IntermediateQuestion 30 of 112

Why should you prefer the -from/-to form of an exception over the -through-plus-clock form?

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

The special form — a startpoint or endpoint given as -through while a clock object is supplied to -from/-to — supports every valid start and end type, but it costs more to evaluate, especially on a large design. Plain -from/-to (SDC) is cheaper and should be used whenever it says what you mean.

Two equivalent set_false_path forms side by side, plain -from/-to naming pins directly versus the more expensive -through-plus-clock form naming a whole clock's fanout — Technical Reference DiagramWhy should you prefer the -from/-to form of an exception over the -through-plus-clock form?

Technical Reference DiagramWhy should you prefer the -from/-to form of an exception over the -through-plus-clock form?

Technical Explanation

There's a special way to write a path exception in which the startpoint or endpoint is given through -through while a clock object is supplied to -from or -to (SDC) instead of a pin, port, or cell.

Why the special form exists: it supports every valid start and end type, letting it express specifications the plain -from/-to form can't reach on its own.
What it costs: it's more computationally intensive to evaluate, and that cost grows with design size, because the tool has less structure to narrow the candidate path set from a clock object than from a specific pin or port.
The default guidance: use plain -from/-to whenever it can express what you mean, and reserve the -through-plus-clock form for cases that genuinely require it.
The broader habit this reflects: several exception forms are functionally correct for a given case, and the right choice is the cheapest one that still says what's needed — the same instinct that shows up throughout exception writing.
Where the cost compounds: it matters most in an exception that lives in the signed-off SDC rather than a one-off report command, because that exception gets re-evaluated on every path search, in every mode and corner scenario, multiplying a small per-run cost across the whole MCMM (multi-mode multi-corner) matrix.

Common Mistake

The Trap: defaulting to the -through-plus-clock form out of habit because it "always works," without checking whether plain -from/-to would express the same intent.

A designer copies a working -through-plus-clock exception as a template for a new, simpler case that only needed a pin-level -from/-to.
Runtime on a large SoC creeps up across dozens of such exceptions once they're all re-evaluated per scenario, and the slowdown gets blamed on the design's size rather than traced back to avoidable exception cost.

Follow-up Question & Model Response

How would you tell, on an existing SDC file, whether a given exception actually needed the -through-plus-clock form or could have been written as plain -from/-to?

Candidate Model Response: Check what object each -from/-to argument names — if it's a clock rather than a pin, port, or cell, that's the special form, and the question is whether the intended startpoint or endpoint set could instead be named directly as pins or ports. If the same set of paths can be captured with set_false_path -from [get_pins ...] -to [get_pins ...] or similar, the plain form is preferable and should replace it. report_exceptions (PT) applied to both versions, comparing the resulting expanded pin lists, is the practical way to confirm they're equivalent before making the swap.

Practical Example

An SDC file for a 2 million instance SoC contains set_false_path -through [get_pins mux_sel/] -from [get_clocks CLK_A] (SDC), written as the special form even though every path from CLK_A through that mux happens to originate from a known, small set of registers. Rewriting it as set_false_path -from [get_pins {regA/CK regB/CK}] -through [get_pins mux_sel/] drops the clock-object dependency from -from, and a full-chip timing run afterward shows a measurable drop in exception-evaluation time across the signoff scenario set, since the tool now narrows from named pins instead of an entire clock's fanout.

Complete STA Handbook