Talk:Real-time computing

From Wikipedia, the free encyclopedia

 This article is within the scope of WikiProject Computer science, which aims to create a comprehensive computer science reference for Wikipedia. Visit the project page for more information and to join in on related discussions.
Start rated as Start-Class on the assessment scale
High rated as high-importance on the assessment scale

Contents

[edit] Hard vs. Soft

The distinction between hard and soft is incorrectly stated. An RTOS is hard when the consequences of not meeting time constraints are severely unacceptable -- typically, medical and other life-threatening situations. Most commercial RTOSs form soft systems, not hard. (I have the time neither to correct the mistakes nor to provide the requisite references.) P00r 04:03, 1 May 2007 (UTC)

[edit] Too soon can be as bad as too late

I'm not going to edit the entry, but I want to make an observation. There are real-time systems in which failure can occur if an operation starts too soon. I deal with such systems frequently. One example is a half-duplex serial connection where one side must delay sending data until the other side has had time to tri-state the shared communication line. Another is a system where one process reads data that must arrive from another from within recurring assigned time windows. An example might be a simple star network, where a central processor takes data from a number of external processors, each of which has well-defined time windows assigned to it. Jive Dadson 21:10, 20 June 2006 (UTC)

[edit] Proposed merger or other reorganizations

What's up with real-time? Its talk page links here, and most of the useful information in it is better explained here or in the RTOS entry. Some of the information on it (Gartner forcasts a fast increase...) is out-of-place, or at the very least needs to be properly cited. And the whole entry needs better organization. I think either we should try to merge "Real-time computing" into a section of "Real-time", or delete the latter. piman 03:09, 2005 Feb 27 (UTC)

My vote is to merge Real-Time computing" to Real-time" Symphonic
I'm much more active on the Dutch Wikipedia than I am here, but I can't help myself today: The real-time-article is also used to explain some other real-time activities, like [[real-time]] chat on WP:IRC. So you guys might want to take this into account in discussing wheter these two articles should be merged. MigGroningen 20:15, 19 February 2006 (UTC)
I agree, merge. This article primarily relates to computing. It should be merged, and "Real-time" should either be what's Real-time (media) now, or some overall explanation of the concept (which I don't think has been written here yet), or just a disambiguation page. --Howdybob 13:53, 16 May 2006 (UTC)

I should elaborate. I think there should be a conceptual definition of "real-time" which applies to everything. I should be able to say "this works in real time" or "it speeds up to thirty times real time." I don't think that's anywhere now. The same article should also cover what it means as an adjective in general, and links to the various things like computing and media. Nobody has really defined the concept of "real-time," just given examples and applications. Howdybob 14:49, 16 May 2006 (UTC)

Realtime currently is going to a disambiguatiuon page, which I feel is incorrect. Being bold, I'll have realtime land here. Electrawn 01:37, 13 September 2006 (UTC)

I feel the disambiguation is necessary, as there exist many different concepts of real-time. Why should it be incorrect?? Fuffzsch 13:52, 7 November 2006 (UTC)

[edit] Other subjects

I would like to ask how does the real time operating system achieve the task in real time.How does it make sure that the given task is finished in the given amount of time.

Different RTOSs may do different things when a task does not complete in its alloted time (e.g., kill it and log an error, abort it this time and raise its priority for the next time, kill less critical tasks). In some cases they may do different things for different tasks (e.g., a flight critical task may be allowed to complete even if it runs over its alloted time, but non-flight critical tasks get killed to give the flight critical ones more time in the future). These are usually designer/programmer decisions and the RTOS must be properly configured to meet the system requirements.
But sometimes the RTOS doesn't even get a say in the matter, if the task doesn't complete in the alloted time it will never complete due to external "real world" events. The best example I ever heard of this was someone's description of a real time data collection computer they were working on that came up in a conference on design methodologies: "I don't think you understand what I mean when I say real time system. The system I am working on is attached at one end of a one meter long steel rod, at the other end of which is a nuclear bomb. It begins collecting data and transmitting it over a high speed link to another computer just before detonation. This system MUST complete all tasks before the radiation from the explosion destroys the processor. That is what I mean by real time system." That is an example of a true hard real time system. -- RTC 00:20 Apr 24, 2003 (UTC)

Real-time can be changed to make operating systems much stronger, if only we had time.


How does it make sure that the given task is finished in the given amount of time.

An excellent question. Say I ask the computer to do some complicated task, and I see with my stopwatch that it takes 2 seconds to finish. Clearly, if I give it a deadline of 0.1 second, the task is going to overrun the deadline.

With most desktop PCs, one demands that the computer do this, that, and the other thing, and we expect it to (eventually) do all of them.

What happens in a realtime system is that one asks the real-time system to do this, that, and the other thing, with a particular deadline (and perhaps on a repeating schedule). Every time you give it a task with a deadline, it checks -- Is there plenty of time to meet the deadline, even in the worst case, taking into account all the other tasks and their deadlines ? If there is any potential for problems, then it politely refuses that task up front. Only if it is certain that it can finish that task (and all the other tasks it has already committed to) will it accept and commit to that task.

Do I need an example ?

--DavidCary 05:16, 20 Jul 2004 (UTC)

[edit] Firm Real-time System

Can anyone explain what is a firm real-time system and how is it different from hard and soft real-time systems? One definition I found was that systems which are soft real-time but in which there is no benefit from late delivery of service. Confused me further...

Why the gratuitous plug for Gartner? - unsigned (please sign your comments)

The concept of "real time" is one which is too general to localise to computing. I agree that two articles about the same thing (RTC) should be merged, but that is best done here and not not at RT, which should either be a philosophical treatment of the abstracted concept (with tangents in psychology {perception}, physics {relativitity}, not to mention any colloquial usage, etc). No specific concept can dominate a more general one, unless its a very old one (hence a seminal term and part of the language tradition); we dont let gamers cant move the Halo (video game) article to Halo. In this case, the RT article has yet to materialize in the abstract form due to its bind to the specific. -Ste|vertigo 20:06, 10 February 2006 (UTC)

The last paragraph is indeed very questionable. Wikipedia is the wrong place for wild guesses about the future and it sounds like pure marketing bullshit blather. I bet whoever added that is affiliated with Gartner or just loves bullshit bingo. --82.141.57.73 01:00, 8 July 2006 (UTC)

[edit] Real-time not a subset of computing!

I cannot follow the idea to merge real-time with real-time computing, as real-time is a broader subject than real-time computing. Although, the matter of real-time is not only a matter of computation. So, neither of these themes are a subset of the other.

There are real-time networks, real-time secure networks. Of course, they do have some software, but that is only part of it.

SAE1962 13:24, 29 June 2006 (UTC)

[edit] The 68000 Was a CPU, not a Computer

Part of the History section reads, "Once when the C64 and later Motorola 68000 were popular...." This seems less than wonderful because the C64 (the Commodore 64) is a model of computer, and the Motorola 68000 is model of CPU (used in the Macintosh and Amiga). I'm going edit this, changing "C64" to "MOS Technology 6502," which was the CPU of the C64 (and the Apple II).

[edit] Modern Air Travel

"It would not be possible to offer modern commercial air travel if these computations could not reliably be performed in real time." This is simply untrue, since "modern commercial air travel" existed long before the development of the microchip; indeed, before even ENIAC. Depends what's meant by "modern commercial air travel"...as a reader, I think "modern commercial air travel" certainly existed when people were boarding DC-3s in the 1930s. Perhaps the original author can clarify.

There has been tremendous misuse of the expression "real time" Most people confuse online rapid response with real time. "Real time" means synchronized with the clock. Notice a NASA missile launch. "T-10 and counting". The missile does not ignite when ready but is poised to ignite at just the right second and every operation is the synchronized with the clock. "Shut off" occurs at just the right time. Collecting telemetry live from an aircraft is not real time. It is rapid and with minimum delay but it is not synchronized with the clock. 68.50.127.210 19:02, 6 July 2007 (UTC)Tom Amenta tom.amenta@aya.yale.edu

[edit] Firm Real-time System

Firm real-time system is a system in which a completion of an operation after its deadline is considered useless, but the system can withstand a few missed operations.

Ehudhal 03:33, 12 September 2007 (UTC)

[edit] Milli second is real time?

What is real time? Is it the interval of one milli second between input and output? One micro second? One nano second? One pico second? One femto second? One atto second?

What is real time to one system may not be so for another. Time is relative. Anwar (talk) 10:07, 11 June 2008 (UTC)