Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Not anymore, but it's only recently that that's been spun as a good thing.

Tell someone around 2000 that a 7ish year old PC would be adequate and you'd get laughed at. 1993 to 2000, for instance: You'd be comparing a Pentium at 60Mhz (some quick googling says 60Mhz was first hit this year) on the high end to a 1Ghz Athlon. Probably >20x real performance improvement (the clock alone is 16.7x).

Mobile devices will catch up and stagnate too, sadly.



There was an odd hole where machine lifetimes got really short. They didn't used to be. An Apple //c (manufactured 1984) with an ImageWriter printer and a second floppy drive was happily in use for school papers, spreadsheets, correspondence, basic accounting, casual games and kids learning to program through about 1996. And it was passed on to another family after that, where it was used for another 3 years or so. Mainly as a Number Munchers/Carmen Sandiego/Oregon Trail/ChopLifter/Moon Patrol machine, I think. And while I'll grant that was a longer-than-average useful life for the time, we weren't the only ones I knew to keep an 80s machine with a 70s core in service well into the 90s.

Then somewhere around when people started to expect Windows on a computer the lifespans got quite a bit shorter, and that seemed to stick into the late 00s. And it's swung back the other direction now... my late 2011 MacBook Pro is good with current Mac OS or current Windows 10 installations, and is only being retired because its soldered-on discrete GPU has started to flake in a way that makes it unusable. (And a replacement motherboard would be stupid expensive.)


I think part of that is the abstraction layers.. The apple][ had a really long lifespan. The IIe had a production run of 11 years and was a slight evolution on a 7 year old design when it was released. During that time you were free to run nearly any apple][ application (excluding the IIgs ones) despite many of them being 10+ years old. Their functionality wasn't reduced and you didn't have to deal with an OS that changed under the applications and dropped support for them. People understood the target and the applications were tuned for it (similar to game consoles).

The advent of hard disks, and installable OS environments changed the model. Outside of zOS, and maybe IBMi, there simply aren't any OSs that actually have provided a fairly stable OS API over time. Even windows has decided they don't care about backwards compatibility despite the fact that for most purposes the core win32 API present in win2k provides basically everything needed to host the vast majority of modern applications. Games might be the exception but people forget the directX was an install-able component until MS decided to tie to to the OS with vista).


> Even windows has decided they don't care about backwards compatibility

I can run 25-year-old Windows 3.1 applications on a 32-bit Windows 10 box and they work correctly. If I have upgraded to 64-bit, I am merely constrained to 22-year-old Windows 95 applications.


Unisys has ClearPath, going all the way to 1961.


You say "recently" and then compare computers from 18 and 25 years ago. That's just about a generation in human years. The comment I replied to said, "For what it's worth, PCs are subject to the exact same slow decline". ARE subject, not WILL BE or USED TO BE.

"Mobile devices will catch up and stagnate too, sadly."

Uh, if they catch up to my point, then they will still be useful at 5 - 7 years, thus the opposite of stagnant.


I almost typed up the comment that you responded to, probably for the reason that 2000 is still "recent" in my mind :/


>Uh, if they catch up to my point, then they will still be useful at 5 - 7 years, thus the opposite of stagnant.

I believe the GP meant that the rapid progress of feature improvement will stagnate, which means older devices won't 'age' as quickly.


My parents kept using their 1989 Mac SE to do most of their typed work up through at least 2005 (a job for which it arguably worked a lot better than many later machines, with simple predictable software and a fantastic keyboard).

Plenty of people used 7-year-old computers in 2000.


Just because someone did it doesn’t mean it wasn’t horrible. You are comparing people who are obviously way behind in technology. There was and is a large portion of people who aren’t tech savvy but not a tech dinosaur.

My grandpa never got a smart phone. But he was never more than 5 years behind in Home computing well into his 80s.


It wasn’t horrible. It was a very well made little machine, and it did just what they needed it to do. I would argue it was quite a bit nicer for what they wanted than a typical PC of 2005. (For one thing the keyboard was dramatically better.)

Gadget nerds have criteria for computers which are not necessarily related to predictability, sturdiness, or fitness for a particular narrow purpose. Instead there is a mix of excitement about technology qua technology, new toys to experiment with, status signaling, etc., in addition to functional improvements for doing real work.

Not everyone shares a gadget nerd’s sensibilities. It’s fine for people to be happy watching TV on a 18" CRT, listening to music on a cassette player, reading paper books instead of a screen, rolling down their car windows with a little crank, playing card games instead of VR games, etc.


Well, yes and no - the Mac SE had a really tiny screen, native resolution was 512x342[1], which I would say is pretty small for composing text of any length on (though I wrote plenty of undergrad papers, including a 110 page thesis on mine).

1: https://everymac.com/systems/apple/mac_classic/specs/mac_se....


But it's largely because today's improvements have gone into lower CPU consumption than core count. But it could have gone core count, it just 99% of users wouldn't know what to do with more cores and would rather take the battery life.


>it just 99% of programmers wouldn't know what to do with more cores

Fixed that for you. Remember that every time you experience an unresponsive GUI anywhere, and that's the easy case.

Beyond that, parallelization is generally not easy.


I think the vast majority of the time unresponsive GUI are because of blocking network or IO calls. A faster CPU won't do anything. I rarely max out my CPU.


A UI shouldn't be unresponsive in that case. It should present a progress bar, or other visual indicator that the UI [thread/actor/process] is waiting for the IO [thread/actor/process]. A UI should only become unresponsive when only one [thread/actor/process] is trying to do everything (I.e. both IO and UI), which should only occur if your application is not written with concurrency in mind.


Mobile SoCs still increase core count but it doesn’t need to increase power consumption.

E.g. Qualcomm SDM845 has 8 cores in total, 4 faster ones that are only used rarely, and 4 power efficient ones.

Also they offload increasing amount of work from less power efficient general-purpose CPUs to other on-chip hardware, DSPs (for many years), and now GPUs (marketing folks call them neural engines or AI chips).


In the year 2000 you could still find mainframes in some companies. They still worked and did their jobs.


As you can in 2018.


Not exactly mainframes, but the amount invested in AS/400 still today astounds me.




Consider applying for YC's Winter 2027 batch! Applications are open till November 2.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: