Point of Alignment – Submitted by Shabtay Matalon

Maybe this question has been already discussed and I don’t recall the response, but what is defined as the point of alignment in (a pure) SCE-MI 2.0 use model? Assume several transactors needing each a different number of clock cycles to complete their initialization; how each individual initialization requirement is coordinated with the other transactors and how the global initialization time is coordinated with the test on the C side?

Discussed April 27^th

Point of alignment has not changed. It is only dependent on the use of clock macros. If no SCE-MI clocks then it is time zero. Reset/Init sequence is completely visible in 2.0  Point of alignment and completion of reset/init are independent. In this way it becomes the same as a software simulator.

If a transactor cannot accept any messages until it has finished initialization, then it should provide a way to let the testbench know when it is safe to send transactions. This is done at the user level. However, it could be nice to let a testbench know about this in a general manner. Most people believe that this is too small a problem to warrant infrastructure capabilities be added.

Create an IM. Close without a change in the specification.