throwaway31131

Which lawsuit was that?

I think it’s been a while since there was a MBP MLB lawsuit but maybe I’m forgetting about one.

This is the last one I recall.

“Apple faces class-action lawsuit over 2011 MacBook Pro GPU issues“

https://9to5mac.com/2014/10/28/apple-class-action-lawsuit-20...

TSMC in Taiwan has a significant share of the wafers produced by the world every month. If those wafers were not produced the global economy would suffer badly.

It takes years to bring a fab online. Fab 21 in Arizona took 5 years to enter mass production from ground breaking. Some believe it could be done in two but that’s yet to be demonstrated. Then there are the wafers themselves. The total time it takes to process one wafer at the single nm scale is around 100 days.

So realistically, even if one makes up their mind to make a fab fast, you’re looking at 3 years before you have your first sellable wafer.

For the cores working hardest to achieve the absolute lowest cpi running user code, this is true. But these days the computers have computers in them to manage the system. And these kinds of statements aren’t necessarily true for these “inner cores” that aren’t user accessible.

“ RTKit: Apple's proprietary real-time operating system. Most of the accelerators (AGX, ANE, AOP, DCP, AVE, PMP) run RTKit on an internal processor. The string "RTKSTACKRTKSTACK" is characteristic of a firmware containing RTKit.”

https://asahilinux.org/docs/project/glossary/#r

How old is the kid?

When my kids were less than 10 I took them to the Computer History Museum in Mountain View CA and showed them how all the mechanical computers work. They have some exploded and simplified exhibits that are very visual so kids can get it pretty easy. Especially if they’re the kind of kid that builds with blocks. Games like Turing Tumble are good too. Then you just wave your hands and say chips work like this on the inside, but with electricity instead of gears because it’s faster, smaller, cheaper and uses less energy. But at the end of the day, it’s basically all clockwork. Then they smile and say, “makes sense”. And I leave it at that.

When the kids were teenagers I gave them a more thorough explanation.

If you’re not in the Bay Area then the Computer Museum does have YouTube videos. There are many others as well.

https://m.youtube.com/watch?v=XSkGY6LchJs&pp=ygUbbWVjaGFuaWN...

Validation that a PCB was manufactured correctly is... easy. Disagree, but how about VLSI. It's hugely automated. Moore's Law is exponential but team sizes aren't. That productivity gap is made up for with huge amounts of automation. And nothing is easy about manufacturing validation of an ASIC.

I do think one primary difference between physical objects and software is we bother to have precise specifications that one can validate against, and I think that's what you're trying to get at. If all software had that then software could have an "easy" validation story too, I suppose.

I have mixed feelings about precise specifications in software. On the one hand the hardware engineer in me thinks everything should have an exact specification. On the other hand, that's throws away the "soft" advantage which is important for some types of software. So there is a spectrum.

Data maybe somewhat dated and I haven’t measured it myself but,

“Per his findings, Chrome used 290MB of RAM per open tab, while Safari only used 12MB of RAM per open tab.”

https://www.macrumors.com/2021/02/20/chrome-safari-ram-test/

Doesn’t that make sense though as each manipulates a different layer in the memory hierarchy allowing the programmer to control the latency and throughput implications. I see it as a good thing.

Maybe not all Bavaria, but the capital has been trying since 2004.

https://interoperable-europe.ec.europa.eu/collection/open-so...

But all that information is in the engineering files that are neither the schematic or the layout. You might guess some information from the layout, but on some crazy HDI PCB like the iPhone uses, you'll be missing most of the information needed to take advantage of it.

Any signal that uses passive termination, and would therefore show up in the schematic, is basically a tacit admission that this line needs something, but basically anything remotely close to 50Ω will do. Which is something you could've gathered by inspection of the final product.

You don't need a schematic to do that.

That's assuming they can get their hands on the "black box" ASICs

The Lantertronics YouTube channel has a bunch of good stuff. https://www.youtube.com/@Lantertronics

Also, Small Signal Audio Design by Douglas Self is really good, but probably not as a first EE textbook, and you have to really want to go into the weeds, as an electrical engineer would. But there is an entire chapter (chapter 12) on just electric guitars (pickup, preamps, effects, direct injection, etc.)

> no more engineering for engineering’s sake

I'm not sure many successful engineering orgs did much of that but also the environment our creations live in is much different now.

It was a huge task just to get our programs to run at all not very long ago. When I first started coding I was working on some code for a VLSI problem and I spent like a month just figuring out how to get the problem to fit on disk (not main memory, but on disk!). Now there are similar things that run on a laptop [0]. And it's not like I'm unique in having to come up with crazy solutions to problems tangent to the primary problem in this space. [1]

Now the infrastructure for our code is amazing and the machines that execute them abound in resources, especially in the cloud. Now that the yoke has been cast off it makes sense more time is spent on the solution the actual problem you set out to solve in the first place.

[0] https://github.com/The-OpenROAD-Project/OpenLane [1] How Prince of Persia Defeated Apple II's Memory Limitations https://www.youtube.com/watch?v=sw0VfmXKq54

In my opinion you've actually hit upon the interesting part of AI.

If you can create an application easily with AI, then someone else can also create that application easily with AI. This means the value is no longer in that application merely existing.

To me, this means that the value, i.e. the things people will pay for, in software will increasingly come from branding on the low end and from the applications that AI can't yet create on the high end.

Maybe not a railroad but I imagine the data centers won't be worth nothing...

I imagine the AI bubble bust will be like the .com bubble bust. Of the 20 (or whatever number) of companies that have a shot, something like 3 will survive and do well. The problem is we don't know which 3.

Many, many, dot-com era companies died during the .com bubble and tons of money was lost, but not everything. For example: Amazon.com, eBay.com, Intuit, etc.

Also, the big challenges in software organizations I deal with are getting a group of a dozen or so “code-slingers” to work effectively as a team. That’s really hard and it’s been the fundamental management problem for the entire time I’ve been in the industry, which is decades. Not enough LOC has never really been a thing.

If you have more bugs than you can fix in a given amount of time then you have to prioritize somehow.

“This bug didn't block release X-1, why should it block release X?” Is actually a pretty strong argument and tough, but not impossible, to counter.

And the bug backlog only gets longer with time. It’s the price of greatly increased software complexity.

I’d say it’s not even problem solving and it’s more pattern recognition.

I actually love LC and have been doing a problem a week for years. Basically I give myself 30 minutes and see what I can do. It’s my equivalent to the Sunday crossword. After awhile the signals and patterns became obvious, to me anyway.

I also love puzzlerush at chess.com. In chess puzzles there are patterns and themes. I can easily solve a 1600 rated problem in under 3 seconds for a chess position I’ve never seen before not because I solve the position by searching some move tree in my mind, I just recognize and apply the pattern. (It also makes it easier to trick the player when rushing but even the tricks have patterns :)

That said, in our group we will definitely have one person ask the candidate a LC style question. It will probably be me asking and I usually just make it up on the spot based on the resume. I think it’s more fun when neither one of us know the answer. Algorithm development, especially on graphs, is a critical part of the job so it’s important to demonstrate competency there.

Software engineering is a hugely diverse field now. Saying you’re a programmer is kinda like saying you’re an artist. It does give some information but you still don’t really know what skill set that person uses day to day.

That’s a good point I hadn’t considered.

I get it. But on the other hand, if you in a situation that requires a six month food supply, because there is no resupply available and you can’t relocate, you have bigger problems. Like the mob of hungry people outside your house.

Los Angelinos would disagree https://en.m.wikipedia.org/wiki/1992_Los_Angeles_riots

That’s certainly true but at the same time, when I was a kid in the early 90s, we watched TV but cartoons ended (we did not have cable or a computer). I came home from school, ate a snack, watched TV for about an hour with a friend, cartoons were over and we went outside. With the internet and YouTube etc. you’re never “done”

Cost is relative. LIDAR maybe be expensive relative to a camera or two but it’s very inexpensive compared to hiring a full time driver. Crashes aren’t particularly cheap either. Neither are insurance premiums.

Percent utilization for most operating systems is the amount of time the idle task is not scheduled. So for both workloads the idle task was never scheduled, hence 100% "utilization".

Junior engineers are getting hired but there are definitely fewer positions available. It’s scary for sure but it’s also a normal part of the boom & bust cycle that is inherent to the tech industry.

If anything, AI investment is propping us up as some money is still getting invested even though money is expensive at the moment.

In 2002 friends of mine that graduated top 10 at UC Berkeley were struggling to get interviews, never mind jobs. That was the worst dry spell I’ve seen in my career. But they stayed busy and were under employed for a bit and eventually got thier first jobs. One even got picked up by a growing startup named Google.

I think most industries are like this. In hardware engineering we definitely get clobbered roughly once a decade for one reason or another.

When the majority of people want what you want, democracy is great.

When the majority of people don’t want what you want, democracy sucks.

The main issue is we’re not there today and it’s not obvious what that world looks like.

We all had junk drawers of useless charging cables, everyone agreed it was stupid, hence a universal charging connector standard along with the promise that the charger junk drawers will be freed.

Even if we mandate the “POSIX of smart phones”, for lack of a better term, what problem today, for everyday users, does it solve? It might even make interactions with various government technology worse as that API will likely only be begrudgingly supported, which won’t win any hearts or minds.

Basically until you have a one line slogan that most people can relate to which, and is a problem they have today, movement will be very slow.

Also, in the short term, if these various site are AI coded, and thus follow existing software patterns, expect this to get worse.

> To push further, Google and Apple have basically as much power as the US.

If this is true then why is Tim Cook visiting Trump? Shouldn’t it be the other way around.

> The problem isn't that locked-down devices exist - it's that we don't have enough truly open alternatives for those who want them.

Not for lack of trying. See for yourself

https://en.m.wikipedia.org/wiki/List_of_open-source_mobile_p...

The list is not short.

Plenty of companies have attempted this over the years but it’s not obvious that a big enough customer base exists to support the tremendous number of engineering hours it takes to make a phone. Making a decent smart phone is really hard. And the operations needed to support production isn’t cheap either.

> I'm not sure it's that our job is the most automatable

I don't know. It seems pretty friendly to automation to me.

When was the last time you wrote assembly? When was the last time you had map memory? Think about blitting memory to a screen buffer to draw a square on a screen? Schedule processes and threads?

These are things that I routinely did as a junior engineer writing software a long time ago. Most people at that time did. For the most part, the computer does them all now. People still do them, but only when it really counts and applications are niche.

Think about how large code bases are now and how complicated software systems are. How many layers they have. Complexity on this scale was unthinkable not so long ago.

It's all possible because the computer manages much of the complexity through various forms of automation.

Expect more automation. Maybe LLMs are the vehicle that delivers it, maybe not. But more automation in software is the rule, not the exception.