Real-Time Operating System (RTOS)

Now, in most of our computer systems there can be some delay in the way things get processed, and that's not a big deal. In fact, when we build certain applications and programs, and in certain ways that we incorporate this into our Internet of Things and the way we build our devices, we just expect that there's going to be a delay. In fact, we program for that — that is, we make our programs in such a way that if there is a delay, it doesn't cause any kind of problems. But sometimes those delays can cause serious issues, and for that we need some sort of real-time operating system.

Where the Delay Comes From

In a typical computer, any of the inputs, anything that comes from storage, anything that comes from our network interface card, the GPU — most of the stuff will go into RAM first. They'll go into the RAM, and then it will be scheduled to be run through the CPU. And then once it's done being run through the CPU and it gets processed, it goes back into the RAM before it gets sent out to an output.

So there can be this delay, because a lot of different stuff could be going into the RAM at any given time and waiting for the CPU to process. At any given time there could be a lot of things that are waiting to be processed on the processor, delaying the output of whatever is being processed.

When Delay Is Catastrophic

But there are some applications where a delay can cause significant, really catastrophic issues, that require some sort of precise timing. A good example of this is if you have a computer that's running your car and you step on the brakes, and there's a delay between the time you step on the brakes and the time that the brakes actually kick in. That could be devastating.

And so there's certain applications where there's a precise timing that needs to happen, cases where we need very quick responses, where between the input and output there's a very small delay — just the delay it takes to process the information. This is where we use a real-time operating system.

Classifications

Now, when it comes to real-time operating systems, there are some classifications. All of them have to deal with the tolerance. Sometimes we have really tight tolerances that are very critical, and sometimes they're a little bit more loose and not as critical.

An example of this might be, well, we have a hard classification. That hard classification could be an airplane. A lot of the controls now, when we are controlling an airplane, go to a processor, and then the processor adjusts things accordingly. You can imagine that if I turn the wheel and then there's this significant delay between that and when the plane turns, that could be catastrophic. It could cause a lot of problems, a lot of harm. And so the delay with that needs to be almost as real time as it could possibly be. You can't get 100% real time, because with the processor there's still going to be a delay with it, but it needs to be almost instantaneous. So there's a hard delay with that.

If I'm talking on the phone, we call that a firm. Where we're talking on the phone, there can be a slight delay with it, but for the most part, if there's too much delay, then that's going to be problematic as well. And if we're streaming something like movies, then a lot of times there's some delay that can occur with that, because there's some buffering that can occur.

So when it comes to real-time operating systems, there's these different classifications on how critical it is that there is very little to no delay, or where sometimes there's a little more forgiving with these systems.