We need a hard drive replacment!

[Sparda]

So I was talking to a friend who pointed me to this artical and he argued that it would replace the hard drive, even though it clearly won't becasue optical media is always hellishly slow. Even though he realised he was wrong I did agree with him that hard drives need replacing, and not with opticval storage. I'v alredy come up with an idea to make ultra fast optical storage, obviusly it wouldn't be compatible with any of the current media, but would be based on the same technolagy.

Idea for ultra fast optical storage

Basicly the disk the data is on is transparrent, the 1s and 0s consist of transparant bits on the disk and blacked out bits on the disk (this will of course make manurfacturing these things harder). To retrive the data off the disk direct light is shon stright down on to the disk where underneith it a ultra high res light thing (the things used in digital camera to allow the camera to 'see'), takes a snap shot of the light been shon though it and alayses the image and alocates the bits it took a photo of in memory, so you have ultra fast access to all the data on the disk. Good idea huh?

The only problem with this is that it means the drive has to have as much memory as the disk offers in storage. The other option of course is to just have it remeber the image and alays the image for each bit as requested. THis would mean that it's faster then it would normly be reading the disk, but at the same time doesn't offer maximum speed. There is no reason there couldn;t be a mixture of drives that do remeber each bit in RAM and drives that remebr the image and analys it when requested.

Idea to make hard drive hold twice as much data in the same amount of physical platter space

So I was trying to come up with an idea for replacing the hard drive, and insted I came up with a concept of making the device I was trying to replace hold alot more data, so what? Any way, it's like this see:

So, a bit is either 1 or 0 right? Well, what if you used diffrent numbers to represent diffrent numbers of bits? For example 0 = 00, 1 = 01 (or 10) and 2 = 11. You have alot more strage space. How do we get each bit on the plater to represent 0, 1 or 2? We eitehr have it negativly magnatised (0), no magnatisation (1) or posativly magnetised (2). So, we have worked out how to store twice as much data in the same amount of space...

I just realised this desighn is flawed... i'll continue any way, and if any body can see a way to make it work then good...

How do we make a normal x86 computer work with it? With emulation of course. make the firmware emulate a normal hard disk while in the back end it's actualy doing alot of work to figur out wheather a bit is a 0 or a 1 based on the bit of the hard drive that is either 0, 1 or 2.

Any way, this desighn doesn;t work becasue of this bit "0 = 00, 1 = 01 (or 10) and 2 = 11" You can't have a bit that can represent two bits, so we need a forth bit 0 = 00, 1 = 01, 2 = 10 and 3 = 11, trouble is though, do do you represent 4 bits in a single space on the hard disk?

Any thoughts?

[metatron]

I thing something based on this technology ('Magnetic memory' chip) could be used in the future.

[Sparda]

That seems likly. Hard drives are the major slow down in computers now.

[Arsenic]

I'm sorry if I don't quite understand your idea, but from what I can tell, you don't quite understand how binary works...

[Sparda]

I do understand how binary works, but my idea was representing 2 bits with 1 bit, there by halfing the amount of physical platter space needed to hold the same amount of data.

[barrytone]

I do understand how binary works, but my idea was representing 2 bits with 1 bit, there by halfing the amount of physical platter space needed to hold the same amount of data.

So basically, you're saying use a base 3 numerical system, instead of base 2?

It's an interesting idea, but I'm not sure it would be very practical.

One reason digital communication can be done at very high frequencies (and is very useful) is that it only has two states: on and off. Because of this: there isn't really much chance of error, as the bits are unambiguous.

A problem with using 3 bits is that the electrical signals for transmitting digital data are all based on tolerances. For example: for CMOS based logic, (I can't remember the values for TTL off the top of my head :P ) a '0' is any voltage falling below one third of the supply voltage (Vdd), and a '1' is any voltage above two thirds of the supply voltage. The reason for this is that it would be very hard, if not impossible to use exact voltages.

By introducing a third state, the voltage to signify the said state would have to fall between one third and two thirds of the supply voltage.

For the amount of benefit you would gain from using a base 3 numerical system, the circuitry required would be way too complex, potentially slow, and more prone to errors. If you think about it: if you just want to keep adding bits, you may as well be using an analouge signal ;)

It's an interesting idea, but I think you would be sacrificing speed and simplicity for storage space.

[armadaender]

I understand your rational but here's my problem with the design:

1. I can't figure out a way for the material to be write, re-write capable.

2. Also, if I'm not mistaken you would need the emulator to sit between he drive and the northbridge, thus creating another piece of equipment to be on the mobo. Additionally, this piece of equipment would need it's own bios (os, something - not sure what exactly what) to control and run the device.

I could be wrong about the second part, havn't hit that section yet in my A+ studying.

[Sparda]

The way I thought the media would be writen is by placing the 'disk' (that is the transparent plastic) on to a surface which can have each bit of it heated seperatly, and by apply insense heat for a short amount of time the plastic will become chard, aditionaly, these disk don;t have to be thin, but if they are thicker then normal they must be tested to be sure that direct light does not bend in side the plastic.

Since this is not a normal hard drive the firmware of the hard drive must emulate a normal hard drive and (efectivly) translate the normal instruction that it recives in to it's own special 4 base system, complex yes, but it saves on physical space need to store data. Basicly, the idea is that you can store two bits of data in the same space it takes a normal hard drive to store 1 bit of data. Thus doubling the amount of data that can be stored.

[armadaender]

I like the idea on paper but as barrytone noted, I believe you would be sacrificing speed for storage.

[VaKo]

I can't see a need for the complexity, its just another level to go wrong IMO. If you look at market trends I think the next big leep in storage tech will be solid state optronics.

I'm just wondering how long we're going to be using x86 in hardware though, and if, with your idea it would be worth emulating a normal x86 interface for the drive, or emulating the entire x86 layer in a new archtechture. Itanic was intels shot, but it turned out to be a bitch to program for. Does AMD have any non-x86 plans at all?

[Sparda]

It wouldn;t just applie to the x86 arch, any arch that suppoerts PATA. To that end, yes it would reduce performance. Any way, we need a hard drive replacment. The hyperdrive 3 is prety good, and now supports 16GB of storage, just that you need some form of extrem redundancy so that it's fast to restor the restore the data on the hyperdrive if it losses power.

[Duelus]

Giga-byte iRam? seems like a good idea, but if you go for 12 hours without power, your data is lost.

**edit**

http://www.youtube.com/watch?v=1PiYgBhAkAM&search=i-ram

[Sparda]

Ye, thats why I said you need extrem redundancy.

[Duelus]

A fuel cell powered RAM drive 8)

[Sparda]

Replace it's normal battary with a lithyum battary which lasts a month or so with out recharging. Hopfully it would last a month, since it will only be out puttin 12V and an un knowen amount of wattage, I guess the amount of wattage requeried chnges depending on what the device is doing.

[Snowy©]

Trinary FTW I have considered it many times complex yes but I figured it could be done by adding current flow direction into the equation (negative voltage)

Why has no one yet made a dual laser DVD/CD drive? Raid style :)

I think I read somewhere about a guy who has storage down to the magnetic readings of quantum spin pair (oh yeah that's the MRAM stuff right?)

Also you could use the same principle as the new touch screen tech and shine the light refractively inside the medium then any surface destortions would refract some of the beam downwards (it's the multizonal touch tech)

Edit Oh there was an Isralie Prof who was at CES and E3 everyone made a joke out of him he showed off 1TB compact flash cards and processors that had speeds out of this world atomchip.com

Not a very convincing website though.

Magnetic technology is something that was deliberatly lost for a very long time... there's just so much you can do like free(ish) power

[Jordban]

Yeah I don't think you can leave the 0 and 1s, but good thinking. I mean re-thinking things that we take as concrete ideas today is what makes large advancements in any field.

I think RAM will be the future, like the very large flash chips someone mentioned. A good example of this now is having a lot of RAM and then having auto start programs that load needed files into RAM. Like I believe Word does this, making the program boot extremely fast, even the first start up.