Welcome guest. Before posting on our computer help forum, you must register. Click here it's easy and free.

Author Topic: Fact or False? Windows vs Linux OS interrupts  (Read 6467 times)

0 Members and 1 Guest are viewing this topic.

DaveLembke

    Topic Starter


    Sage
  • Thanked: 662
  • Certifications: List
  • Computer: Specs
  • Experience: Expert
  • OS: Windows 10
Fact or False? Windows vs Linux OS interrupts
« on: August 23, 2014, 07:31:47 AM »
So I was on the phone with a engineer from a international automation company that I cant specify here, because I dont want to bash the company name for what I feel is a engineer who might not know what he is talking about. While I myself do not know everything, the conversation I had about a problem with software with statistical reporting from sensors throughout the entire automation machine seemed wrong.

So the conversation started about me asking if there was any known bugs with the error reporting software that keeps statistics on the entire machines hundreds of sensors which record a run log as well as tach speeds, and ms delays between objects traveling between light barriers and what caught me as wrong, but maybe this engineer is correct is that he stated the following:

Windows is horrible for that machine because Windows OS ( XP ) can only handle a maximum of 10 interrupts at a given period of time to record up to 10 pieces of data sampled per sample interval, and yet he claims that Linux is better because it can handle a maximum of 1000 interrupts at a given period of time to record up to 1000 pieces of data from the various sensors. And so if you have your machine screaming at you from more than 10 locations with data reported back, it has to consolidate the errors as general error codes vs specific error codes, meanwhile if the machine was controlled by Linux OS, it wouldnt be overloaded and it would be able to handle say the 20 items at the same interrupt interval.

To me this 10 interrupt per sampled interval seems very small, and I would think that better programming is needed really which I was being blown off with the 10 interrupt smoke screen that others with less computer electronic and OS experience would just go with as it came from an engineer and so it must be true.

I would think that if you had such a limitation in how many pieces of information that could be reported back at a given period of time, that you would engineer a buffer of some sort either in programming code to store it locally or an intermediary device that is a data logger that communicates say serially to the PC. This data logger then can consolidate or spill its entire run of error logs and statistical information about speeds of  tachs and delays in ms etc.

BUT, the fact that the Hardware itself IS capable of handling all this info and it comes down to the difference of Windows XP and Linux OS in which there is a claim that the Windows XP OS is acting as an information gathering bottleneck basically, just had me shaking my head while I was on the phone like... ARE YOU SERIOUS!!! ?

To me it seems like they have some coding issues and they are just trying to sell us on the Linux version vs fixing flaws in their statistics portion of the software. Unfortunately while I do like Linux personally better, the business I work for is not going to be swapping out Windows XP for Linux based systems on their isolated from internet secure LAN automation machinery.

But for the fact that I do not know everything, I figured I'd ask if anyone knows about such an Operating System related bottleneck of Windows XP vs Linux in ability to gather information using exactly the same hardware between OS's?

BC_Programmer


    Mastermind
  • Typing is no substitute for thinking.
  • Thanked: 1140
    • Yes
    • Yes
    • BC-Programming.com
  • Certifications: List
  • Computer: Specs
  • Experience: Beginner
  • OS: Windows 11
Re: Fact or False? Windows vs Linux OS interrupts
« Reply #1 on: August 23, 2014, 07:50:56 PM »
It sounds like their XP version is simply crappier. Neither XP nor Linux have any control over how frequently interrupts occur, gathering information from sensors would not entail trapping interrupts anyway. I agree that it was a smokescreen- make up some gobbedygook and hope you go away.

Quote
To me it seems like they have some coding issues and they are just trying to sell us on the Linux version vs fixing flaws in their statistics portion of the software.
Bingo.
I was trying to dereference Null Pointers before it was cool.

Geek-9pm


    Mastermind
  • Geek After Dark
  • Thanked: 1026
    • Gekk9pm bnlog
  • Certifications: List
  • Computer: Specs
  • Experience: Expert
  • OS: Windows 10
Re: Fact or False? Windows vs Linux OS interrupts
« Reply #2 on: August 23, 2014, 08:55:22 PM »
I used to work in automation software. But I do not have a quick answer to the question.  An Intel CPU can handle a number of hardware interrupts.

An IBM compatible Desktop PC is NOT well suited to  get time-sensitive data over a network. Unless the data itself is "smart" and has time stamps embedded in the data.

In any case, for a critical application, no standard off-the-shelf  system software should be used except a package that has been made for machine automation. The people who do this use tools available from Intel or IBM. An often a custom motherboard has to be used.

Here is a general reference about the hardware interrupts for  an Intel system.
http://retired.beyondlogic.org/interrupts/interupt.htm#2

System software may disable interrupts for unspecified reasons. This can cause loose of information is some cases. If custom software is out of the budget, then remote controllers should be used to gaiter data cache it for a PC.

At best. expect to have an error of 55 ms on any stock PC. Worse if the OS disables interrupts at random times.

Linux kernel can be customized for a specific application. It is not the same as a typical Linux distribution.

What the OP describes is often called "machine automation" in the industry. Meaning a software and hardware package that controls a factory of plant that requires constant frequent  samples and adjustments.

Here is a relevant link:
http://www.automation.com/library/articles-white-papers/opc-articles-and-white-papers/3-reasons-linux-is-preferred-for-control-systems

The point here is that Windows was never intended to be a  universal tool kit for everything. Still, you can design a complex system using Microsoft software, but some more hardware is needed

Intel suggests using a system with custom hardware and software.
http://download.intel.com/design/network/papers/30405401.pdf

Read the links.
End of my two cents.  :)

BC_Programmer


    Mastermind
  • Typing is no substitute for thinking.
  • Thanked: 1140
    • Yes
    • Yes
    • BC-Programming.com
  • Certifications: List
  • Computer: Specs
  • Experience: Beginner
  • OS: Windows 11
Re: Fact or False? Windows vs Linux OS interrupts
« Reply #3 on: August 23, 2014, 09:13:14 PM »
An IBM compatible Desktop PC is NOT well suited to  get time-sensitive data over a network. Unless the data itself is "smart" and has time stamps embedded in the data.
If only there was a Transfer Control Protocol That operated over The Internet Protocol that requires Time information to be stored in the data packets. Oh well.


In any case, for a critical application, no standard off-the-shelf  system software should be used except a package that has been made for machine automation. The people who do this use tools available from Intel or IBM. An often a custom motherboard has to be used.

Quote
Here is a general reference about the hardware interrupts for  an Intel system.
http://retired.beyondlogic.org/interrupts/interupt.htm#2
That is the information on hardware interrupts on an 8086 CPU. More recent CPUs are compatible in real Mode but they are NOT compatible at that level in either Protected or Long Mode. Not to mention ACPI and the PCI bus making those hardware interrupts less relevant. Additionally, when it comes to automation software, they will typically use an add-on board. such an add-on board will fire non-maskable interrupts that cannot be ignored regardless of the OS.
Quote
System software may disable interrupts for unspecified reasons.
No, it can only mask certain interrupts. Non-maskable interrupts cannot be ignored or masked, as you ought to know.

Quote
Linux kernel can be customized for a specific application. It is not the same as a typical Linux distribution.
based on the Original Post, the choice is between Linux Software and WIndows Software. The company does not provide a specially tuned Linux distribution, it is simply Linux software. Given the scenario it is almost a certainty, I would expect, that there is a hardware dongle (PCI/PCI-E Card) involved. So really the issues have nothing to do with either OS in particular and everything to do with the software (Either drivers or control software) offered by the vendor. And that is fine.

But when Engineers start lying through their teeth making up garbage, that is when you need to reevaluate your relationship with that vendor. If everybody in the company is as trustworthy as a salesman, who do you go to when you want to understand how things work in reality? Obviously the limitation is not one of Windows itself but a limitation of their Windows software.
I was trying to dereference Null Pointers before it was cool.

Salmon Trout

  • Guest
Re: Fact or False? Windows vs Linux OS interrupts
« Reply #4 on: August 24, 2014, 02:12:43 AM »
End of my two cents.  :)

More like twenty dollars. Well posted Geek!

DaveLembke

    Topic Starter


    Sage
  • Thanked: 662
  • Certifications: List
  • Computer: Specs
  • Experience: Expert
  • OS: Windows 10
Re: Fact or False? Windows vs Linux OS interrupts
« Reply #5 on: August 24, 2014, 05:00:14 AM »
Thanks for the detailed and well thought out information everyone.  8)