1. Purpose of the gate
The business case and the first targets from the architecture model say what the chip must do and what it may cost. Before the specification is frozen and the program commits people and money to development, this review asks a narrower question: is the technology the chip depends on ready to deliver those targets? Ready does not mean finished. It means the team has evidence, rather than hope, that the chosen process, the critical IP blocks, the package and the thermal solution can reach the performance, power, area and cost targets within the plan, and that every gap is known and has a credible way to close.
Four criteria apply. Process technology: the target node and its options, such as device flavours and metal stack, are assessed against the performance, power and cost targets, using foundry data at an agreed maturity. Critical IP: the silicon status of each critical block is known, whether it is proven on the target process, on a related process or not yet in silicon, and each gap has a plan. Package and thermal technology: the package technology and the thermal envelope are shown feasible by analysis for the expected power. Open technology risks are tracked, each with an owner.
The review comes before the specification freeze and the product plan agreement for a reason. Changing a process option, replacing an IP block or moving to a different package after the specification is frozen means reworking the architecture, the floorplan estimate and the schedule. Finding that a technology is not ready here costs a re-plan; finding it after tape-out costs a re-spin. The decisions the specification freeze will lock, such as process node and foundry selection, IP make or buy and package technology and substrate, are only as sound as the technology evidence this review examines.
10 more sections for registered members.