thing in the world to duplicate — the software has bloated into something that does lots of things I don’t need.
So I’m reclaiming hundreds of megs or RAM that were used by the Dropbox client to sync some files, and will used iCloud Drive instead.
It’s built-in MacOS, and after using it for a few years in parallel with DB, I now trust it enough to make it my primary cloud drive.
Funny thing is, I never paid a cent to Dropbox for a decade+ of use.
In the early years, I got some extra space for referring others, so I had 7.5 gigs on the free plan, and that was enough for the kind of files I wanted to keep there. ¯\_(ツ)_/¯
💚 🥃 If you would be bummed out to wake up one morning to a email saying that I can’t do the newsletter anymore because it’s a full-time job and the opportunity cost is too high with only 4.5% of readers supporting it, you know what to do.
Please sign up to become a supporter, we’ll both be happy about it. Thank you 🧡
A Word from our Legendary Sponsor: 🏰 Greenhaven Road Capital 🏰
This is definitely the craziest thing I’ve done in this newsletter, and I love it.
Back in edition #159, I wrote as a joke that I’d love to review Parmigiano-Reggiano (aka real parmesan). I just love the stuff, and was probably hungry at the time, so it seemed like the ultimate gig.
Dreams *do* come true — I was contacted by a reader and supporter (💚 🥃) at Greenhaven Road Capital (check out their investor letters) who wanted to sponsor my parmesan adventure. So after recruiting my wife to the project, here we are.
The format of this gameshow is simple: I’m going to put various cheeses that the nice lady at the cheese-shop recommended to me as parmesan-like against the reigning king, the 👑 Parmigiano-Reggiano 👑, and let you know what I think.
Will we witness regicide? Stay tuned.
⚔️ Round 1: Parmigiano-Reggiano vs Grand Padano Cheese ⚔️
My wife and I tried both, without eating or drinking anything else to muddy things up.
What struck me first was the difference in texture. The Reggiano just has a more interesting texture, with some little crystals that are kind of crunchy, in an otherwise very firm cheese.
The Grand Padano lacked the tiny crystals, and the overall texture was a bit softer.
My wife had a good summary of the difference in taste: "It's a bit like having the original and an inferior no-name alternative. Other than the lower price, I don't see the attraction for the Padano."
I tend to agree. The Padano isn't bad per se, but everything about it is less sharp and less intense than the Reggiano. It probably would seem better if it wasn't directly compared to the king, but hey, this is a deathmatch, and so far, the king remains on the throne after round 1 👑.
Update a week later: I tried the Grand Padano again, but this time on its own. I think it’s one of these things where humans form relative opinions, because it’s delicious. It only suffers in direct comparison to the king, but if you ever offer me a piece, I’ll take it for sure. 🧀
I can’t close without showing you this badass picture of how they stamp Grand Padano (who knew cheese-making was this cool, eh?):
Investing & Business
Toyota Waking Up on EVs (?) 🚘🔋
I always felt like Toyota snatched defeat from the jaws of victory when it came to the long obvious transition to electric vehicles that we’re in the middle of.
They were early to hybrids with the Prius (the first production model came out in Japan in *1997*), and were the first to scale up HEVs with over 18 million sold cumulatively so far across more models than anyone else.
But then 🦗🦗🦗
They had a bunch of Homer concept cars at auto shows, lots of noise about hydrogen, and they even had a short-lived partnership with Tesla to provide some powertrain tech for a low-volume electric Rav 4… but nothing too serious, mostly just compliance cars to meet some local requirements or get some PR.
Hopefully that’s starting to change now:
Toyota announced that it will spend $13.6 billion (¥1.5 trillion) on batteries between now and 2030. Of that money, $9 billion (¥1 trillion) will go toward battery production, with a planned output of 180 to 200 GWh/year by the end of the decade.
Some of that will go to the US:
[Toyota] will spend $3.4 billion (¥ 380 billion) on high-voltage batteries for electric vehicles in the US between now and 2030.
The photo at the top of this is the Toyota bZ4x (I want to know who named it that? that person should not be in charge of naming things…), an EV scheduled for 2022.
Not much is known about it so far, and I’ll be curious to see how good it is, and how fast they can ramp up production of it as well as roll out more models.
Apple further embarrasses Intel…
Nerds like me were quite aware that the M1 chips we had seen so far were all for low-end Macs, and that we could expect something quite a bit beefier for the Pro lines, but even with these expectations, Apple crushed it on Monday.
The M1 Pro and M1 Max SoCs are beasts, and they fall into that fairly rare category of having-cake-and-eating-it-too.
They crush Intel chips on performance *while* using a lot less power.
Today, the real axis of competition is performance/watt, and in that world, the M1 family of chips reigns supreme.
The Max has 57 billion transistors, 400gb/s of memory bandwidth, 32 GPU cores with direct access to 64GB of RAM… That’s just bonkers.
Apple is also focusing a lot on real-world workloads rather than just benchmarks, so there’s plenty of stuff they’ve spent their silicon budget on that may not show up in Geekbench numbers, but that in real-life will make things faster and more energy-efficient, like the new decoding cores for ProRes video, which video professionals will find magical when encoding or doing playback on multiple streams of 8K video or whatever it is they do. Same with the machine-learning accelerator cores.
Apple’s GPU work is also very impressive. There’s a slide where they compared the Max to a mobile Nvidia GeForce RTX 3080 and it’s competitive with it at 100W less power (60W vs. 160W). Just bonkers 🤯
And remember. These are *laptop chips*. I can’t wait to see what they put in the iMac Pro (which has fewer thermal limitations and no battery to worry about) and the big daddy Mac Pro.
I didn’t even mention the redesigned Macbook Pro yet.
Let’s just say that they addressed almost everything people had been complaining about with the old models, and that’s a rare thing to get in one fell swoop. Screen looks great, battery life seems great, keyboard seems great, audio input and output seems much improved, webcam much improved, and speed should be out of this world thanks to the M1 Pro/Max.
Hard to complain…
Alibaba Making its own Custom ARM-based Chips
Alibaba Group Holding will this week launch a server chip based on technology from British firm Arm Holdings [...]
The design process for Alibaba’s central processing unit (CPU), which has been under development since 2019, concluded in the middle of this year. It will be minted using the 5-nanometre process [...]
That basically means TSMC will fab it.
Alibaba, which in 2018 founded a semiconductor design subsidiary PingTouGe, also known as T-Head, has also been using foreign IP to craft its own chips. They include Hanguang 800, a neural processing unit unveiled in September 2019 that is designed to accelerate machine learning and artificial intelligence (AI) tasks. It was fabricated on a 12-nanometre process, packing up to 17 billion transistors. XuanTie 910, another Alibaba-designed chip, can be used for IoT applications in sectors such as 5G, AI and autonomous driving. (Source)
Backblaze files for IPO
The cloud backup service that I personally use has filed an S-1.
If you want my thoughts about the service, and about why you should use cloud backup, check out what I wrote in the intro of edition #22. My thoughts on it haven’t changed, so I don’t really have anything to add…
I was surprised to see that their gross margins are that high, but I guess most users have relatively little data and it's a huge power law.
They’re at the center of a push-pull between people generating more data and the Moore’s Law equivalent for storage density (Kryder's Law?).
I guess that so far, they’ve been able to keep ahead of the curve while keeping their prices pretty low — personal unlimited backup on a 2-year plan is about $5/month.
What I really hope is that they stay focused on being the best backup/cloud storage company they can, and don’t feel like they have to expand in 15 different directions like Dropbox did after it went public and had trouble meeting expectations.
🛢 History of Standard Oil, Part 1 ⛽️
Normally I’d wait for the full series to be out before recommending it here, but this was enough fun that I’ll go ahead with just part 1:
Maybe I’m just biased because friend-of-the-show Ben gives a shoutout Deadwood near the end, though… 🤠
Update: Part 2 is out! I haven’t had a chance to listen yet, but I’m sure it’ll be great.
Science & Technology
Safest & Cleanest Sources of Energy, Trade-offs Edition
Our World In Data compiled some info about various energy sources.
They put the numbers in the context of what it would mean for a mid-sized town:
Looking at deaths per terawatt-hour can seem a bit abstract. So let’s try to put it in perspective.
Let’s consider how many deaths each source would cause for an average town of 187,090 people in Europe, which – as I’ve said before – consume one terawatt-hour of electricity per year. Let’s call this town ‘Euroville’.
If Euroville was completely powered by coal we’d expect 25 people to die prematurely every year as a result. Most of these people would die from air pollution. This is how a coal-powered Euroville would compare with towns powered by other energy sources:
Coal: 25 people would die prematurely every year;
Oil: 18 people would die prematurely every year;
Gas: 3 people would die prematurely every year;
Nuclear: In an average year nobody would die. A death rate of 0.07 deaths per terawatt-hour means it would take 14 years before a single person would die. As we will explore later, this might even be an overestimate.
Wind: In an average year nobody would die – it will take 29 years before someone died;
Hydropower: In an average year nobody would die – it will take 42 years before someone died;
Solar: In an average year nobody would die – only every 53 years would someone die.
People are often misled because they only look at one aspect of things, and not at the whole picture.
The fact is, life is trade-offs, and when we don’t do thing X, we have to do things Y instead. If you don’t do X because there are problems with it, but do Y which is even worse, have we made a good decision?
Contrary to popular belief, nuclear power has saved lives by displacing fossil fuels
The disaster in Fukushima killed 574 people. In 2018, the Japanese government reported that one worker has since died from lung cancer as a result of exposure from the event. No one died directly from the Fukushima disaster. Instead, most people died as a result of evacuation procedures. According to Japanese authorities 573 people died due to the impact of the evacuation and stress [...]
In the wake of the 2011 Fukushima nuclear disaster, Germany announced plans to phase out nuclear power generation: over the period from 2011 to 2017 it shut down 10 of its 17 nuclear facilities, and plans to close the remaining reactors in 2022.
Because nuclear is safer than its main alternatives this policy decision cost lives.
Replacing nuclear energy with fossil fuels kills people. This is likely to be the case in the recent example of Germany. Most of Germany’s energy deficit from scrapping nuclear was filled by increased coal production – which is, as we just saw, the most polluting source with the largest health impacts. Analysis by Stephen Jarvis, Olivier Deschenes, and Akshaya Jha (2020) estimates that Germany’s nuclear phase-out has come at the cost of more than 1,100 additional deaths each year as a result of air pollution. Germany’s plan to make its energy systems safer has done exactly the opposite. (Source)
The Arts & History
Hell Let Loose (Video Game)
I was telling a friend about my days of playing COD4 (the first ‘Modern Warfare’) on hardcore tactical servers that aimed for realism... He told me about this game he’s playing, and so far I’ve only seen videos of it on Youtube, but it’s pretty cool.
Almost like you’re a character in Band of Brothers. It’s fairly realistic and environments are based on real maps from that era.
And if you want more, this guy’s channel if full of videos of this game.
If any would-be sponsors are reading this and would like to partner on future reviews of random cool things, please send me an email! (you can just reply to any newsletter edition or DM me on Twitter) I think my wife would love to do dark chocolates, and I wouldn’t mind reviewing a couple well-rated kitchen knives to figure out which is best…
Apple's success with the M1 has made me wonder whether x86 chips need to exist at all. Apple's first generation ARM chips run x86 apps (through their rosetta translation layer) better than they do natively.
Is it crazy to think that before long Azure, GCP, AWS will have their own ARM chips capable or running x86 workloads through a layer of emulation/translation with better outright performance and price performance than a comparable x86 chip? If this is possible what is the raison d'être of x86?
Maybe I am missing something fundamental. curious to get your thoughts.