Hacker News new | comments | show | ask | jobs | submit login
Random access memory (2016) (rlfbckr.org)
90 points by bryanrasmussen on June 11, 2017 | hide | past | web | favorite | 22 comments



This is not random access; the latency depends on the position of the read/write head over the spinning platter. Other analogies may be more apt.

Also I guess they could have just as well built that with a Pick and Place machine out of an electronics manufacturing company, but I guess UdK has an emphasis on things being hand-made and looking artsy.


It is random access because the device doesn't have to touch every grain of sand on the way to the one it intends to deal with next.

It's very similar to a spinning-platter hard drive: latency depends on rotational velocity of the platter and radial velocity of the read/write head. That the latency is quite high in this device doesn't make it any less random-access.


> It is random access because the device doesn't have to touch every grain of sand on the way to the one it intends to deal with next.

what you described is called "sequential access"

> It's very similar to a spinning-platter hard drive: latency depends on rotational velocity of the platter and radial velocity of the read/write head.

hard drives & etc. are called "direct access"

> That the latency is quite high in this device doesn't make it any less random-access.

the problem is not that the latency is high but that the latency varies.

from wikipedia: "A random-access memory device allows data items to be read or written in almost the same amount of time irrespective of the physical location of data inside the memory. In contrast, with other direct-access data storage media such as hard disks, CD-RWs, DVD-RWs and the older drum memory, the time required to read and write data items varies significantly depending on their physical locations on the recording medium, due to mechanical limitations such as media rotation speeds and arm movement." https://en.wikipedia.org/wiki/Random-access_memory


On the other hand, latency of DRAM also depends on the physical location of data accessed (precharge, bank selection...). But for DRAM case it only depends on sequence of previous operations and does not change with wall time, for direct-access media it changes with time.


> It is random access because the device doesn't have to touch every grain of sand on the way to the one it intends to deal with next.

This isn't what random access means.

"In computer science, random access (more precisely and more generally called direct access) is the ability to access any item of data from a population of addressable elements roughly as easily and efficiently as any other, no matter how many elements may be in the set. It is typically contrasted to sequential access."

https://en.m.wikipedia.org/wiki/Random_access


I fact hard drives and similar rotating disc media are not random-access according to your otherwise reasonable definition.

Modern (ie. anything that was ever intended for personal computers and probably every harddrive ever sold by IBM as "DASD") hard drive does not known where given sector is physically located on the platter and does not care what is the current angular position of the heads. Instead it seeks to correct track and reads whatever data is there until it gets to sector header with correct sector address and then either reads or overwrites the sector (ie. it does have to touch irrelevant data).


> I fact hard drives and similar rotating disc media are not random-access according to your otherwise reasonable definition.

I don't see that as an issue with GP's definition. Hard drives aren't random access.


This project doesn't just look "artsy" - it is an art project in the first place.

While you could go on about the technicalities of what kind of memory is being represented here, IMO the main reason they called it "RAM" is because it's a term that is known to a layperson; so it immediately rings a bell.


Never mind, just put a small, variable delay after each operation to make it constant time.

Or a large, fixed delay - that way the jitter will be a small proportion of the total access cost.

That should make it random access.


Modern RAM access time is not deterministic either. Certainly not on the modern desktop.


Utterly brilliant! Anyone studying computer engineering should understand that Turing Machines exist outside of electricity, silicon and logic gates. This should be the first lesson of the first class of any computer engineering course.


Your comment made me think of this project that used water to build logic gates and half adders: http://www.blikstein.com/paulo/projects/project_water.html


Except his gates are not regenerative, meaning they lose their function after a few stages.


Are you sure? So long as the water stays in the system, you should be able to chain as many as you want, since only the boolean "water flowing/no water" matters.


Here's one point where it fails. The gates are not perfect: the output of a gate needs to stabilize. During this stabilization, the gate is leaking water in the wrong output pipe.


Something like that time lapse video at the end could be shown to kids even in preschool.


But how is this relevant to HN except as a curiosity?


I'd be curious to know how many bits this fun memory stores :-)


I love how soothing that sound is. I would love to have it going while working in a library study hall or something.


Relevant xkcd: https://xkcd.com/505/


Isn't this what Greg Egan talks about when he explains his "dust theory"?

http://www.gregegan.net/PERMUTATION/FAQ/FAQ.html


Have just looked at some of his other work. Would love to buy a print!




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

Search: