Home Knowledge Base Chip Floorplanning

Chip Floorplanning is the early-stage physical design process that defines the chip's physical organization — determining die size, placing hard macros (memories, PLLs, ADCs, I/O pads), partitioning power domains, defining clock regions, and establishing the top-level routing topology — where decisions made during floorplanning propagate through every subsequent design step and can improve or destroy timing closure, power integrity, and routability.

Why Floorplanning Matters

A bad floorplan cannot be fixed by downstream optimization. If two blocks that communicate intensively are placed on opposite sides of the die, no amount of buffer insertion or routing optimization can recover the wire delay penalty. Conversely, a well-crafted floorplan places communicating blocks adjacent, minimizes critical path wire lengths, and provides sufficient routing channels to avoid congestion — making timing closure straightforward.

Floorplanning Decisions

1. Die Size Estimation: Total cell area + macro area + routing overhead (typically 1.4-2.0x cell area, depending on metal layer count and routing density) + I/O ring area. Die size directly impacts cost (die per wafer) and yield (larger die = lower yield). 2. Macro Placement:

3. Power Domain Partitioning: Blocks with different supply voltages or power-gating requirements are placed in separate physical power domains. Each domain requires its own power switches (header/footer cells), isolation cells at domain boundaries, and level shifters. 4. Clock Region Planning: Define which clock domains cover which physical regions. Minimize clock crossings between regions to reduce CDC complexity. 5. Routing Channel Planning: Reserve routing channels between macros for signal and power routing. Insufficient channels create routing congestion that may be unfixable without moving macros.

Floorplan Evaluation Metrics

Floorplanning is the architectural blueprint of the physical chip — the strategic decisions that determine whether the downstream place-and-route flow converges to a timing-clean, DRC-clean, power-clean design, or spirals into an unresolvable mess of violations.

floorplanning chip designmacro placementpower domain planningdie size estimationblock level floorplan

Explore 500+ Semiconductor & AI Topics

From EUV lithography to CUDA optimization — search the full knowledge base or chat with our AI assistant.