Shabtay’s take 04/28 - Just clarification IM. Issue resolved and no action needed.

Committee comment: Still needs to be discussed. Question has not been answered. Do I have to implement it for a simulator.

After further discussion it was agreed that this is not the case and the IM should be closed.

Compliance

Deciding whether simulator support is required for a compliant implementation.

Compliant models can run on a simulator. A simulator does not require an infrastructure linker. For VHDL and older Verilog, something must be done to make the model work in the simulator. i.e. a library that provides all of the necessary functions.

Could think of a public domain implementation for this library. Also could provide an emulation of the emulator running on the simulator.

Using a simulator to debug models - library support. Only bigger vendors will provide the emulator emulator.

Must compliant simulator (and accelerator/emulator) implementations of SCE-MI 2.0 enforce and/or check for adherence to DPI type and function call restrictions?

An infrastructure linker should determine and report any non-compliance in a model. Can only check the HDL side and should only include checks that can be performed statically. Should require no special parser.

Shabtay> Have we agreed on the term "emulation of the emulator"? As we need to support compatibly when moving from one emulation engine to another, similarly we need to provide compatibility between simulation and emulation engines. A vendor may support one/two/three such engines. I don't see a need to draw a distinction between simulators and emulators (besides performance).

Per> This was a term Brian used to describe the difference between pure simulation of a design and simulating an emulator running (emulating) the same design. There is no need to agree on this term as it does not apply to SCE-MI 2.0. Any vendor is free to implement something like this and it will appear to the user as simply another target.

Shabtay> We of course distinguish between a simulator and emulator, but I am concerned if we distinguish between simulating a design and simulating an emulator running the design.

Per> Right. From the point of view of the SCE-MI spec we do not make this distinction.

Shabtay> OK. I agree with this.