BeginnerQuestion 4 of 95

What are startpoints and endpoints of a timing path?

From PDVerse STA Mentor Guide, part of the pdVerse Mentor Guide

By Tabish Iqbal, who has 9+ years of VLSI Physical Design experience

Short Answer

A startpoint is where the tool considers data to be launched — a flip-flop's clock pin or a primary input port. An endpoint is where the tool checks that data arrived correctly — a flip-flop's data pin or a primary output port. These are the only object types the tool will accept as path boundaries.

Valid startpoints (flip-flop clock pins, input ports) and valid endpoints (flip-flop data pins, output ports) in a timing path, with an internal gate pin shown as not a valid boundary — Technical Reference DiagramWhat are startpoints and endpoints of a timing path?

Technical Reference DiagramWhat are startpoints and endpoints of a timing path?

Technical Explanation

Knowing the exact list of valid boundaries matters because timing reports and exception commands both reference them directly.

Valid startpoints: the clock pin of a flip-flop or latch (data leaves the instant the active clock edge hits it), or a primary input port (data arrives from outside the chip, timed against a reference clock).
Valid endpoints: the data pin of a flip-flop or latch (must be stable before and after the capturing clock edge), or a primary output port (must meet the receiving chip's own timing needs).
Everything else is just "inside" the path. A combinational gate's input or output pin can appear in a path report as an intermediate point, but the tool will never treat it as the path's edge on its own.
The clock pin, not the data-output pin, is what launches. It's the clock edge hitting the flop that starts the clock-to-Q delay — the flop's Q output is simply the first stop after launch, not the launch point itself.

Common Mistake

The trap: naming a flip-flop's data output pin (Q) as the path's startpoint.
It looks right because Q is where the data physically comes from on the schematic.
But the tool measures the path from the clock edge, so an exception written -from a Q pin instead of the flop's clock pin (or the cell itself) won't match the path the tool actually reports — the command silently applies to nothing.

Follow-up Question & Model Response

Is a flip-flop's asynchronous reset pin treated as a startpoint or an endpoint?

Candidate Model Response: It's an endpoint. During recovery and removal checks — the setup/hold-style checks that apply to asynchronous control pins — the reset pin is checked against the nearby clock edge to make sure the reset is released (or asserted) safely relative to it. Data isn't launching from that pin; the tool is verifying that a signal arriving there doesn't collide with the clock's active edge, which is exactly the role an endpoint plays.

Practical Example

Suppose an engineer writes a false-path exception targeting -from [get_pins reg_a/Q]. The command applies to nothing, because reg_a/Q isn't a valid startpoint in the tool's model. Writing it correctly as -from [get_pins reg_a/CK], or simply -from [get_cells reg_a], lets the tool resolve it to the actual launch point and apply the exception as intended.

Complete STA Handbook