0
$\begingroup$

For this question, let's say we have a space faring civilization. They are at least 100 years more advanced than us, and are drawing up plans to send colonists to a neighboring star.

The catch is, non volatile computing memory is something that lies outside of their grasp - ie hard drives - and only much more primitive, small, forms of it are available (1950's magnetic tape, punch cards, books, etc.). However, their computing technology is quite advanced, and volatile memory is very potent.

How would their society be impacted by this tech gap, and how would it affect their ability to travel through space?

$\endgroup$
  • 7
    $\begingroup$ Non-volatile memory is older than volatile memory. Before modern integrated circuits, computer RAM was made of magnetic-core memory which is non-volatile. And there is no way on in the universe for a civilization to have integrated circuits but no ROM chips and no flash chips. $\endgroup$ – AlexP Feb 15 '17 at 2:59
  • 2
    $\begingroup$ @AlexP with the reality-check tag, it’s a perfectly good answer to repudiate the premise. So you could develop your comment into a full Answer. $\endgroup$ – JDługosz Feb 15 '17 at 10:34
  • 1
    $\begingroup$ I don't know enough about this to make an answer, but isn't a situation like this why the Dune universe went with human computers (they wanted to limit how powerful computers could be)? How well could this alien society offload calculations and memory storage into themselves, or specially trained and possibly genetically engineered organic beings? There could be a group of people whose only task is to be able to reprogram the core from MEMORY. $\endgroup$ – Jason K Feb 15 '17 at 14:29
8
$\begingroup$

Ah young people these days.

That today we use highly volatile dynamic Random-Access Memory is an accident; volatile RAM is relatively recent, and, I hope, will not last for much longer.

Some of the first commercially available computers used drum memory for RAM. Non-volatile.

For a relatively long time, about 20 years (from the late 1950s to mid-1970s at least, in some places much longer), computers used magnetic-core memory for RAM. Non-volatile. (Friends of UNIX-like operating systems may recall that the file containing the memory dump of a faulty program is called core, because in those days when UNIX was born RAM was magnetic-core memory, and people used to say "core" instead of RAM, for example, to load a program into core, or to dump core.)

Then came large-scale integrated circuits and static semiconductor RAM. Static RAM is only slowly volatile, and it is faster than the highly volatile dynamic RAM which dominates today; unfortunately, it is also more expensive, requiring 6 transistors per bit instead of one transistor and one capacitor.

Coming up to the question, non-volatile static RAM (nvSRAM) is used in some aerospace, automotive, medical etc. applications where non-volatility is important. Non-volatile SRAM enhances the inherent characteristic of SRAM to be only slowly volatile to guarantee that data is kept intact when power is lost. (The more usual solution is to use battery-backed-up SRAM, BBSRAM, which is cheaper than nvSRAM; the energy consumption when not in operation is tiny, so a battery-backed-up SRAM can survive a long time, but it is slower than nvSRAM.)

And then there is flash memory. How on earth a civilization can develop very large scale integrated (VLSI) circuits but somehow miss developing flash is incomprehensible, because flash is the current step in a chain which began with factory-programmed Read-Only Memory (ROM) "as old as the semiconductor technology itself" (Wikipedia), continuing through programmable read-only memory (PROM) which could be written once, through erasable read-only memory (EPROM) which could be erased and rewritten many times but required exposure to ultraviolet light in order to be erased (some of us may remember the EPROM chips with quartz windows), through electrically-erasable read-only memory (EEPROM) which joy of joys could be erased under program control, and culminating with the NOR and NAND flash cells we use today.

The access speed of modern SSDs is improving fast, and the density of other non-volatile RAM technonologies is also improving. Some large computer manufacturers are exploring several ways of making non-volatile RAM large enough, fast enough and cheap enough to do away with the current paradigm and return to the ideal of single-level store computers.

TL;DR: Non-volatile memory is older than volatile RAM, and chances are that the domination of highly volatile dynamic will come to an end in the not too distant future.

$\endgroup$
  • 2
    $\begingroup$ Magnetic core memory was volatile in a way, in that it had destructive read. By reading out a bit, you cleared that bit. Special circuitry was required to reset the bit to the desired value, not entirely unlike how special circuitry is required to regularly refresh dynamic RAM today. $\endgroup$ – a CVn Feb 15 '17 at 12:52
  • 2
    $\begingroup$ @MichaelKjörling: A memory cycle always included the restore-after-read. But the point is that core memory retained its contents indefinitely with no power (and in those days operating systems used to have power off and power on interrupt handlers which essentially made power loss unobservable for a user-level program). $\endgroup$ – AlexP Feb 15 '17 at 13:10
8
$\begingroup$

From a practical standpoint I don't see how they couldn't just use batteries for their volatile memory, it would simply become an additional requirement to ensure your computer stayed powered or you would lose the program.

$\endgroup$
  • 2
    $\begingroup$ More or less this. Computers would be treated more like brains which must be constantly kept 'alive'. If one ever loses power it needs to be retaught everything it knows from scratch. They would likely have very redundant power supplies so they never have to turn their computers off. The biggest problem would not actually be doing calculations, it would be the need to reprogram everything from scratch. Few modern programs are small enough to fit on the kind of storage you allow (though modern magnetic tapes can actually store terabytes of data, they just have atrocious seek times). $\endgroup$ – Alex Stasse Feb 15 '17 at 2:37
  • $\begingroup$ So if the ship completely lost power temporarily, how would they reboot it? $\endgroup$ – Raznarok Feb 15 '17 at 4:11
  • 1
    $\begingroup$ @Raznarok no need to. Event able to "kill" every battery on a ship would kill all living beings anyway. $\endgroup$ – Mołot Feb 15 '17 at 8:51
  • $\begingroup$ Just like the batteries to keep saved game states alive on nintendo game cartridges. $\endgroup$ – Shufflepants Feb 15 '17 at 16:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.