optimization-performance_skill

Overview

This skill helps game developers optimize performance across platforms by providing concrete profiling workflows, CPU/GPU tuning patterns, and platform-specific targets. It focuses on measurable outcomes—stable frame rates, lower memory and draw call budgets, and predictable load times—for smoother gameplay experiences. The guidance covers engine-agnostic techniques and engine-specific tool recommendations.

How this skill works

The skill teaches a repeatable profiling workflow: measure, identify bottlenecks, analyze details, optimize one change at a time, verify across platforms, and repeat. It inspects CPU vs GPU time, frame time breakdowns, draw call counts, overdraw, shader complexity, memory allocations, and loading patterns to pinpoint hotspots. Recommendations include algorithmic fixes, cache- and allocation-aware code, batching/instancing, LODs, and platform constraints for mobile and VR.

When to use it

• When frame rates drop or stutter appears during gameplay
• Before shipping to ensure target platforms meet frame budgets
• When porting to mobile, VR, or lower-end hardware
• During feature development to prevent regressions
• To reduce load times, memory use, or thermal/battery impact

Best practices

• Profile first—never guess where the bottleneck is
• Optimize one subsystem at a time and measure before/after
• Design data-oriented systems and minimize cache misses
• Avoid allocations in hot paths; use pooling and pre-allocation
• Target platform budgets (draw calls, triangles, memory) and test edge cases

Example use cases

• Identify whether a scene is CPU- or GPU-bound and apply targeted fixes
• Reduce mobile draw calls with batching and texture atlases to hit 30–60 FPS
• Implement job systems and async loading to remove frame spikes
• Apply LODs, frustum/occlusion culling, and instancing for high-entity scenes
• Tune shaders and reduce texture samples to lower GPU time on consoles