Talk:Computer

I think there should be a paragraph on digital computers in relation to the fact that currently all what we currently call digital computers are in fact analog computers simulating a digital computer. Too many people think they are actually digital rather than analog with tolerances to infer a digital behaviour. Suggested paragraph...

"While often referred to as 'digital computers,' modern computing devices are, in reality, analog computers that simulate digital behavior. These devices rely on analog components, such as transistors and electrical signals, to process information. By carefully controlling the tolerances and thresholds of these analog components, engineers can create the illusion of digital behavior, where discrete binary values (0s and 1s) are processed and manipulated. However, it's essential to recognize that this digital behavior is an abstraction, built upon the underlying analog nature of the physical components." 2405:6E00:2229:7BB:112A:78DC:C83D:41C4 (talk) 15:16, 20 December 2024 (UTC)

That's completely missing the point and would be a bad change to make to the article. Andy Dingley (talk) 16:20, 20 December 2024 (UTC)

Actually explaining misnomers is very educational and should be in every article. Wikipedia is supposed to be all factual, and so many people grow up not realizing facts because of such and wikipedia is becoming more and more the source of truth for a lot of people.

Is it the actual fact you believe shouldn't be stated in the article or the way I worded it which should be factual and I doubt any computer scientist or physics expert would say otherwise. 2405:6E00:2229:7BB:112A:78DC:C83D:41C4 (talk) 16:48, 21 December 2024 (UTC)

No, the point is to tell you what is the case, and only what is pertinent. Usually that does not require explicitly ruling out everything that is not the case or misconceptions not potentially implied from a plain reading of the text. You think a distinction one could identify is interesting, but that does not deserve to be extrapolated out to a primary point of focus for readers because you've decided to confuse yourself on purpose about it. Remsense ‥ 论 17:05, 21 December 2024 (UTC)

You probably think ROM is not random access memory. Its worth noting that Buchholz himself acknowledged the potential for confusion between the terms "random access" and "read/write," but the terminology had already become widespread by the time he expressed his concerns.

Seems digital is similar, in this analog world there pretty much isn't anything digital outside of conceptual and any electronics engineer will know that. 2405:6E00:2229:7BB:112A:78DC:C83D:41C4 (talk) 06:36, 22 December 2024 (UTC)

• It's not a misnomer. Digital computers are digital, they work by digital electronics.

Digital electronics is fundamentally analogue. Aspects such as signal noise margins, output impedance, fanout and (into the time domain) gate delay are all analogue and crucial to good design of their circuits. But once these low-level circuits are designed, they can be assembled into larger architectures with little thought as to their analogue behaviour, or at least by codifying this as some basic rules to follow. From that point on, they can be treated as digital functional blocks, without having to consider the underlying analogue implementation.

Once we're into computers, then we're even further abstracted. We no longer see signal levels or logic families, we're reduced purely to a 'bit' or 'word', because we'd never get anywhere otherwise. Andy Dingley (talk) 20:22, 21 December 2024 (UTC)

As stated, digital computers are purely simulated on an analog platform. 2405:6E00:2229:7BB:112A:78DC:C83D:41C4 (talk) 06:34, 22 December 2024 (UTC)

Your observation that the physical world is unideal and noisy is not new, interesting, or useful here. You are trying to shove a fixation of yours where it doesn't belong, into a place where everyone already understands it but does not mention it because it doesn't matter. That's why we say what sources do in proportion to how often they say it. Remsense ‥ 论 06:39, 22 December 2024 (UTC)

The basic building block of any digital computer is the logic gate. The vast majority of these are nowadays made using FETs, which are connected in such a way that they operate either in cutoff mode or saturation mode - the FET either conducts, or it doesn't, and so it acts like a switch - either "off" or "on", usually represented by the binary values 0 and 1. Between these two modes, FETs also have a linear mode within which it is purely analogue and can amplify a signal. The same applies to BJTs, which were used to construct the majority of logic gates until the 1970s/1980s. It is perfectly possible to construct a logic gate using relays; one feature of the relay is that there is no linear mode - the output is either "off" or "on", there is nothing in between. Even if a slowly-increasing input is applied, at some point the output of the relay will snap instantaneously from one state to the other, without any graduation. Logic gates that use either FETs or BJTs must necessarily pass through the linear region when the transistor is changing state, but with careful design, this period is extremely short. One reason that CMOS devices get hot is that when a pair of FETs (see diagram of a CMOS inverter) is changing state, the power rail (Vdd) is briefly short-circuited to ground (Vss) whilst the two FETs are conducting simultaneously.

As an aside, but on a related matter: just over 46 years ago, there was an article in Everyday Electronics magazine^[1] showing how to construct a small AM radio receiver that used only two semiconductor devices: a Mullard OA91 germanium diode used as a detector, and a CD4001B integrated circuit, a device that contained four two-input NOR gates. In the finished radio, one NOR gate was unused (and tied to Vss), and the other three were wired as inverters and connected (basically, by feeding the output back to the input through a resistor) in such a fashion that they operated in linear mode, and functioned as amplifiers. So, it can be said that CMOS logic gates are analogue devices, but when used in a digital computer, the fact that the inputs will be at either of the two supply voltages means that the output will also be at one of these voltages.

I should point out that all digital computers emply a clock signal, which is not just used for synchronisation, it is also used to verify the validity of logic levels. In its simplest form, when the clock is in one state, one set of data wires is stable and the other is in an indeterminate state, during which time these will be conditioned from the first set; and when the clock line changes to its other state, the second set then become stable and the first set will become indeterminate and are then conditioned from the second set. By checking the clock state, you can find which data wires are stable - and it will be found that all of these will be at one or the other of the two supply voltages. None will be at an intermediate voltage: their signals are not analogue, but digital. --Redrose64 🌹 (talk) 21:28, 22 December 2024 (UTC)

yes the output is digital but the input is analog. Its inherently analog simulating digital behaviour. 2405:6E00:2229:7BB:112A:78DC:C83D:41C4 (talk) 02:57, 26 December 2024 (UTC)

Next you will likely say digital computers can only be binary? Digital does not mean binary actually which is a fact. Yet the article is focusing on the fact that most digital computers are binary based. They can be but not necessarily are. But all are analog because analog is also digital. 2405:6E00:223C:91A:E5AD:1513:A920:7051 (talk) 08:09, 17 January 2025 (UTC)