[emulation] What's the difference between emulation and simulation?

Possible Duplicate:
Simulator or Emulator? What is the difference?

In simple understandable terms, what is the difference between the two terms?

[I have already looked at this, this and this]

Please forgive me if I'm wrong. And I have to admit upfront that I haven't done any research on these 2 terms. Anyway...

Emulation is to mimic something with detailed known results, whatever the internal behaviors actually are. We only try to get things done and don't care much about what goes on inside.

Simulation, on the other hand, is to mimic something with some known behaviors to study something not being known yet.

my 2cents

This is a hard question to answer definitively because the terms and often misused or confused.

Often, an emulator is a complete re-implementation of a particular device or platform. The emulator acts exactly like the real device would. For example, a NES emulator implements the CPU, the sound chip, the video output, the controller signals, etc. The unmodified code from a NES castridge can be dumped and then the resulting image can be loaded into our emulator and played.

A simulator is a partial implementation of a device/platform, it does just enough for its own purposes. For example, the iPhone Simulator runs an "iPhone app" that has been specifically compiled to target x86 and the Cocoa API rather than the real device's ARM CPU and Cocoa Touch API. However, the binary that we run in the simulator would not work on the real device.

I do not know whether this is the general opinion, but I've always differentiated the two by what they are used for. An emulator is used if you actually want to use the emulated machine for its output. A simulator, on the other hand, is for when you want to study the simulated machine or test its behaviour.

For example, if you want to write some state machine logic in your application (which is running on a general purpose CPU), you write a small state machine emulator. If you want to study the efficiency or viability of a state machine for a particular problem, you write a simulator.

Coming from the hardware development world. . .

Simulation tests functionality. 2+2 = 4 etc

Emulation tests the functionality on the specific environment (64-bit, 16-bit, fingers and toes).

Here is a food example:

You have two pieces of bread, one knife, peanut butter and jelly and will be giving them to a kindergartner. You write instructions on how to make a sandwich.

In simulation, you would act out the process, pretend you opened the jars, pretend spreading the peanut butter etc.

If at the end of the instructions your are left with only jelly and not peanut butter then you failed the simulation and you need to fix your instructions. On the other hand if you have a complete "sandwich" then the instructions should be valid

In emulation, you would use close representations of the actual parts (same bread, knife peanut butter etc). What happens if you gave your kindergartner a cheap plastic knife and really really thick peanut butter?? The knife would break in emulation and the instructions would need to be clarified or fixed to accommodate this problem. In this case you might suggest warming up the peanut butter in the microwave.

In practice: Consider a 64-bit system that you are programming in and a 32bit system that will actually be running the code. You add two very very large numbers and print the result. In simulation everything works (you managed to get the code right to add two numbers) In emulation however you find that you get the wrong answer. What happened? The emulation of the 32-bit system was unable to handle the large numbers. This is an example of correct functionality (i.e. simulation) but not proper support for your runtime environment (emulation)

Both are models of an object that you have some means of controlling inputs and observing outputs. With an emulator, you want the output to be exactly what the object you are emulating would produce. With a simulator, you want certain properties of your output to be similar to what the object would produce.

Let me give an example -- suppose you want to do some system testing to see how adding a new sensor (like a thermometer) to a system would affect the system. You know that the thermometer sends a message 8 time a second containing its measurement.

Simulation -- if you do not have the thermometer yet, but you want to test that this message rate will not overload you system, you can simulate the sensor by attaching a unit that sends a random number 8 times a second. You can run any test that does not rely on the actual value the sensor sends.

Emulation -- suppose you have a very expensive thermometer that measures to 0.001 C, and you want to see if you can get by with a cheaper thermometer that only measures to the nearest 0.5 C. You can emulate the cheaper thermometer using an expensive thermometer by rounding the reading to the nearest 0.5 C and running tests that rely on the temperature values.

I was confused between the two processes. I found this simple explaination about the difference between Emulators and Simulators

1. Simulator:
   Suppose you have written assembly program in a file and corresponding exe file is ready. The simulator is the pc software which reads the instructions from the exe and 'minmics' the operation of the processor.

2. Emulator:
   Emulator is a (PC software + a processor). The Processor can be plugged into the TARGET BOARD when you want to test the developed software in real time to check run time bugs. When not in use it can be unplugged. The Processor will have a parallel or JTAG interface with the PC for downloading the exe file for execution.

Hence, whereas the Simulator is slow in execution, Emulator will be able to give real time verification of the developed code. Generally you will test your developed code on simulator first and then go for checking on emulator.

source : http://www.dsprelated.com/groups/c6x/show/148.php

Based on software and system engineering experience, I'd summarise the difference as follows:

Simulation: for me, this is always in software - every aspect of the real system is only MODELLED by some code and/or mathematics. Simulation attempts to accurately reproduce the behaviour (or predict it) of the real system, but only approximates it.

Emulation: As opposed to simulation, it does not APPROXIMATE the behaviour of the real system, it COPIES the behaviour of the real system. An emulator may involve hardware. But it may also be entirely in software. E.g. you get these software EMULATORS for old game consoles like the Sega Genesis. That is an emulator because it COPIES the real genesis functionality so much so that you can run the original Genesis code in the emulator. A genesis simulator would not be able to run the original code, it would only APPROXIMATE its behaviour, producing similar results, depending on how good the models of the original system were.

An emulator of a system component can be included in a larger system, completely replacing the component it is emulating - a simulator could not because it is not an accurate enough representation of the original component behaviour.

An "emulator" is a term for a software-based hardware-simulator, but in general the two are synonyms.

Here's an example - we recently developed a simulation model to measure the remote transmission response time of a yet-to-be-developed system. An emulation analysis would not have given us the answer in time to upgrade the bandwidth capacity so simulation was our approach. Because we were mostly interested in determining bandwidth needs, we cared primarily about transaction size and volume, not the processing of the system. The simulation model was on a stand-alone piece of software that was designed to model discrete-event processes. To summarize in response to your question, emulation is a type of simulation. But, in this case, simulation was NOT an emulation because it didn't fully represent the new system, only the size and volume of transactions.