|
does anyone have any idea just how many things have to go right for a modern computer to work
|
#
?
Jul 29, 2023 12:18
|
|
|
|
| # ? May 3, 2024 14:32 |
|
|
I can't believe we did it
|
|
#
?
Jul 29, 2023 12:18
|
|
|
those sons of bitches.. they really did it didn't they. they really made computers that work. computers that turn on and just work. they just went and did it
|
|
#
?
Jul 29, 2023 12:19
|
|
|
we bad programmers have a lot to thank good EEs for
|
|
#
?
Jul 29, 2023 20:24
|
|
|
echinopsis posted:does anyone have any idea just how many things have to go right for a modern computer to work yeah in my experience the number of things that has to go right for a computer to work is n+1, where n is the number of things which are currently working.
|
|
#
?
Jul 29, 2023 21:34
|
|
|
pretty wild that something can keep more or less perfect math for hundreds of trillions of operations just to make sure you know the helg status of james
|
|
#
?
Jul 29, 2023 22:33
|
|
|
^^^ this person gets it
|
|
#
?
Jul 29, 2023 22:42
|
|
|
computer stopped working
|
|
#
?
Aug 7, 2023 05:21
|
|
|
beep boop
|
|
#
?
Aug 7, 2023 05:30
|
|
|
i wonder how many billions of dollars have been spent in total on computer hardware development
|
|
#
?
Aug 7, 2023 05:48
|
|
|
Farmer Crack-rear end posted:i wonder how many billions of dollars have been spent in total on computer hardware development 3?
|
|
#
?
Aug 7, 2023 06:01
|
|
|
lol look at this guy who thinks that computers work
|
|
#
?
Aug 7, 2023 06:07
|
|
|
computers are just overgrown rube goldberg machines
|
|
#
?
Aug 7, 2023 13:50
|
|
|
weirdly i find it more amazing that modern cars work, considering how many extremely different engineering disciplines are intermingling, getting manufactured at a vast scale at reasonable cost. for a computer you only really need to figure out a couple of components to a sufficiently high standard of reliability that you can have billions of them without individual failures.
|
|
#
?
Aug 7, 2023 14:09
|
|
|
Cybernetic Vermin posted:weirdly i find it more amazing that modern cars work, considering how many extremely different engineering disciplines are intermingling, getting manufactured at a vast scale at reasonable cost. and you have to arrange billions of them into increasingly larger components that perform within timing constraints billions of time a second for billions of consecutive seconds at a time under a variety of conditions.
|
|
#
?
Aug 7, 2023 14:43
|
|
|
Internet Old One posted:and you have to arrange billions of them into increasingly larger components that perform within timing constraints billions of time a second for billions of consecutive seconds at a time under a variety of conditions. yeah, but still, i have somehow internalized extremely regular replication as more of a thing than the mess of pieces that goes into a ice car e.g. i think if all manufactured goods disappeared tomorrow i kind of suspect we'd get back to very decent computers faster than we'd get a reliable mass market car again.
|
|
#
?
Aug 7, 2023 14:48
|
|
|
Cybernetic Vermin posted:yeah, but still, i have somehow internalized extremely regular replication as more of a thing than the mess of pieces that goes into a ice car Depends on what you mean by a decent computer and a reliable car. I think we’ll be able to cast and machine parts that make a car that can be reasonably expected to make it across the country without a breakdown before we get the MSI required to make a machine with Altair level power. If we can machine with the precision to get secondary storage more advanced than punch cards then we’re doing pretty good. Then what about a display that anyone would want to use? I gotta admit I don’t know a ton about cars but I know hobbyists manage more impressive feats of unnecessary automotive engineering in their garages than what I see computer nerds cranking out when it comes to semiconductors. Internet Old One fucked around with this message at 15:12 on Aug 7, 2023 |
|
#
?
Aug 7, 2023 15:06
|
|
|
reason i think the car will get trickier is that ultimately the most complex parts of a computer are pretty much manufactured in one go in one room. e.g. a cpu is complex, but broadly knowing how we did it last you could put together a team to work on it on one site and it'd be workable. a car is an absolutely monstrous supply chain by comparison. but, indeed depends on what one means, i may be putting a high standard on cars and low of computers.
|
|
#
?
Aug 7, 2023 15:11
|
|
|
Yeah but how are you going to get all that precise lithography equipment?
|
|
#
?
Aug 7, 2023 15:13
|
|
|
i mean, has to get built. otoh where are you going to get a decent ecu before you get a computer? or if we go really mechanical (losing some reliability at once for sure), you have to design all that stuff before the computer team gets in to help you.
|
|
#
?
Aug 7, 2023 15:15
|
|
|
Cybernetic Vermin posted:i mean, has to get built. otoh where are you going to get a decent ecu before you get a computer? or if we go really mechanical (losing some reliability at once for sure), you have to design all that stuff before the computer team gets in to help you. Carburetors were basically mechanical analog computers
|
|
#
?
Aug 7, 2023 15:19
|
|
|
ADINSX posted:Carburetors were basically mechanical analog computers yep, were also a lot less reliable than modern cars though, which admittedly is me randomly making up what i mean by a "modern reliable car" vs. what i mean by "decent computer" e: that's boring though. while i do think a reliable modern car would need electronics (not that obviously ic's), more to the point: i suspect that the level of reliability currently enjoyed is some 50 years of refinement that is actually really hard replicate even having the broad knowledge of what went into them. Cybernetic Vermin fucked around with this message at 15:30 on Aug 7, 2023 |
|
#
?
Aug 7, 2023 15:20
|
|
|
I suspect a lot of older car unreliability came from other aspects of their design and maybe also manufacturing techniques vs a carburetor itself. I do agree though that a lot of older tech required more routine maintenance though (points ignition for example)
|
|
#
?
Aug 7, 2023 15:46
|
|
|
Cybernetic Vermin posted:i mean, has to get built. otoh where are you going to get a decent ecu before you get a computer? or if we go really mechanical (losing some reliability at once for sure), you have to design all that stuff before the computer team gets in to help you. I was thinking mechanically. I know people were driving across country without breakdowns before the first ECU came out. Interesting this guy was the first to do so but apparently he replaced nearly every piece of the car across the course of his journey so I don’t know if that’s dependable. I imagine a dependable automobile to probably get across the country after a shakedown period and servicing before the trip. I see that someone took a Ford Model A an impressive 22,000 miles. It required some reasonable service work across this trip but I can’t find any record of exactly what or how often. The transmission is mentioned in particular but I think a car requiring transmission work in 22k miles would still be of immense utility to a typical person. I imagine a good computer to be at least altair’s speed plus a keyboard and a display with a movable cursor at least 40 characters wide, plus secondary storage at least paper tape and 32kb of ram or maybe 24k ram if you can get floppy disk storage.
|
|
#
?
Aug 7, 2023 15:51
|
|
|
now that computers work we can focus on making printers work
|
|
#
?
Aug 7, 2023 15:51
|
|
|
Samuel L. ACKSYN posted:now that computers work we can focus on making printers work That’s a post-singularity problem. We have a greater chance of developing warp drive.
|
|
#
?
Aug 7, 2023 16:00
|
|
|
i don't think printers can ever truly work in an universe with this particular set of laws of physics
|
|
#
?
Aug 7, 2023 16:44
|
|
|
armpit_enjoyer posted:i don't think printers can ever truly work in an universe with this particular set of laws of physics ive seen printers that work but they look like this and require dedicated workers 
|
|
#
?
Aug 7, 2023 16:49
|
|
|
technically those are printing machines. the printers are the workers that make the machine go, and they're unionised and i want to kiss them on the mouth
|
|
#
?
Aug 7, 2023 17:18
|
|
|
Yeah they used to have some really good greenbar printers back in the mainframe days and if the MTBF on a part was low they’d actually have people sitting around somewhere doing science to make it better. More recently I worked at a place where we had a high volume laser printer shared between like 10 people and though I don’t remember how often it broke I probably changed out every user serviceable part within 5 years and it wasn’t even used that much and we had similar models sitting out in important places where we lost money every minute it was down. I often complained that we shouldn’t have printers involved in such important processes but somehow that sounded like some insane futurist talk about jet packs and flying cars. I’m sure the same model printers are somehow even shittier now.
|
|
#
?
Aug 7, 2023 17:20
|
|
|
i was thinking a lot more ambitious on both computer and car, but tbh i feel like i am moving some very loosely defined goalposts here, you're for sure not wrong with this take either. it is fun to think about though, in particular trying to put some measure to how hard a commonly available mundane thing actually is to make.
|
|
#
?
Aug 7, 2023 17:24
|
|
|
early days of electronic fuel injection were totally analog using almost all discrete components like Bosch's Druck-Jetronic in the mid 60s and AMC & Chrysler's use of the Bendix Electrojector in 1958
|
|
#
?
Aug 7, 2023 17:25
|
|
|
ah, i will add that one bit of defense though: i do think people vastly underestimate the reliability improvements we've seen since. and how hard-won they were.
|
|
#
?
Aug 7, 2023 17:32
|
|
|
yeah it's hosed up that they do op
|
|
#
?
Aug 7, 2023 17:38
|
|
|
Internet Old One posted:Yeah they used to have some really good greenbar printers back in the mainframe days and if the MTBF on a part was low they’d actually have people sitting around somewhere doing science to make it better. printer manufacturers don't want high-reliability printers. the margin is in the consumables and the longer a printer lives the longer someone might just buy third-party ink/toner for it. we had a dell 5230 (which was basically a rebadged Lexmark if i remember right) which ran for several years and racked up nearly a million pages printed before it got replaced, and for most of its life we bought third-party toner. get companies used to the notion that printers just suck, get them to go with managed printing services where everything's covered so long as you keep buying official consumables, and boom there's your consistent revenue stream. keep refreshing printer models so that A) when the crappy 'old' printer breaks down you've got an excuse to say "well, that unit's just EOL anyway, time to 'upgrade' to the new model", and B) keep it harder for third parties to offer reman ink/toner carts
|
|
#
?
Aug 7, 2023 18:05
|
|
|
armpit_enjoyer posted:i don't think printers can ever truly work in an universe with this particular set of laws of physics the old hp laserjets had a reputation for being absolute tanks. i think reliable printers are entirely technically possible, they're just incompatible with the incentives of capitalism.
|
|
#
?
Aug 7, 2023 18:15
|
|
|
Captain Foo posted:beep boop
|
|
#
?
Aug 7, 2023 18:21
|
|
|
error: 5pm, no longer able to support user, forfitting connection
|
|
#
?
Aug 7, 2023 23:31
|
|
|
Farmer Crack-rear end posted:the old hp laserjets had a reputation for being absolute tanks. i think reliable printers are entirely technically possible, they're just incompatible with the incentives of capitalism. I had a $5 yard sale hp LaserJet with 250k previous impressions on it and it was perfect, good drivers and never failed to print a solid page quickly from cold in a few seconds
|
|
#
?
Aug 7, 2023 23:43
|
|
|
|
| # ? May 3, 2024 14:32 |
|
|
i leave a gift for the elves in the sauna and they make my computer work good
|
|
#
?
Aug 8, 2023 09:30
|
|
