Primate Labs has just launched Geekbench AI 1.0, a cutting-edge benchmarking tool for measuring AI performance across different platforms.
Building on the earlier preview versions of Geekbench ML and now ready for general use, Geekbench AI 1.0 provides a comprehensive set of tests tailored to the rapidly evolving landscape of machine learning, deep learning, and AI-centric workloads.
Software developers can use the benchmark to ensure consistent performance across different platforms, while hardware engineers can use it to measure architectural improvements and device performance.
Measuring AI performance is a complex challenge due to the different approaches and hardware used in different devices. Primate Labs has addressed this complexity by providing three different performance scores in Geekbench AI 1.0. These scores take into account the different ways AI hardware is developed and how developers use that hardware.
Geekbench AI 1.0 is available for download for Windows, macOS, Linux, as well as through the Google Play Store and Apple App Store and is accessible to a wide audience.
To accommodate the diversity of AI hardware, Geekbench AI 1.0 offers three performance scores. This approach is similar to the way benchmarks evaluate the single-core and multi-core performance of CPUs.
In addition to speed, accuracy is a critical factor in AI performance. Geekbench AI 1.0 includes a new test-based accuracy measure that underscores the importance of not only running workloads quickly but also producing reliable results.
Primate Labs has ensured that Geekbench AI 1.0 is based on best industry practices. Detailed technical descriptions are available in the Inference Workloads document.
You can download Geekbench AI 1.0 for your device from this link.
