h a l f b a k e r yNot so much a thought experiment as a single neuron misfire.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
Today's microcircuits might better be called
"nanocircuits"
--the "line spacing" they can do, while manufacturing
computer chips, today is about 25 nanometers (billionths
of a meter, abbreviated "nm"), and progress is being
made
toward 15nm lines (also sometimes called "rules", like
"ruled
paper" that has lines on it).
Let's do some quick computations assuming 25nm rules.
If
a line is 25nm wide, then the space between it and the
next line is probably also 25nm wide. So we need to
think
about 50nm per "spaced" line. That means 2 lines per
100nm and 20 per 1000nm, which is also one
"micrometer"
or "micron", a millionth of a meter. If we multiply by
1000, then that is 20,000 lines per millimeter and
multiplying again by 10 gives us
200,000
lines per centimeter --the average computer chip is
about
a one-centimeter square.
A square allows lines to cross it in two directions (like
the
grid of a multiplication table), and if 200,000 lines cross
each other, then there are 40 BILLION intersection
points.
So now let us imagine a computer chip constructed in 3
layers (on top of the foundation or "substrate"), as
follows:
The first layer is a simple set of grooves crossing the
chip
surface, and we fill those grooves with wires
("metalization"). The second layer is "epitaxially grown"
(see link) on top of the first layer, and we want this layer
to basically be an electrical insulator, like silicon dioxide
or
perhaps pure silicon.
We now place a "mask" close to the second layer, and
bombard it with appropriate ions. The mask has holes in
it allowing ions to enter the second layer at specific
locations. These ions have the purpose of forming
vertical
electrically-conductive channels through the second
layer.
The third layer is simply a metalization of wires,
completing the grid of potential intersections.
When we send an electric signal down any one wire of
the
first layer, that signal can only reach various wires of the
third layer if conductive channels exist between the
wires
at the "intersection" points. Obviously the mask controls
which intersections become conductive.
In computer terminology, each one of those wiring
intersections can represent "one bit" of data (the
fundamental unit of information theory). If the
intersection is conductive, that could qualify as a "1",
and
if the intersection is nonconductive, that could qualify as
a
"0". Eight data-bits constitute one "byte" of
information,
and so our 40 billion intersections are equivalent to 5
gigabytes of data-storage space.
Well, the average DVD movie disk can hold about 4
gigabytes of data, and is often enough space for just
about any single movie that exists.
As you know, a movie DVD is a "read only" data-storage
device. Similarly, a computer chip constructed as
described above would also be a read-only device. The
data it contains is controlled by the pattern of holes in
the
"mask"; that's where the name "mask ROM" (Read-Only-
Memory) comes from.
Mask ROMs were very popular in the 1980s, early in the
personal-computer industry. They would hold software
that didn't need to be changed, and the data-storage
medium, the chip, was quite a tough and durable device;
the data never got lost through wear-and-tear.
They DID tend to have a problem in that if the software
contained a bug, the company had to make all-new mask
ROMs at a significant expense, and this is the main
reason
the industry has mostly switched to using Programmable
ROMs, such as modern "flash" devices. I'll get back to
this
point in a bit.
Mask ROMs can be pretty cheap if they are made in large
quantities, and it is known that the movie industry
routinely sells copies of movies by the million. It
logically
follows that if a "player" device was constructed to read
chips like those described here, movies need not cost
more than they do now, on DVD. There is also the
advantage that such movie-players would have no
moving
parts, and could last a long time.
Also, from the viewpoint of the copyright holder, this
particular technology will resist piracy. It costs an
enormous amount of money to build a factory to make
computer chips with 25nm lines! The average pirate will
have to look for something else to copy illegally.
Now, what if the mask is flawed, or the chip-
manufacturing process works imperfectly? In this case
the
consequences are far less severe than when an old-
fashioned mask ROM holds a computer program. Here
all
the chip holds is video data. If the data is imperfect,
you
see a momentary sparkle or some other visual glitch on
the
screen. The movie plays on, regardless!
Finally, this Idea offers room for expansion. Consider
again that third layer of wires. Suppose we epitaxially
grow a fourth
layer on top of it, another non-conducting layer. Then
we
use another mask to make conductive spots in that
layer,
and on top of that fourth layer we put another layer of
wires, running the same direction across the chip as the
very first layer. Now we can use the middle layer of
wires
independently with respect to the wires in either the
first or fifth
layer,
and now the chip can hold 10 gigabytes of read-only
data,
not just 5.
Obviously, the above can be repeated for even more
layers, and even more data storage. A 20-gigabyte chip
probably can hold the average Blu-Ray movie, for
example.
And, of course, when they start mass-production of even
finer lines than 25nm, the data storage capacity can go
up
even more.
About epitaxial growth
http://www.mansic.e...ents/PAM1/Ferro.pdf
As mentioned in the main text [Vernon, May 07 2013]
storage with a similar structure
http://www.theregis...l_micron_3d_xpoint/
Unfortunately with respect to this idea, it's ram not rom [Loris, Jul 29 2015]
[link]
|
| |
[bigsleep], so far as I know, they don't actually
make
chips as described here, for mask ROMs,
specifically
for video data. I THINK it can be done easily with
today's tech. To the extent I might be mistaken,
that is the extent to which the details presented
here qualify as "Half-Baked". |
|
| |
I'm aware that the basic concept for this Idea is
kind of old in Science Fiction, but what is new
here are the details I've presented. That should
save this Idea from the dreaded MFD. |
|
| |
[bigsleep], 8GB flash drives are much more
expensive than blank writable DVD disks. As long
as players exist that use such disks, there will be
pirates taking advantage of the fact they can make
cheap copies. |
|
| |
But very few pirates will be rewarded for copying
movies to expensive flash drives. So, if the
technology of chip-movies replaces the technology
of disk-movies (and those machines WILL
eventually wear out, because they have moving
parts), then the pirates will no longer be able to
have a business of making and distributing cheap
permanent copies. |
|
| |
Yes, they can still make limited-lifespan copies on
ordinary hard-disk-drive space for very little cost.
But those are convenient only so long as the hard
disk doesn't break, or the "cloud" doesn't
evaporate. Any time someone has to rebuild a
collection of lost data, which had been
ephemerally stored and hadn't been sourced from
"hard" copies, that is someone who will have a
reason to obtain permanent copies of the data.
Which will be non-pirated, in the future described
here! |
|
| |
The problem is that, for copy protection, it
doesn't matter where the data comes from, it
matters where it goes. As long as it's in a format
that can be displayed (by a computer or TV, or
whatever), the display data (which is digital) can
be copied and re-encoded with minimal loss to
whatever format you choose. |
|
| |
Oh, and I'm pretty sure that it doesn't encode out
to one junction, one bit in a grid, simply because
there ends up being no way for the device to
determine which grid point is active. Think of five
points in a cross pattern, even with the most
intelligent sensing at the edges, you can't tell if
the center point is active or not if all four outer
ones are. It's closer to say you need one layer or
one path per bit in a row. |
|
| |
The main enabler of piracy in terms of films (etc) is that the pirate can produce an essentially equivalent product at a much lower cost - they don't have to make the movie. |
|
| |
So, if you can design a format which costs very little at large scale and inescapably more at small scale then it will work to mitigate that advantage.
I like this idea - don't stop people copying the data, just make it uneconomic to distribute. |
|
| |
Flash disks (and other storage media) still seem to be increasing in data volume at an exponential rate. For this to work you'd have to carry a large enough volume of data for it not to become cheap to copy in a few years time. And also, the data must be dense enough to prevent people simply transcoding to a lighter format. Maybe proper 3D image movies would have a large enough filesize. |
|
| |
Also, as MechE points out, some flash disks already are bigger than the numbers mentioned in the idea body. I suspect that this is down to the described grid structure storage being rather inefficient at scale. |
|
| |
[MechE], yes it does code out to one bit per grid
junction. I've worked with this sort of scheme
before, 25 years ago, where the computer
keyboard was such a grid, and the computer
software used a signalling chip to feed current
into one wire at a time, and then tested the
cross-wires one at a time to see if a key was
pressed (which formed a grid-intersection
connection). |
|
| |
In the case of one of these movie chips, you feed
current into one wire on one side of the square,
and as you test each crossing wire, you can detect
whether or not (1 or 0) current can reach it
through an ion-implanted pathway,
thereby accumulating sequential data. Then you
feed current into the next wire on the first side of
the square, and again test all the crossing wires
one at a time.... |
|
| |
Regarding piracy, I do agree that there is a
window of opportunity here with respect to the
fact that flash drives are currently too expensive
for distributing piracy --and the costs are falling.
For a while, the simplest solution may be as [Loris]
described, encoding movies to occupy more
space, requiring higher-density mask-ROMs, and
even-more-expensive flash drives for illegal hard
copies. |
|
| |
Again, it doesn't matter how the video is encoded
on the device, if it can be translated to a display
format, it can be translated back to any other
encoding you want, at normal density for that
encoding. |
|
| |
Okay, I was thinking of reading as a state rather
than sequentially, and yes that will work. I would
have concerns about error checking and error
handling and the required data speed since it's a
purely serial output, but I'm not certain enough to
make a case of it. |
|
| |
The reason why I suggested "proper 3D" films is that such images essentially require an image for what in 2D films is a pixel. Hence they must have enormous data sizes.
Sure, you could convert such a file to a 2D format, but you'd lose the 3Dness. |
|
| |
Regarding the grid encoding - it seems to me that if one raises the voltage of one column then (potentially) data on every row can be read at once.
Now, this parallel reading in part explains why the storage density is low. Essentially what we have is a chip-wide bus of as many bits as rows, where rows is intended to be some large number.
But to actually read that data one would need either a hellavalota pins on the chip, or some transistor logic to serialise the output. So why not put that logic closer to where it's needed? Use an addressing bus of, say, 64 bits, with each position addressing, say, 256 bits - and you can access a maximum of 2^72 bits.
Which is: 2^69 bytes, or 590 exabytes (590*1000^6 bytes).
I'm not claiming that this is enough storage for everything, but it would require only 64+256+power pins, (instead of the 137438953472 pins of a pure grid array of that size)
The real saving in chip real-estate is that the lines to address each bit now don't have to travel across the entire chip. Instead most of the length is communal bus line. Admittedly fabrication would be more difficult (we need switching logic, which means transistors). But I can imagine that this structure could be implemented as a binary tree, so could be quite elegant. |
|
| |
Wildy off-topic but this is cool: open the first link,
and scroll down to page 5, with the TEM image.
Scroll the page up and down, and you get a very
strong illusion that the pattern of black and white
dots is wobbling from side to side. |
|
| |
As Loris pointed out, a pure mask array as described will have way too many pins to interface off the chip. Therefore there will need to be some logic in the chip to allow the data to be read out over a reasonable number of pins. |
|
| |
Adding logic will of course add cost, but might not derail this idea completely. A chip with just a couple interconnected metal layers would be singificantly cheaper than the same size chip that had to go through the processing steps to create transistors. Since the full chip has to go through each procesing step, once you are putting some transistors on the chip, it doesn't really make a big diffrence if you add a few more. So you might as well throw on a little more loic so there can be some encripted handshaking between the chip and player. That way the chip will only give up its data to a trusted player. Of course any system can be compromised, but I suspect this would allow anti-piracy measures that would put anything on static media like blu-ray to shame. And because of the production cost of pirated media, it's possible that no one would even bother to try to break the system for quite some time. |
|
| |
In addition, this format could easily support home movies on flash or write-once media. I think a micro-SD compatible format would be a great, but for movies you buy, they should have an extended case, making the micro-SD card look more like a popsicle stick. That would make it slightly more difficult to misplace the movie, and give enough space for the title in a readable font and maybe a small picture. The player could be desiged to support the entire length of the stick in a groove in the face of the player to protect it and make it easy to see what movie is loaded. A standard small micro-SD could simply be placed in the business end of the groove. |
|
| |
Then again from what I can tell, streaming is fast replacing physical media anyway so I'm not sure it's worth the bother. |
|
| |
Oh yeah, I almost forgot. Aren't most of the world's high tech silicon fabs are in china anyway? Therefore a pirate doesn't need to copy anything. They just need to pay off the right person in china to run a few more batches. The only thing "fake" about the pirated movies would be the fact that no money got sent to Hollywood in the process. |
|
| |