1. Purpose of the gate
Before the specification is frozen, the team shows that the macro can meet its targets on the chosen process and metal stack, from analysis, earlier silicon or simulation on the process models, and that the process design kit is mature enough to start.
A hard macro is built for one process. If the process cannot give the speed, the voltage headroom, the noise or the matching the macro's hardest circuits need, no amount of design effort later will fix it. TR1 checks this before the specification is frozen and the design is committed.
Three criteria apply. Process models and design rules must be available at a maturity fit for the macro, with the foundry's statement of what may still change. Critical circuits must be shown feasible by analysis, simulation on the process models or earlier silicon, with the assumptions written down. Open feasibility risks must be tracked, each with an owner and the evidence that will close it.
The review pays particular attention to the metal stack. A macro uses some of the process's metal layers and leaves the rest for the SoC; the choice affects the macro's area and performance and every integrator's routing. It also looks at kit maturity: a macro designed on early models must be re-checked when mature models arrive, and the plan must allow for it.
The review does not need every risk closed. It needs each one stated, owned and scheduled to close before the commitment it threatens.
10 more sections for registered members.