Runtime Insights

Module: Tech · Category: Performance · Tool ID: RuntimeInsightsSuite

Four-tab end-to-end runtime profiling pipeline. Profiler monitors stat unit live with CPU vs GPU bottleneck classification. Trace Analyzer parses .utrace files with per-thread event breakdown. Benchmarks runs spline-based automated benchmarks with CSV/JSON export for CI/CD regression tracking. AI Analysis does LLM-powered trace interpretation (Coming Soon).

Runtime Insights on Profiler tab with live frame metrics, CPU/GPU bottleneck classification visible.

When to use it

• Runtime performance investigation during PIE / packaged builds
• Trace file analysis: opening .utrace files captured during play
• CI/CD performance regression tracking: automated benchmark + JSON export
• Bottleneck classification: “is this CPU-bound or GPU-bound?”

What it isn’t for

• Don’t expect asset-side analysis. Runtime profiling only.
• Don’t expect this to replace Unreal Insights. Complementary; UE Insights handles deep traces.

Opening the tool

1. Open the EQLabs Hub and search for Runtime Insights (or browse to Tech → Performance).
2. Click the tool card.
3. Use the tab bar: Profiler / Trace Analyzer / Benchmarks / AI Analysis.

Tab 1: Profiler

Live frame monitoring with stat unit parsing + bottleneck classification (CPU-bound / GPU-bound / Sync-limited).

Tab 2: Trace Analyzer

Parse .utrace files. Per-thread event breakdown, hot-function detection.

Tab 3: Benchmarks

Automated benchmark runner:

• Spline-based camera path
• Capture frame-time per sample point
• CSV/JSON export
• Suitable for CI/CD perf regression detection

Tab 4: AI Analysis

Coming Soon — LLM-powered trace interpretation. Feed a trace, get plain-English diagnosis.

Benchmark tab with spline path defined, frame-time graph visible after a run.

Workflow

Live profile

1. Switch to Profiler.
2. Start PIE.
3. Watch frame metrics + bottleneck classification.

Trace analysis

1. Capture a trace (UE’s Insights or via console trace.start).
2. Switch to Trace Analyzer.
3. Open the .utrace file.
4. Walk per-thread events.

CI benchmark

1. Set up a spline path through the level.
2. Switch to Benchmarks.
3. Run — captures frame-times along the path.
4. Export JSON; plug into your CI’s regression-detection pipeline.

Output

• Live in-panel metrics. Profiler tab during PIE.
• Trace analysis. Read-only display from .utrace.
• Benchmark CSV / JSON. For CI/CD pipeline ingestion.

Tips and gotchas

• Profiler is a stat unit consumer. Same data UE shows in console, presented with bottleneck classification.
• Benchmark is reproducible. Same spline path = same benchmark, useful for regression detection.
• AI Analysis is Coming Soon. V1 leaves the tab as a placeholder with a “coming soon” banner.

Related tools

• Rendering Systems Auditor — pre-runtime rendering audit
• Scene Optimization Advisor — actor/component-level optimization