Benchmark Dataset: Sustained performance and thermal efficiency of Minecraft Java workloads on a fanless ARM64 and an actively cooled x86-64 laptop
Description
This dataset contains the experimental data used in a comparative study of gaming performance, power consumption, thermal behavior, and performance-per-watt efficiency between an ARM64-based Apple MacBook Air M4 and an x86-64-based Lenovo ThinkPad E14 Gen 5 with an Intel Core i5-1335U processor. The experiments use Minecraft Java Edition 1.21.1 as a sustained and reproducible gaming workload, with identical core configurations across both platforms. The workload includes Sodium, Fabric API, and Baritone, with Baritone used to provide automated and consistent ground-sprint traversal. Each platform was evaluated through five independent 20-minute runs while connected to AC power. The dataset includes measurements of frame rate, power consumption, and temperature collected at approximately one-second intervals. FPS data for macOS were recorded using a custom Fabric client mod, while Windows FPS measurements were obtained using HWiNFO64 and PresentMon. Power and thermal measurements were collected using platform-specific monitoring tools, including powermetrics and TG Pro on macOS and HWiNFO64 on Windows. The dataset is provided to support reproducibility and further analysis of the reported experimental results. It includes the raw benchmark workbooks for both platforms, containing protocol information, individual experimental runs, summary statistics, and time-series data. The dataset also supports the generation and verification of the figures and statistical analyses presented in the associated research paper. The reported study compares sustained performance and thermal/power efficiency across two consumer laptop platforms with different CPU architectures and cooling approaches. The results should be interpreted within the stated experimental limitations, including the use of a single physical unit per platform, five repeated runs per condition, different measurement tools between operating systems, and differences in the definition of the 1% Low FPS metric.
Files
Institutions
- Binus UniversityJakarta, Jakarta