What is the difference between a flat and a hierarchical timing run?

Short Answer

A flat timing run loads a design's entire netlist into one analysis, with every gate and wire visible to the tool at once, regardless of which block it belongs to. A hierarchical run instead analyzes some blocks using compact stand-in models — an extracted timing model or a HyperScale boundary model — instead of their full netlists, trading a small amount of representational detail for a large reduction in memory and runtime.

Technical Explanation

Both approaches aim to answer the same question — does the design meet timing — but they differ in how much of the design the tool actually has to hold in memory at once to answer it.

• A flat run treats hierarchy as organizational only. Blocks and sub-blocks exist in the netlist's naming and structure, but the tool sees straight through every boundary — a path can be traced gate by gate from anywhere to anywhere, because nothing is abstracted away.
• A hierarchical run replaces some blocks with models. An extracted timing model (ETM) or a HyperScale block model stands in for a block's internals, so paths crossing that block use the model's input-to-output arcs instead of tracing through real gates.
• Flat is simpler to reason about but costs more. There is no question of whether a boundary model is stale or accurately built, because there are no boundary models — but the memory and runtime cost scales with the entire design's size, all at once.
• Hierarchical is cheaper but adds a dependency. Analysis cost drops significantly, but the results are only as accurate as the block models in use, which means a process for keeping those models current has to exist alongside the analysis itself.
• The choice is not all-or-nothing. A design can be partially flat, with only a few large or especially stable blocks represented by models, matching the approach to which blocks actually need the memory savings.
• Why it matters: as designs grow, a fully flat run eventually becomes impractical, and hierarchical becomes not an optimization but the only way to get a signoff-quality result in a workable time.

Follow-up Question & Model Response

Would you ever need to run a flat analysis on a block that normally uses a hierarchical model, even after the model has been verified?

Candidate Model Response: Yes, specifically when debugging a violation whose root cause lives inside that block. The hierarchical model can tell you a path crossing the block has a certain delay, but not which internal gate or net is responsible, since the model has no internal representation at all. I would switch to a full-netlist analysis of just that block to trace the internal path, then return to the model-based view once the issue is understood and fixed.