Primate Labs, the creators of Geekbench, have launched the Geekbench AI app, a cross-platform benchmarking tool designed to assess the performance of AI workloads through real-world machine learning tasks.
Previously known as Geekbench ML during its preview phase, this innovative tool evaluates the capabilities of CPUs, GPUs, and NPUs in machine learning applications, assigning scores based on accuracy and speed.
Each test run on Geekbench AI generates three scores: single precision, half-precision, and quantized. Additionally, it measures accuracy to determine how effectively models complete tasks and provides insights into relative efficiency over time. Each workload is tested for at least five iterations to ensure optimal results, allowing devices to reach maximum performance.
Geekbench AI is compatible with various AI frameworks, including CoreML for macOS and iOS, OpenVINO for Windows and Linux, QNN for Snapdragon-powered Arm PCs, and vendor-specific frameworks for Android devices. This versatility allows for a thorough evaluation of AI performance across multiple platforms, facilitating easier comparisons of results.
Furthermore, Geekbench AI delivers a detailed report after each benchmarking run, featuring model accuracy, inference time, and insights into hardware utilization. This information is invaluable for developers and researchers seeking to optimize their machine-learning models for enhanced performance.
As AI continues to infiltrate various industries, Geekbench AI is a valuable resource for companies and organizations looking to evaluate the capabilities of different devices before making investments tailored to their specific AI needs. It also aids in informed decision-making when considering different hardware configurations.
The introduction of Geekbench AI underscores the increasing significance of benchmarking in the rapidly evolving landscape of artificial intelligence.
