What's wrong with larger than binary code? (Base 4?)

eupfhoriaeupfhoria
Posted:
in Future Apple Hardware edited January 2014
For all of you enlighened people here, why couldn't a company move to say Base 4 code rather than base 2?



Concievably it wouldn't be that tough for wiring to be accurate enough to support 4 levels of light/eletricity intensity, plus base 4 will save alot of space when considering a HD.

(Considering DNA uses base 4 with its ATCG codes)



I'm asking, what is the probability of this happening in the future (yeah, we'll prolly get yetaherz G28s first, but....)
«134

Comments

  • Reply 1 of 68
    jpfjpf Posts: 167member
    I think I read this post 3 times and I'm still confused. Does anybody understand what's he is saying?



    I'm too lazy right now to do any research at google.com
  • Reply 2 of 68
    *l++*l++ Posts: 129member
    [quote]Originally posted by Eupfhoria:

    <strong>For all of you enlighened people here, why couldn't a company move to say Base 4 code rather than base 2?</strong><hr></blockquote>



    I don't think this is the forum for such question, but here is an answer:

    The Russians had computers working in base 3 (simpler than base 4). This is first helpful from a storage perspective where you can store a 32 binary bit integer in ~20 triary bits, a 37% memory savings. However, the complexity of the memory and the CPU ends up costing you more.

    We will probably only see base 4 computing when we move to quantum mechanical computing, as a quantim bit is really a base-4 number.
  • Reply 3 of 68
    nolelinoleli Posts: 29member
    The way I understand it, a chip works essentially by being a huge network of switches. A switch has only 2 possitions (on/off), so all machine code must be in base 2 (0/1).

    I could be wrong, but I'm pretty sure that's the reason in a nutshell.
  • Reply 4 of 68
    krassykrassy Posts: 595member
    as i said... apple to introduce the first quantum computer in august running os x
  • Reply 5 of 68
    agent cooperagent cooper Posts: 157member
    [quote]Originally posted by Eupfhoria:

    <strong>For all of you enlighened people here, why couldn't a company move to say Base 4 code rather than base 2?



    Concievably it wouldn't be that tough for wiring to be accurate enough to support 4 levels of light/eletricity intensity, plus base 4 will save alot of space when considering a HD.

    (Considering DNA uses base 4 with its ATCG codes)



    I'm asking, what is the probability of this happening in the future (yeah, we'll prolly get yetaherz G28s first, but....)</strong><hr></blockquote>



    The main reason computing takes place in binary, is due to the binary nature of a switch - it is either on or off, 1 or 0. So in terms of processing, the switches on the CPU are desendants of large physical switches from the very very very early days of computing.



    Another reason is Magnetics, if you want to store data on a magnetic medium, such as a hard disk platter, you can store each bit in one of two states - Positive and Negative.



    Four and three bit processing are possible, there are chips and switches that have been developed to hold more than two states, but the storage medium, magnetics, is still postive or negative - so there is a translation process from the binary stored in the memory device before it is processed, and it slows down the whole system to the point of inefficiency.



    As another poster pointed out, quantum computing offers new vistas of concievable processing systems, and some very interesting things have been done with quartz crystals for storage of more than two states (imagine having a bit for every shade of colour in the raindow).



    Theres also a lot of talk about optical computing, which would transmit light through fiber optics and not electricity through semi conductors, which would mean much less heat and a lot more flexibility, and instead of having two states to compute with, you would have any color of the rainbow, and any shade inbetween.



    It'll be interesting to see what happens in the future.
  • Reply 6 of 68
    quaremquarem Posts: 254member
    [quote]Originally posted by *l++:

    <strong>



    I don't think this is the forum for such question, but here is an answer:

    The Russians had computers working in base 3 (simpler than base 4). This is first helpful from a storage perspective where you can store a 32 binary bit integer in ~20 triary bits, a 37% memory savings. However, the complexity of the memory and the CPU ends up costing you more.

    We will probably only see base 4 computing when we move to quantum mechanical computing, as a quantim bit is really a base-4 number.</strong><hr></blockquote>



    A Quantum bit is not a base-4 number. Instead a qubit stores both a 1 and a 0 at the same time in a quantum state called superposition. The idea of superposition is what make quantum computer incredibly fast in some instance, such as factoring, and database searching.
  • Reply 7 of 68
    jet powersjet powers Posts: 288member
    I would think the move to a 3-base would be possible in a magneto arena: north/south/null.



    Elctrical: pos/neg/null.



    I would imagine that one reason why things are still 2-base is because they notion of computation stated that way. Call it inertia.



    ting5
  • Reply 8 of 68
    mrbilldatamrbilldata Posts: 489member
    It is too expensive to change now.



    Its like saying if you can seat four across in a car you could get more people in it. It is in fact true, but there aren't any roads to drive that new car on.



    The most tried idea that I have heard of is using tri-state logic. It is already used in many places, but not for things like memory storage. You get 1, 0, and -1. It gets complicated trying to comform to the current 2 state logic since a positive voltage is normally the zero state.



    Of course why stop at 3 or 4 states. Many analog computer models currently exist. Some have actually worked.
  • Reply 9 of 68
    agent cooperagent cooper Posts: 157member
    [quote]Originally posted by Yet Another Registration:

    <strong>I would think the move to a 3-base would be possible in a magneto arena: north/south/null.



    Elctrical: pos/neg/null.



    I would imagine that one reason why things are still 2-base is because they notion of computation stated that way. Call it inertia.



    ting5</strong><hr></blockquote>



    Good point regarding Inertia, if you've had trouble switching from OS 9 to OS X, imagine ehat kind of a nightmare would be involved in switching code from base2 to base3, base4, or any other kind of numeric system.



    As for your thoughts on tirtiary states in magnetics and electronics, i have no idea if that possible or not.. My source is the Discovery channel, and thing are always over-simplified in TV-land. :^)
  • Reply 10 of 68
    jeremiah richjeremiah rich Posts: 853member
    I am by no means an expert here.. But I do not believe that null would work on a harddrive. (Correct me if I am wrong) Does the search head thingy know exactly what its position is? Or something like that.. Because if there was no charge whatsoever, how would it not determine it as a bad sector, or know it exists there? Maybe I am just not making sense...
  • Reply 11 of 68
    animaniacanimaniac Posts: 122member
    This is pretty much fuzzy logic. (not akin to "fuzzy math") DNA based computers a long ways down the road, though they will one day see light since base-4 would allow for more accurate AI, since the decision cylce has not two (0 or 1) but four choices (A, T, C, G). I don't see apple ever using this. By then computers will be no where what we think they are now. This is 50 years down the road at least.
  • Reply 12 of 68
    snoopysnoopy Posts: 1,901member
    Even if a good three of four state logic media existed, it would still have to interface with everything else, which is binary. The complexity burden far outweighs any small benefits.



    Such a discussion belongs with other such topics, like why do we have a base 10 number system? A base 16 numbering system would be much better. It works great for logarithms and computers. Base 10 is a bummer, but you know we will never change.
  • Reply 13 of 68
    eupfhoriaeupfhoria Posts: 257member
    Thnx for the psots everyone, I just wanted people to be thinkin about it.



    Plus, this IS future hardware, just alot in the future



    [ 07-30-2002: Message edited by: Eupfhoria ]</p>
  • Reply 14 of 68
    plague bearerplague bearer Posts: 68member
    Running binary code on a trinary or quadratic system is concevably possible. As long as 2 of the states for the higher powered language equated to the original binary states.



    Trinary could be

    +1, 0, -1

    Easily done, Computers only use half a sine wave now. This just uses the lower half as a negative volt.



    Quadratic could be

    Frequency: 0, 1

    Amplitude(current) 0, 1

    Modulate the voltage and the current.

    This gives you 4 discreet states

    0, 0

    0, 1

    1, 0

    1, 1

    This still only uses 1/4 of the available states in AM/FM signals.

    0, 1 = 0 binary

    1, 1 = 1 binary



    Holographic storage would be a necessity. OK children turn in your data crystals so I can check your homework.



    Edit: Just did the math, If you include negative voltage and negitive current, then you get 9 discreet states, allowing you to do base9 math.



    [ 07-29-2002: Message edited by: Plague Bearer ]</p>
  • Reply 15 of 68
    overtoastyovertoasty Posts: 439member
    [quote]Originally posted by Eupfhoria:

    <strong>For all of you enlighened people here, why couldn't a company move to say Base 4 code rather than base 2?



    </strong><hr></blockquote>





    Lot's of reasons, not the least of which is signal to noise ... in DNA the 4 steps are each fully differentiated, it's a fully quantized system, you can't have an amino acid which is half one and half another ... thus base 4 works.



    But you start trying to slice computer voltages even finer than they already are (to the limit as it is), and the ability of the system to differentiate states breaks down pretty quick - now not always, it depends on the medium you're working with; but hell, to remain consistent, it's best just to slice it as fine as you can across the board ... and reap the energy and storage space bonus anyway.
  • Reply 16 of 68
    myahmacmyahmac Posts: 222member
    hey my dad tried working on a tri based computer back when he was smaller. the biggest advantages you guys havent mentioned are neural networks. neural networks running on dna style computers which arer already being developed by a drop out from stanford would be sweet. they can learn from your data. it doesnt matter what yoiu give them they learn it. almost like a baby. ex if you gave a network several pics that looked like a symbol the network would begin to remember the signal. my friend built one last year, it was slow but sweet. anyway my poin is that a dif. style of logic would allow comp. to think more like humans. since humans neurons work around the same way as a trit based chip would. i am onnly 18 so if any of you pros can better explain what i am trying to say please do.
  • Reply 17 of 68
    tjmtjm Posts: 367member
    Another issue is the actual microchip construction. They are made of transistors, which are essentially electronic switches, with two states: on or off. Transistors have three leads: the source, the drain, and the gate. It is the gate voltage that is varied to turn the transistor on and off, so the clock frequency determines how fast it turns the transistors on and off.



    I'm contemplating how I would go about building a microchip that would have three (or more) stable states and how I would go about setting and reading the state, and frankly I can't figure it out. I'm not sure how to construct a device that could be integrated to the level of millions per square cm, and still keep everything straight. If you have three states, for example, you end up with, say, -1, 0, and +1 as your three states. The problem I see is that "-1" represents a current flowing in the opposite direction of a "+1", so we need a device which can spit out bits of current in both directions. But then how do you keep a "-1" from cancelling out a "+1"? i.e., how do you detect the difference between a "+1/-1" superposition and a true "0"? My head starts spinning when I try to sort this all out, as I have reached the limit of my own electronics education. I'm sure it can be done in principle, but it seems hugely complex for whatever benefit may arise from it.



    The strength of binary logic is in its simplicity. It may require more basic units to do computations than a ternary (or higher) logic system, but the number of ways you can go wrong with the higher number systems seems to rise exponentially. Others who know more may have better insights, but that's my take on it.
  • Reply 18 of 68
    woozlewoozle Posts: 64member
    The reason is pretty straight forward.

    In any given chip process the system is capable of distinguishing a certain voltage level. Current chips are running just over one volt internally.

    That cant be split any further, so having three or four levels means having a higher voltage.



    Then it becomes a trade off of speed ( more data per clock ) versus the increased compplexity of the design and its greater power requirements, and therefore heat dissapation.



    I dont think that those issues weigh in favour of greater than binary systems at the moment. However, if we reach some physical limit ( maybe quantum computing ), then advances in speed must come from an approach other than making the chip smaller.



    All the other comments about interfacing with other elements are valid. To get people to migrate you have to show a significant improvement over their current systems in some way ( performance is easy to measure ), eg: twice as fast for the same price ( see PC's versus Macs ), half the price for the same speed ( see PC's versus Macs ) or whatever.
  • Reply 19 of 68
    smirclesmircle Posts: 1,035member
    Originally posted by Eupfhoria:

    [quote]For all of you enlighened people here, why couldn't a company move to say Base 4 code rather than base 2?<hr></blockquote>

    What a great idea. This would put Apple into the front of the competition immediately. Since 4 is a larger number than 2, the resulting CPUs would be much faster than x86 - incidentially because of the same reason 64Bit CPUs are necessary to stay competitive on the desktop.



    I just wonder why no one ever implemented that?
  • Reply 20 of 68
    jasonppjasonpp Posts: 308member
    The superposition of electrons in up and down states simultaneously in quantum computers increases potential processing power by a huge amount.



    Instead of an array of on/off switches in current machines, qubits multiply the number of operations..



    a 1 qubit qc has 2 "switches" a 2 qubit qc has 4, 3 qubit qc has 8, 4 qubit qc has 16 ...



    you can see at around a few hundred qubits the processing power of a quantum computer accelerates exponentially. We have 7 qubit quantum computers today. There was a discovery recently which may very well lead to massive thousand qubit quantum computing. In that range it's likely we'll be able to simulate every atom in the Earth's weather systems.. so we can predicts weather out a few hundred years with accuracy.
Sign In or Register to comment.