Neural processing units (NPUs) designed to accelerate artificial intelligence algorithms were originally found in mobile platforms and Apple’s PC processors, but are now becoming commonplace in Intel and AMD chips. This means there’s a need to measure their performance, a task that Geekbench AI aims to address.
Primate Labs’ main Geekbench app is designed to test CPU and GPU performance, but in recent years the developer has also been experimenting with an alternative, Geekbench ML. With the launch of Microsoft’s Copilot+ initiative and the start of the race for the fastest neural processing units between Intel, AMD, Qualcomm, and Apple, Primate Labs has released version 1.0 of the new benchmark and renamed it Geekbench AI.
Features and Supported Platforms
“Geekbench AI introduces its own set of workload tests that run on single-precision, half-precision, and quantized data, covering the various types used by developers in terms of both precision and purpose in AI systems,” Primate Labs explained. Geekbench AI supports several AI frameworks: OpenVINO on Windows and Linux, ONNX on Windows, QNN for Qualcomm Snapdragon processors on PC, Apple CoreML on macOS and iOS, and a number of Android frameworks offered by different chip manufacturers. On Windows PCs, the test supports NPUs from Intel and Qualcomm, with AMD support coming later.
Geekbench AI is available on Windows, macOS, Linux, iOS/iPadOS, and Android. The basic version is free, but there is also a Pro license for $99, which includes command-line tools, the ability to run the test without uploading results to the Geekbench database, and some other features.
The developers intend to update the benchmark as needed, adding support for new hardware, frameworks, and workloads. We’ll keep you updated on any significant changes or additions to Geekbench AI as they become available.
As the landscape of AI acceleration continues to evolve rapidly, tools like Geekbench AI will play a crucial role in helping users and developers understand the capabilities of different NPUs across various devices and platforms, notes NIX Solutions. This benchmark not only provides valuable insights into the performance of current NPU implementations but also sets a standard for measuring future advancements in AI processing hardware.
The introduction of Geekbench AI reflects the growing importance of AI acceleration in consumer devices, from smartphones to desktop computers. As more applications begin to leverage AI capabilities, having a reliable and comprehensive benchmark for NPU performance will become increasingly valuable for both consumers and manufacturers.
By offering support for multiple platforms and AI frameworks, Geekbench AI aims to provide a unified testing methodology that can be applied across a wide range of devices and use cases. This approach should help foster a better understanding of NPU performance across different ecosystems and enable more informed decisions when it comes to hardware selection for AI-intensive tasks.