Unreal Engine Research & Evidence-Based Practice 4 — Questions and Answers
Question 1: Which metric in 'stat unit' most directly indicates that the game thread is the performance bottleneck rather than the GPU?
- Draw call count exceeds GPU budget
- Frame time is high and Game thread time equals or exceeds Frame time (Correct answer)
- GPU time is greater than CPU time
- Render thread time is zero
Correct answer: Frame time is high and Game thread time equals or exceeds Frame time
When Game thread time equals or exceeds overall Frame time, the CPU game thread is the limiting factor, not the GPU or render thread.
Question 2: A researcher wants to compare memory usage of the Sparse Class Data optimization against default UPROPERTY storage. Which tool shows class instance memory footprints?
- Unreal Insights Memory track with LLM tags (Correct answer)
- P4V depot browser
- Content Browser size column
- Project Settings memory budget sliders
Correct answer: Unreal Insights Memory track with LLM tags
The Low-Level Memory (LLM) tracking in Unreal Insights tags allocations by subsystem, enabling direct comparison of per-class memory before and after Sparse Class Data.
Question 3: When evaluating community-proposed Blueprints optimizations on the Unreal forums, which criterion most determines whether the advice is evidence-based?
- Number of upvotes on the post
- Whether the post includes before/after profiler screenshots with methodology described (Correct answer)
- Whether the author has a high post count
- Whether the suggestion involves C++ rather than Blueprint
Correct answer: Whether the post includes before/after profiler screenshots with methodology described
Evidence-based advice includes measurable before/after data and a described methodology; upvotes and post count are social signals, not empirical evidence.
Question 4: Which Unreal Engine console command generates a full memory report saved to disk, suitable for longitudinal research tracking?
- stat memory
- memreport -full (Correct answer)
- obj list -class=Texture2D
- showmemory
Correct answer: memreport -full
`memreport -full` writes a comprehensive memory snapshot to the Saved/Profiling directory, capturing all allocations for offline analysis and trend comparison.
Question 5: A researcher discovers conflicting advice: one source says to disable World Partition for linear levels, another says to always use it. What is the evidence-based approach?
- Follow the advice from the source with more social media followers
- Design a controlled test with both configurations on a representative level and measure streaming and memory metrics (Correct answer)
- Default to disabling World Partition for all projects
- Ask on Discord and accept the first answer
Correct answer: Design a controlled test with both configurations on a representative level and measure streaming and memory metrics
Designing a controlled test with measurable metrics on a representative level provides project-specific empirical evidence that resolves the conflicting general claims.
Question 6: How does Unreal Engine's Automation System (UAutomationTestBase) support evidence-based quality assurance?
- It generates marketing screenshots automatically
- It provides a framework for writing repeatable automated tests with pass/fail assertions (Correct answer)
- It compresses pak files for distribution
- It manages source control branching
Correct answer: It provides a framework for writing repeatable automated tests with pass/fail assertions
UAutomationTestBase lets teams write repeatable automated tests with defined pass/fail criteria, creating an evidence trail for code correctness over time.
Question 7: Which approach correctly isolates Niagara particle system CPU cost from other game thread work during a profiling session?
- Disable all Niagara emitters and compare stat unit Game time to baseline (Correct answer)
- Increase particle count until FPS drops
- Enable Nanite on particle meshes
- Reduce draw distance to zero
Correct answer: Disable all Niagara emitters and compare stat unit Game time to baseline
Disabling all Niagara emitters and comparing the resulting Game thread time against the baseline isolates exactly how much CPU cost Niagara contributes.
Which metric in 'stat unit' most directly indicates that the game thread is the performance bottleneck rather than the GPU?