Timeline for Why are old CPUs like MOS Technology 6502 and Motorola 68000 considered better for real time systems applications than modern x86 based CPUs?
Current License: CC BY-SA 4.0
31 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 28, 2020 at 7:53 | answer | added | ghellquist | timeline score: 2 | |
| Oct 19, 2020 at 17:48 | answer | added | Chris Stratton | timeline score: 4 | |
| Oct 19, 2020 at 15:32 | comment | added | bobeff | @Technophile "Considered better" at least by the author of the quoted paragraph in Wikipedia. :) | |
| Oct 19, 2020 at 15:28 | comment | added | Technophile | "Considered better" by who or what? "Consider" is a verb; someone or something is doing the considering. Their knowledge base, the way they think and any agendas will be part of any conclusions they reach. | |
| Oct 19, 2020 at 2:17 | comment | added | user4574 | The 8086 itself had only a small instruction cache of a few words. And it was quite easy to completely disable interrupts (other than NMI). So timing for that processor can be very predictable. But for 80386 and onward timing certainly becomes less predictable. Though windows is not real time. There are real time operating systems for x86 processors. | |
| S Oct 18, 2020 at 18:49 | history | suggested | Peter Mortensen | CC BY-SA 4.0 |
Copy edited (e.g. ref. <https://en.wikipedia.org/wiki/Operating_system>). Fixed the question formation - see e.g. <https://www.youtube.com/watch?v=t4yWEt0OSpg&t=1m49s> (see also <https://www.youtube.com/watch?v=kS5NfSzXfrI> (QUASM)).
|
| Oct 18, 2020 at 17:38 | review | Suggested edits | |||
| S Oct 18, 2020 at 18:49 | |||||
| Oct 18, 2020 at 3:04 | comment | added | R.. GitHub STOP HELPING ICE | They're not (the premise is incorrect). Someone just didn't know what they were talking about when they wrote that paragraph. | |
| Oct 17, 2020 at 23:59 | comment | added | Robbie Goodwin | Don't you think taking three paragraphs out of a significant work begs you to elaborate why there's a problem? | |
| Oct 17, 2020 at 20:59 | comment | added | Ken Shirriff | I've tagged that paragraph in the Wikipedia article as "disputed". It's a mush of uncited claims of varying accuracy that haven't been fixed since 2012. | |
| Oct 17, 2020 at 15:23 | vote | accept | bobeff | ||
| Oct 17, 2020 at 13:19 | comment | added | Tonny | It is perfectly possible to run a realtime OS on top of Intel hardware. You just need the right OS. In the mid 90's I worked on a industrial controller. This used a 25MHz 486sx with AMX as OS. I also used OS9000 (an Intel port of OS9) back in the day, but I absolutely hated that one. (The assembler used Motorola syntax even with Intel assembly. So all the source/destination were swapped in the assembly code. I can do M68000 and x86 assembly, but this pastiche was a nightmare to use.) | |
| Oct 17, 2020 at 12:27 | comment | added | dave | Careful with the terminology though - the IBM PC AT had a "real time clock" which simply remembers date and time when the OS is not running. The timing-interval device was the "programmable interval timer". These days there's a "high-precision event timer". | |
| Oct 17, 2020 at 10:51 | comment | added | Thorbjørn Ravn Andersen | Old in itself is not a bad thing. | |
| Oct 17, 2020 at 10:26 | comment | added | marshal craft | I think real time clock is one important thing. You could have a faster, more powerful processor, but if you can't accurately know it's frequency, you can not in any way insure periodic handling of anything beyond the error in the frequency. Maybe someone can elaborate? | |
| Oct 17, 2020 at 10:17 | comment | added | marshal craft | Modern day mpus and cpus designed for applications which require real time, often have a real time clock, which is different then the operating clock. While it may not be faster, I presume it does have a more precisely defined frequency. Allowing time values to be accurately known. They usually have interrupts features. | |
| Oct 17, 2020 at 10:01 | comment | added | UncleBod | When you read Wikipedia, it is important to remember that the information often can come from unverified sources. Especially when on the top of the page you have "This article includes a list of general references, but it remains largely unverified because it lacks sufficient corresponding inline citations. Please help to improve this article by introducing more precise citations. (April 2014)" The information is already at least 6 years old, and still don't have citations to verify the information. This question ought to be something like:"How to rewrite the following Wikipedia article?" | |
| Oct 17, 2020 at 9:22 | comment | added | None | More than the CPU, it's also because of the type of memory used. Unless SDR, a DDR access is nondeterministic and varies. | |
| Oct 17, 2020 at 9:06 | comment | added | Mark Morgan Lloyd | I agree with @anotherdave there. It would be interesting to look at the edit history of that article, I suspect that that paragraph has been added by a home-computer fan rather than somebody appropriately skilled in the art. | |
| Oct 17, 2020 at 7:19 | comment | added | Cort Ammon | I think one thing that's really important to keep in mind here is that if you have a real time problem that fits on a 68000, you can probably run it on a modern chip and accidentally meet those realtime requirements, simply because the chips are so much faster. 'course if you have a problem that needs the power of a modern CPU, it will be hard to make it realtime with the complexities of a modern CPU... but you'd also simply fail said problem on a 68000! | |
| Oct 17, 2020 at 3:51 | comment | added | Todd Wilcox | I think "modern" doesn't belong in the question title. Nothing about the body of the question seems to relate to what I would call "modern" x86 (really x64 but that's just confusing) processors. For one thing, the PIC mentioned in the quotes has been superseded by a much more sophisticated system. For another, the clock speeds and parallel processing cores in a modern Intel compatible CPU means you can't really compare with a 68000 series architecture any more. | |
| Oct 17, 2020 at 1:59 | history | became hot network question | |||
| Oct 17, 2020 at 0:13 | answer | added | Chromatix | timeline score: 10 | |
| Oct 16, 2020 at 22:35 | comment | added | dave | That Wikipedia article seems to mix up interrupt architecture and OS structure, The second is irrelevant, you don't use a non-real-time OS if you want real-time response. But the paragraph in question is about "using your home computer", so perhaps their self-imposed restriction is using it with the vendor-supplied OS. | |
| Oct 16, 2020 at 19:53 | answer | added | tofro | timeline score: 27 | |
| Oct 16, 2020 at 19:42 | comment | added | tofro | A 8086 to 80586 CPU isn't any more "modern" than a 68k CPU - they were contemporary. | |
| Oct 16, 2020 at 18:40 | comment | added | Brian H | Real-time systems aren't about the latency - it just needs to be low enough. It's mostly about the predictability. The jitter in the latency, if you will. So controlling every aspect of execution so that timing predictions are accurate and consistent... | |
| Oct 16, 2020 at 18:26 | answer | added | RETRAC | timeline score: 22 | |
| Oct 16, 2020 at 18:25 | comment | added | Justme | I don't think it has nothing to do with the CPU at all, or even the interrupt controller, but running on a real-time OS or on bare metal without OS at all, and Windows is not a real-time OS. If you make a 8086 system that runs with code on ROM and data on SRAM, it would be a very real time system, as not even DRAM refresh would steal CPU cycles. | |
| Oct 16, 2020 at 18:23 | answer | added | supercat | timeline score: 52 | |
| Oct 16, 2020 at 17:54 | history | asked | bobeff | CC BY-SA 4.0 |