Agenda

The Delta Force development team tackles the challenge of cross-platform performance optimization for highly complex, large-scale gameplay—particularly achieving 32v32 real-time battles on mobile devices.
Their in-house tool, Metaperf, overcomes the limitations of traditional instrumentation by combining hybrid sampling and AI-driven attribution to deliver high-frequency, low-overhead diagnostics across PC, mobile, and console platforms.
The system automatically classifies stuttering patterns through frame-by-frame analysis and intelligent correlation, precisely identifying bottlenecks in high-concurrency and large-scene environments.
Validated over 30 months of production, Metaperf has processed 15,000+ game sessions and identified 3,000+ performance defects.
By empowering QA and engineering teams with expert-level analytical capabilities, it establishes an industrialized, intelligent workflow for performance optimization—setting a new benchmark for large-scale, cross-platform game development.

Takeaway

Attendees gain insight into building a unified cross-platform performance profiling pipeline integrating data from PC, mobile, and console.
Through Delta Force case studies, they learn low-overhead hybrid sampling, AI-driven stutter attribution, and data processing methods that transform manual profiling into intelligent, industrial-scale performance optimization.

Intended Audience

This session is intended for performance engineers, technical directors, and QA leads who work on large-scale PC and console titles. Attendees should have basic knowledge of profiling and CPU/GPU performance tools. Engine developers and tool designers seeking to build scalable, data-driven performance analysis pipelines will gain the most from this talk.