SUOrangeman
Diamond Member
- Oct 12, 1999
- 8,361
- 0
- 0
maybe not analog directly, but why just a digital signal, you could store data in a tri level or quad level signal, the reason digital is good is its accuracy, but a digital audio signal takes alot more space than its analog counter part, if you could draw a sine wave for a sound analog, it would take up 1/8th or less space than its digitized equivilent. Dont make me explain rudimentry physics here to prove myself
SUO,
Those C64 audio tapes were badass. Sometimes if I wanted to drive myself crazy I'd put them in the tape player and listen to them. Waaaay trippy.
Those C64 audio tapes were badass. Sometimes if I wanted to drive myself crazy I'd put them in the tape player and listen to them. Waaaay trippy.
ahh yes.. the reason Digital is NOT everything!
why? well to be specific, if you want to store a specific piece of data, say the number -5, you need to have MULTIPLE bits to show that the number is -, and that it is 5. if the number was just 5, it would be 101 in binary. if you're in 16 bit mode, and you wanted say a -5, then you have 1111111111111011 to represent it.
quite the waste huh? well you really cannot get much better then that (ok go to 8 bit mode, then u get 11111011). in the decimal numbering system, you automatically have 1 'bit' representing the number 5, and perhaps 2 'bits' representing -5.
as u can see, using multiple levels for data storage is immensely better, becuase it makes more sense.
so why don't we do that then? well first, you need storage mediums which can ONLY be 1 of 4 levels, no more, no less.
however, one thing u can do today is this; say you were burning a magneto-optical disc with audio data. what you might do is this: heat up the disc so that the magnetic part can be influenced through a magnet on the 'burner'. then the magnet passes over that area, and the disc cools and KEEPS that data forever (till u heat it again, or break it).
the ONLY possibility for data corruption, is in the accuracy of the magnetic alignment of the magnetic material within the disk. but as u can see, that possibility in devices such as hard drives, is currently pretty small. what you get by going to multiple levels is a higher possibility that that inacuracy will screw up your data.
as u can see, this is only an Audio disc example. this is because it's currently very feasible. however, you can store MUCH MORE data on something the size as a CD through multiple levels of data possibilities.
this is all why when I was about 3 years younger, I concluded that Analogue/multilevel data is destined to be the format for storage in the future, HOWEVER because of the binary numbering system being so great at accuracy per 'bit', it can keep errors out of communication FAR BETTER then other current known methods of communication. therefor, digital for a while, will be the best way to communicate (analogue 'farmer vision' TV gets snow, however at the same time and place, Digital TV would not show nearly as many problems, unless the signal is too far gone. the border between signal, and no signal, is much thinner with binary numbering systems for communication).
why? well to be specific, if you want to store a specific piece of data, say the number -5, you need to have MULTIPLE bits to show that the number is -, and that it is 5. if the number was just 5, it would be 101 in binary. if you're in 16 bit mode, and you wanted say a -5, then you have 1111111111111011 to represent it.
quite the waste huh? well you really cannot get much better then that (ok go to 8 bit mode, then u get 11111011). in the decimal numbering system, you automatically have 1 'bit' representing the number 5, and perhaps 2 'bits' representing -5.
as u can see, using multiple levels for data storage is immensely better, becuase it makes more sense.
so why don't we do that then? well first, you need storage mediums which can ONLY be 1 of 4 levels, no more, no less.
however, one thing u can do today is this; say you were burning a magneto-optical disc with audio data. what you might do is this: heat up the disc so that the magnetic part can be influenced through a magnet on the 'burner'. then the magnet passes over that area, and the disc cools and KEEPS that data forever (till u heat it again, or break it).
the ONLY possibility for data corruption, is in the accuracy of the magnetic alignment of the magnetic material within the disk. but as u can see, that possibility in devices such as hard drives, is currently pretty small. what you get by going to multiple levels is a higher possibility that that inacuracy will screw up your data.
as u can see, this is only an Audio disc example. this is because it's currently very feasible. however, you can store MUCH MORE data on something the size as a CD through multiple levels of data possibilities.
this is all why when I was about 3 years younger, I concluded that Analogue/multilevel data is destined to be the format for storage in the future, HOWEVER because of the binary numbering system being so great at accuracy per 'bit', it can keep errors out of communication FAR BETTER then other current known methods of communication. therefor, digital for a while, will be the best way to communicate (analogue 'farmer vision' TV gets snow, however at the same time and place, Digital TV would not show nearly as many problems, unless the signal is too far gone. the border between signal, and no signal, is much thinner with binary numbering systems for communication).
WELL, i had digital cable for a while, when we got rid of it though we still kept the digital box, but now after watching tv the picture gets shortened and aline of snow appears across the top, and the focus goes a little, we haven't bugged the cable company about it, but yeah digital aint all its cracked up to be, and i think that tri level (trigital??) electronic signals could be used and have better size than digital signal's, and you'ld hafta redesign computers from the ground up, or have dac's and trac's, and i think maybe you could make the signals better from their, have the trigital signal easily go thru a converter to 2 digital signals, and if you went with a quadlevel signal it would convert to 3 digital signals, as for storing the resulting tri level bits, well magnets seem to only have north and south, but i think the probly have east and west pole (or east west poles derivved from the facing of the north and south, and storing a tri level signal on optical media will be easy to adapt to, why you could have lasers read pits for deapth width and length, within a range so optical storage has a huge possibility, due to the acurateness of light
<< well to be specific, if you want to store a specific piece of data, say the number -5, you need to have MULTIPLE bits to show that the number is -, and that it is 5. if the number was just 5, it would be 101 in binary. if you're in 16 bit mode, and you wanted say a -5, then you have 1111111111111011 to represent it. >>
Technically speaking, in a 16-bit computer, 5 is 0000000000000101 and -5 is 1111111111111011, so there's no real advantage to abandoning the 2's complement approach. Integer values takes up space up to the maximum value of the allocated entry. So a 64-bit number - either positive or negative - will always take up 64 bits - whether or not the value is merely "1" and you are 'wasting' 63 of them. I know you know this, Soccer, I'm just pointing out that it's not fair to say this is a wate.
You could simply have a pos/neg bit separate to the number - which would be 101-0 for "5" and 101-1 for "-5". Everything would look better on paper, but you'd pay a hefty price in hardware (and chip cost) by having to implement subtractors in addition to adders. Besides, like I said, this doesn't save you anything.
Intel Flash memory (and maybe a few other companies, but I know Intel for certain) uses multiple levels to encode more than one bit of data into a memory cell. There are noise concerns with doing it in a CPU, and I don't see it happening any time soon - especially not with voltage levels continuing to drop in these submicron days. I don't think that I can agree with you, Soccerman, that it's destined to be the format of the future. In specific devices like Flash, and maybe DRAM, but I definitely don't think it would work in a CPU - not without a huge amount of engineering effort, anyway, and this effort almost certainly wouldn't equal the gain.
Engineers use binary because it's easy. CMOS is binary, it's either on or off and everyone designs everything to switch to rails. At best you might be able to convert the cache arrays on chip to multi-level but this would be at a cost of latency. Probably a substantial latency cost (as a percentage of total latency) if you did the L1 array of a chip, a lesser cost if you did the L2. But it wouldn't be pretty to design or debug... and I wouldn't want to guess what an alpha particle would do to the values - since you would now be corrupting two bits instead of just 1, so ECC would be harder to implement.
Multi-level data storage may be our destiny, but I don't see it in CPUs or DRAMs any time soon. Still, it's a good point to bring up since it may offer one way around the superparamegnetism limit.
As far as the actual subject of the thread, first off there's enough money involved that we will switch to something else when we hit this limit. There might be a capactity plateau (where, for a few years, sizes are constant while everyone works on spin rate or something else), but it's definitely not "all over". Besides, these are TeraByte harddisks we are talking about... that's still a long way in the future.
Edit: I must have been asleep when I first wrote this. There were a bunch of grammar and spelling errors. Anyway, I think I've cleaned it up now.
ahh, you are correct! my mind wasn't running properly!
anywho, a better example would have been a fraction, like 1/100 correct?
as for it (multilevel data) not being feasible currently, you are correct (which is why we don't see it in front of us now).
it will require immense engineering, as well as brain power to solve problems, but I have to say that if those problems become solved (they probably will be if multilevel data becomes a priority), they will change the industry quite a bit.
and finally, with Quantum computing coming around (hey I barely know anything about it), that could also change alot of things, becuase due to quantum physics, the atom can be in more then 2 states (ala binary), which gives it it's ENORMOUS computing power.
anywho, a better example would have been a fraction, like 1/100 correct?
as for it (multilevel data) not being feasible currently, you are correct (which is why we don't see it in front of us now).
it will require immense engineering, as well as brain power to solve problems, but I have to say that if those problems become solved (they probably will be if multilevel data becomes a priority), they will change the industry quite a bit.
and finally, with Quantum computing coming around (hey I barely know anything about it), that could also change alot of things, becuase due to quantum physics, the atom can be in more then 2 states (ala binary), which gives it it's ENORMOUS computing power.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter