[pecenipicek]

say what you want about it, it works. hardware causes less headaches than software i think. you know the hardware is fine when the system posts, or when you pass memtest86. seems to me there are more layers in software than in hardware, especially on high end systems. i wonder how much cpu power gets squandered on just making every software subsystem plays nice with every other software subsystem.  hardware on the other hand is just vast arrays of the same old stuff. same old logic gates over and over again. increase the word size? no problem, just stick 2 32 bit adders together and you got a 64 bit adder. your just making the bus wider. you might get some proprietary hardware (most gpus) but its all made out of the same old transistors, making the same old logic gates on the same old silicon die.

like with fpgas, a single generic logic cell can be connected to large numbers of other completely identical cells in varying ways, allowing you to spin a custom cpu out a generic assembly line part. someone even spun a feature compatible cray-1 on an fpga which could run native cray code. same chip could easily be reconfigured to be feature compatible with a completely different processor architecture, like x86 or arm (granted im not sure these things have enough cells to be configured as cutting edge chips). and this is all from an array of completely identical parts. its a tech i want to get into eventually.

with software you have an os, which by nature of trying to do everything is always gonna be bloatware. you have to supoport hardware with drivers, you need a framework in which everything operates, then you need to provide ui, multitasking, the ability to load software, and to talk on the net. for some reason the bundle an os with a bunch on applications that you could have obtained yourself. why use wordpad when you can get notepad++? when all this is bundled into an operating system, you dont really have control of the layers, you just trust the software to deal with it all for you. your not allowed to get rid of layers you dont want. just by shear brute force of processing power all this bull**** is instantaneous and you dont notice most of the time that its happening. you just have a machine patiently waiting an eternity for input occasionally entertaining itself with the bull**** as usual. you only ever notice it when you absolutely need the whole computer's resources to solve a single complex problem (like encoding a video) and it takes multiple times longer than should be theoretically possible given the speed of the hardware..

Have fun

[Ghostavo]

say what you want about it, it works. hardware causes less headaches than software i think. you know the hardware is fine when the system posts, or when you pass memtest86. seems to me there are more layers in software than in hardware, especially on high end systems. i wonder how much cpu power gets squandered on just making every software subsystem plays nice with every other software subsystem.  hardware on the other hand is just vast arrays of the same old stuff. same old logic gates over and over again. increase the word size? no problem, just stick 2 32 bit adders together and you got a 64 bit adder. your just making the bus wider. you might get some proprietary hardware (most gpus) but its all made out of the same old transistors, making the same old logic gates on the same old silicon die.

Two words, cache coherence.

[Polpolion]

say what you want about it, it works. hardware causes less headaches than software i think. you know the hardware is fine when the system posts, or when you pass memtest86. seems to me there are more layers in software than in hardware, especially on high end systems. i wonder how much cpu power gets squandered on just making every software subsystem plays nice with every other software subsystem.  hardware on the other hand is just vast arrays of the same old stuff. same old logic gates over and over again. increase the word size? no problem, just stick 2 32 bit adders together and you got a 64 bit adder. your just making the bus wider. you might get some proprietary hardware (most gpus) but its all made out of the same old transistors, making the same old logic gates on the same old silicon die.

like with fpgas, a single generic logic cell can be connected to large numbers of other completely identical cells in varying ways, allowing you to spin a custom cpu out a generic assembly line part. someone even spun a feature compatible cray-1 on an fpga which could run native cray code. same chip could easily be reconfigured to be feature compatible with a completely different processor architecture, like x86 or arm (granted im not sure these things have enough cells to be configured as cutting edge chips). and this is all from an array of completely identical parts. its a tech i want to get into eventually.

with software you have an os, which by nature of trying to do everything is always gonna be bloatware. you have to supoport hardware with drivers, you need a framework in which everything operates, then you need to provide ui, multitasking, the ability to load software, and to talk on the net. for some reason the bundle an os with a bunch on applications that you could have obtained yourself. why use wordpad when you can get notepad++? when all this is bundled into an operating system, you dont really have control of the layers, you just trust the software to deal with it all for you. your not allowed to get rid of layers you dont want. just by shear brute force of processing power all this bull**** is instantaneous and you dont notice most of the time that its happening. you just have a machine patiently waiting an eternity for input occasionally entertaining itself with the bull**** as usual. you only ever notice it when you absolutely need the whole computer's resources to solve a single complex problem (like encoding a video) and it takes multiple times longer than should be theoretically possible given the speed of the hardware..

I'd say they're pretty similar, to be honest honest. I don't need to worry about operating temperature in software, but I need to worry about memory usage (ooh wait you need to worry about that in hardware too). I don't need to worry about self-inductance in software, but I need to worry about parallelisms (oh wait). Anyway, the list goes on. What it boils down to is what you're more comfortable with.

ed: no, actually I'd say that hardware is more headache prone just because if you're dealing with gen-purpose computer hardware you really need to make sure that you're making a viable target for software.

[z64555]

especially the old skool 40-pin dip packages that a lot of keyboards still use for some reason.

It's just like with the 555 timer chip, why design a new chip when you've already got one that works perfectly fine?

[Mongoose]

What I take away from this is that, if civilization goes tits-up and we wind up in a post-apocalyptic wasteland, we're all screwed, because no one will be able to figure out how the hell everything used to work.

[Nuke]

especially the old skool 40-pin dip packages that a lot of keyboards still use for some reason.

It's just like with the 555 timer chip, why design a new chip when you've already got one that works perfectly fine?

well my point was why not use a much smaller/cheaper qfp instead. actually i think the reason is that a lot of peripherals are still manufactured by manual labor. hince throughhole parts as opposed to smt.

[redsniper]

What I take away from this is that, if civilization goes tits-up and we wind up in a post-apocalyptic wasteland, we're all screwed, because no one will be able to figure out how the hell everything used to work.

We figured it out once. We can figure it out again.

Plus now we'll have all kinds of cool stuff lying around to reverse engineer.

[Flipside]

The thing about modern society is that it works on a 'no man is an island' mentality. Almost all high level tech solutions require several experts in several fields.

There was a book called 'Strata' by Pratchett that was quite interesting with regards to this subject, stating that something like a simple Space Rocket is built on a huge pyramid of agriculture, mining, chemistry, physics, communication etc, etc. You need that foundation in place first.

Yes, if we lost everything, we could get it back, but it wouldn't take much less time than it did the first time round.

[The E]

Mostly because we would have to rediscover intermediate steps we have long since discarded from active memory while at the same time not losing sight of the eventual goal. There may be a few shortcuts we can take (or rather, dead ends we can avoid). For example, getting back to the steam age and electricity might not take as long as it did the first time around, but rebuilding the infrastructure needed to make integrated circuits will probably not be that much faster.

[Mongoose]

Yeah, the jump from 4000 B.C. to 1870 or so presumably wouldn't be all that hard, but we'd have our work cut out for us to move through the 20th century again.

[MP-Ryan]

Yeah, the jump from 4000 B.C. to 1870 or so presumably wouldn't be all that hard, but we'd have our work cut out for us to move through the 20th century again.

I beg to differ =)  The knowledge derived from the advances in the 5870 years you've mentioned in agriculture, biology, chemistry, physics, and mathematics are all prerequisites for the advances of the 20th century.  If you wiped out a large chunk of the human race and most of our infrastructure today, there are some major elementary hurdles we'd have to reconquer.

[The E]

Yes, but unlike our distant ancestors, we know that certain things are possible already. And personally, I believe that a catastrophe that manages to wipe out such a large portion of the human race AND our printed knowledge base is pretty unlikely.....

[redsniper]

Yeah, I didn't mean to say that recovering from some cataclysmic disaster would be a walk in the park or anything. I just think it would go a little easier than the first time around when you actually have examples of what could be done lying around. It's marginally easier to look at a thing and think "how can we build this again?" rather than "how can I invent and build something that has never existed before ever?"

[Mongoose]

Yeah, the jump from 4000 B.C. to 1870 or so presumably wouldn't be all that hard, but we'd have our work cut out for us to move through the 20th century again.

I beg to differ =)  The knowledge derived from the advances in the 5870 years you've mentioned in agriculture, biology, chemistry, physics, and mathematics are all prerequisites for the advances of the 20th century.  If you wiped out a large chunk of the human race and most of our infrastructure today, there are some major elementary hurdles we'd have to reconquer.

Well, I was thinking more along the lines of the complexities required to potentially reverse-engineer a machine or process.  Like, I'm a complete mechanical layman, but I at least know the basics of how a simple steam engine works: heat up water, use the steam to push against a piston, do work.  It wouldn't be very sophisticated, but I think a lot of people would be able to jury-rig something basic along those lines.  In the same way, while your average person might know much about genetics, but there are a lot of do-it-yourself gardeners who understand the basics of selecting good crops.  Most people could tell you that boiling water before drinking it kills off germs, or that letting cattle do their business upstream of a town's water supply is a bad idea.  There's a lot of basic knowledge ingrained in people today that would still be there if we had to start over, and I think we'd be able to reconstruct a lot of the basics along those lines without massive difficulty.  It's when you get into the more recent technical advances, things that required many experts in specific fields to develop, that you'd run into problems.

[Nuke]

nuclear warheads tent to point at cities and other missile bases, i figure the rural areas will mostly be unaffected by anything, maybe a little fallout. plenty of small towns and and agricultural areas will remain to keep essential knowledge going. you dont really need to worry about the stuff thats well documented. for example there have been people who could produce diy semiconductors. its the closed proprietary technologies that will be lost. any tech that is documented from theory to production will probibly survive without being need to be made from scratch.

there will be a lot of book burning, not to destroy knowledge but to make fire for heat/cooking. there will be some desire to save books deemed essential for immediate survival, stuff on agriculture, trades manuals, **** like that. i very much doubt a manual on c++ will last very long. for that preservation of knowledge to be ongoing, literacy must be maintained, because when its not you get indiscriminate book burning.

there will also be a number of working computers left lying around. a computer really isnt essential to survival, so they might be salvaged for parts or just used for their raw materials. like making arrow points out of bits of case metal, or hammering the sheet metal flat with a rock to make building material. the surface mount technology is a ***** when it comes to component level work, so most of the boards will probibly be thrown into rubbish heaps. if computers do manage to survive it will be unlikely that they will be used for anything due to lack of power. of course you can fuel a genny with wood, or build a wind turbine with junk, so its possible to have a few places where the technology will be preserved.

engine technology will survive for a couple reasons. the first being that they can be found everywhere. they will be extremely useful for survival as they can do work humans can not and allow for a degree of industrialization. all be it limited by the ability to find fuel (though wood gasifiers are easy to build). what ultimately determines if a technology is preserved or not is ultimately its usefulness for survival.