Controlled evaluation of layered interventions on CS2 P99 frame time

Determine whether the layered interventions B1–B4—CS2/display configuration, disabled capture paths, background cleanup, and system tuning—lower session-level P99 frame time relative to the preceding intervention layers and the restored baseline B0 on the specified Windows gaming laptop under a frozen workload.

Background

The paper presents Spike-Killer as a same-day, single-machine experience study rather than a controlled performance-effect evaluation. The completed pilot collected system-state changes, microbenchmarks, telemetry, and integrity-checked Windows Performance Recorder traces, but it did not obtain valid per-frame intervals, a matched baseline, or a randomized comparison across intervention conditions.

RQ1–RQ3 received pilot evidence, whereas the prospective performance question, RQ4, was explicitly left unresolved. Answering it requires a frozen workload and a repeated controlled comparison of B0 through B4, with valid frame-time measurements suitable for estimating session-level P99 behavior and conditional intervention effects.