Hacker News Reader

Comment by Underphil | Hacker News Reader

I often think about this. We do clearly waste CPU cycles by the boatload these days, but what would the carbon footprint be like if we were still using these machines now? How does the power consumption compare? How would it compare if we were still using the protocols and languages of the past?

(I'm not an 'everything was better in the old days' kinda guy, but I do think about this just as a curiosity).

View on HN

It was not better. Power supplies in the 80s were not very high efficiency. They did not have much in the way of low dropout regulators meaning the boards blew waste heat constantly with few sleep modes invented yet. It wasn't even cpu cycles generating that heat, just resistive waste. All of the peripherals like disks and of course the CRT had quiescent draw as well.

So the 80s were not good. The 60s and 70s -- even worse. Gas had lead. And it spewed everywhere and we'll never get it out.

I remember reading that the Apple 2e was the first device with a switching power supply and the power supply design revolutionized power supply design for years to come. Is that true?

It's not true. Woz's genius switching power supply was what made the Apple ][ special long before the 2e was even dreamed of.

Apparently not, see link provided in the other reply:

http://www.righto.com/2012/02/apple-didnt-revolutionize-powe...

Jobs lied and, additionally, Woz did not design that power supply as you suggest.

In the book "I, Woz", Steve Wozniak explains that Jobs hired a random hacker Rod Holt to design switching mode power supply unit for their Apple II computer. Jobs paid him $200 a day which was a lot of money by that time. Rod did the best work he could.

https://en.wikipedia.org/wiki/Rod_Holt

http://www.righto.com/2012/02/apple-didnt-revolutionize-powe...

This was amazing to read. Thanks for correcting me and string the record the straight for me. Jobs was not telling the truth.

I'm glad you liked the article. If you want something more "official", here's a link to the IEEE Spectrum: https://spectrum.ieee.org/a-half-century-ago-better-transist...

Love your website, Ken. I discovered it some time back from HN (long before I registered an account here). You are an inspiration.

But computers use a lot more power nowadays. The early home computers were all passively cooled. My Atari ST power supply actually broke down on me so I had to buy a new one from Great Britain. It's a quite small affair with quite minimal venting through the grills at the top of the chassis.

Some computers use more power today, some use less.

If you walk down to your local electronics store, there's probably a few passively cooled laptops on the shelf. Not to mention mobile devices.

And we don't use CRTs anymore which is also a significant power savings.

More passive cooling nowadays would be great, I am totally with you there. But the 80s were awful. I drank water from lead pipes. The fuel economy was atrocious. Murder rates through the roof. They didn't have statins so my dad's friends had heart attacks in their 40s, despite not smoking or anything. Colonoscopy? What's that? Everything is truly better. Sadly, democracy is not advancing as fast as technology though. I would say the 80s had some fun tunes and cute artistic tastes that I still enjoy. Hope not to forget those.

One of my dream projects is a graphical, single-user operating system for, say, small ARM microcontrollers. You can easily get 10x the computing power of an Atari that runs in tens of milliamps.

The problem with microcontrollers for this purpose is memory, not processing power. It's typically measured in KB.

I think the problem is those protocols were arbitrary implementations due to the desire of companies to try to lock-in developers and consumers. This led to the constant churn of hardware and software we all experienced even with our smartphones. Now, it just seems silly but now I think we need as a community (developers, even electrical engineers who work on hardware) need to hammer out the core implementations but that's just a pipedream obviously. :)

Regarding the languages it is easy, hence the pressure of going back to AOT or at very least JIT caching across runs.

https://www.johndcook.com/blog/2012/04/11/facebook-and-cpp/

> One reason Facebook is shifting more work to C++ is to reduce operating costs such as power consumption per user.

Good reference, but that post is from 2012. There are better choices nowadays. Language-provided memory safety also allows you to run more of your code within a single address space (barring information disclosure concerns, aka Spectre/Meltdown of course) - this is effectively what modern "async" programming is. Not that different from the kind of code you would write for a random 16-bit machine.

What is modern async programming? I/O multiplexing with the select system call came out in 1983 and async programming has been exactly the same ever since.

Edit: I misunderstood your 16-bit comment. Yes, in the 80s and 90s async was popular like it is today because processes were expensive. Even on 32-bit. So multiplexing was a good strategy whether in a preemptive OS like SunOS or a cooperative OS like 16 bit Windows. Then Java and various web server architectures came along and people got comfortable with threads, and the OSes got relatively efficient with them.

Reply on news.ycombinator.com