Path reports

Hint: Three defaults decide what you get: which analysis, how much detail per path, and how many paths per group.

With no options, -path_delay is max (setup), -format is full, and one path is reported per path group. Paths are grouped by the clock that captures them, with separate groups for unclocked and unconstrained paths. OpenSTA reference

Hint: In a timing tool, min and max name the two analyses, not the size of anything.

min uses the earliest arrivals, which are checked against hold requirements; max uses the latest arrivals, checked against setup. OpenSTA reference

Hint: Read it as min plus max: the two analyses in one report.

min_max runs both analyses in one report: the setup (max) paths and the hold (min) paths. OpenSTA reference

Hint: There are two counts in this command: one per path group, one per endpoint. This is the first.

The default is 1 per group. Because -endpoint_path_count defaults to 1, each endpoint contributes at most one path, so these are the 10 worst endpoints of each group. OpenSTA reference

Hint: The group count caps the whole group; the endpoint count only decides how those paths may be shared out.

-group_path_count caps the paths per group; -endpoint_path_count caps how many of them may end at the same endpoint. Many of those paths can differ only in rise and fall edges; -unique_paths_to_endpoint keeps one path per set of pins. OpenSTA reference

Hint: The end of a path is its endpoint, and the format keeps only that.

The end format drops the stage-by-stage detail and keeps one line per path end: required time, arrival, and slack. It is the usual way to see how many endpoints fail and by how much. OpenSTA reference

Hint: The default report compresses each clock into a single `clock network delay` line. Ask what expanding it shows.

The default full format shows the clock network delay as a single line. full_clock expands it into the clock buffers on both sides, which shows where skew comes from. full_clock_expanded also shows the path from a generated clock's source. OpenSTA reference

Hint: Fields add columns to each stage; they do not filter which paths are reported.

The extra columns explain a slow stage: a large load, a high fanout, or a slow input slew. input_pin and net add the input pins and the nets as rows. OpenSTA reference

Hint: Read `-slack_max` as an upper bound on the slack of the paths you want listed.

-slack_max keeps paths with less slack than the value, and -slack_min keeps paths with more. Together with a large path count, -slack_max 0 lists the failing paths. OpenSTA reference

Hint: `-from`, `-through`, and `-to` mean here what they mean in a timing exception.

-from, -through, and -to narrow the report the same way they narrow a timing exception: from start points, through intermediate pins, to endpoints. OpenSTA reference

Hint: A path with no requirement has no slack, so ask where such paths are normally put.

They are reported in their own group. Without this option the unconstrained group is not reported at all. An unconstrained path usually means a constraint is missing, for example an output port with no output delay. OpenSTA reference

Hint: By default paths are listed group by group, sorted inside each group.

By default paths are listed group by group, worst first within each group. -sort_by_slack merges the groups into one list ordered by slack. OpenSTA reference

Hint: A path's group is the clock that captures it, unless `group_path` moved it.

A path's default group is the clock that captures it, so this reports paths captured by core_clock. Groups made with group_path are selected the same way. OpenSTA reference

Hint: The name says find, not report. Ask what a script would want back.

find_timing_paths takes the same selection options as report_checks but returns objects for a script. get_property reads their fields, such as slack, startpoint, and endpoint. OpenSTA reference

Hint: Between the full path and one line per end, there is a middle.

summary keeps the start point, which end drops, and omits the required and arrival columns that end shows. It is the quickest way to see which registers the failing paths run between. OpenSTA reference

Hint: Ask what remains when the stages and the numbers are both dropped.

What is left is the start point, end point, path group, and path type. That alone answers which paths would be reported, which is how a selection is checked before a long report is run. Confirm it by reading the header lines: the path type says whether it is a setup or a hold path. OpenSTA reference

Hint: Ask who reads this one.

A script that post-processes timing should not parse the human report, whose layout changes between versions. The JSON (JavaScript Object Notation) form is stable, and -fields is ignored because every field is present. OpenSTA reference

Hint: The report wanted here is the hold report.

Minimum delay paths are the hold checks and maximum delay paths the setup checks, and min_max asks for both. Other timers spell the same option -delay_type, which is why a script moved between tools trips here rather than anywhere more interesting. OpenSTA reference

Hint: Two of these show the clock. Only one expands it cell by cell.

The plain full format shows the data path with the clock arrival as a single number. Adding the clock shows the clock edges, and expanding it shows every buffer in the tree, which is what you need when the skew itself is the suspect. OpenSTA reference

Hint: The other counting option limits how many paths may share one endpoint.

By default one path is reported per endpoint and a handful per group. Raising the group count gives a longer list of the worst paths; raising the endpoint count gives several paths through the same endpoint, which is what you want when one register is failing for more than one reason. OpenSTA reference