GPUScore vs MLX Benchmarks

GPUScore: The Expedition is graphics performance benchmark for mobile devices. It is part of the GPUScore family of benchmarks and is the latest and most modern graphics benchmark for mobile devices, offering the most demanding workload for that device range. The Expedition measures GPU performance in a 3D mobile game scenario. The content could represent an action or real-time strategy game. The Expedition runs precisely the same content on every run, which guarantees maximally comparable results. The Expedition utilizes state-of-the-art rendering effects, like those seen in the latest mobile games. These include Volumetric lighting, Ground Truth Ambient Occlusion (GTAO) and Temporal Anti-Aliasing (TAA). The application contains several different modes. The Official and Official Native and benchmarks use fixed settings and thus always produce comparable results. The Official uses a fixed rendering resolution of 2560x1440 for rendering, while Native uses the displays resolution. The Custom benchmark can be used to run customized benchmarks. Additionally, the application contains an Experience mode where the content can be previewed in a smooth game-like way. GPUScore: The Expedition submits results from Official benchmarks to Powerboard 4.0 online platform, where users can compare their results. To run Official benchmarks one must have an active internet connection.

MLX Benchmark helps you understand how large language models perform on your hardware. Select a model and quantization, set context length and iterations, and run repeatable benchmarks that capture load time, tokenization time, time to first token, throughput, memory usage, and jitter. View detailed metrics grids, visualize comparisons in charts, and manage downloaded model repos directly in the app. Built for developers and AI enthusiasts who want clear, on-device performance insights without leaving their device. Features: - Model selection with quantization details - Configurable prompt, context length, and iterations - Metrics for load, tokenization, first-token latency, throughput, jitter, and memory - Charts to compare runs and devices - Downloaded model management (list, delete, refresh) - Optional Supabase upload for aggregating benchmark results