Compare Runs

Single-run results tell you what happened. Compare mode tells you what changed — pick a Base run, add one or more runs to compare against, and Tigrister lays the metrics side-by-side with percentage deltas, an overlay chart, and a per-step diff table. This is how you spot regressions between deploys, validate optimizations, or prove that a config change actually moved the needle.

Entering Compare Mode

The Compare button in the action bar is enabled when you have a selected run in the Results view and no test is currently running. Clicking it switches the run history list into selection mode and reveals two extra action bar buttons:

  • Cancel — exits compare mode without loading anything
  • Compare (N) — executes the comparison against the N runs you picked; disabled while no runs are selected

The currently selected run in the history becomes the Base. Every other run you tick is added to the others list — you can pick one, two, or as many as you want.

Compare Header

Once you execute the compare, the Results view is replaced by the Compare view. At the top, two cards summarize the selection:

Base Card

Outlined with the primary color. Shows the short run ID, the start date, and the VU count + duration from the run's stored config — a quick reminder of what the base run actually did.

Comparing N Runs Card

One line per comparison run with its short ID and timestamp. The heading pluralizes based on count (e.g. Comparing 1 run vs Comparing 3 runs).

Metrics Comparison Table

The delta table is the headline of the Compare view. Rows are aggregate metrics, columns are runs — Base first (primary color) followed by each comparison run in the order you picked them.

MetricUnitBetter when
Avg ResponsemsLower
P95 ResponsemsLower
P99 ResponsemsLower
Throughputreq/sHigher
Error Rate%Lower
Total RequestscountHigher

Delta coloring: next to every comparison-run value, Tigrister prints the percentage change vs. Base with an arrow:

  • Green ↑ or ↓ — the change is an improvement for this metric
  • Red ↑ or ↓ — the change is a regression
  • Muted (no arrow) — change is below 0.5 %; treated as noise

Response Time Overlay Chart

Below the delta table, the overlay chart plots every run's response-time time-series on the same axes so you can eyeball shape differences — does the spike happen at the same elapsed second? Does the new run keep a flatter tail? Is there a step change halfway through?

The Base line is drawn in the primary color and sits on top; comparison runs use distinct hues assigned in selection order.

Per-Step Comparison Table

The final section of the Compare view is a per-step diff. Every step in the spec gets a row, and each row is expanded across all selected runs so you can see how that specific request behaved in every run side-by-side. This is the fastest way to answer the question:

"Our P95 went up by 40 % between runs — which step is responsible?"

Glance down the per-step table and the culprit stands out immediately — the row whose delta columns bleed red while everything else stayed flat.

Exiting Compare

To leave the Compare view and return to a single-run Results view, just pick any run from the history sidebar — compare mode is cleared automatically. Alternatively, press Cancel in the action bar if you're still in selection mode before executing the comparison.