As artificial intelligence is rapidly becoming widespread in the technology world, evaluating the performance of devices in this field becomes of great importance. In line with this need, the popular performance measurement tool Geekbench has introduced Geekbench AI, a new cross-platform test, to users. This tool measures how efficiently devices work under AI-heavy workloads by testing components such as CPU, GPU and NPU (neural processing unit). Geekbench AI was developed by Primate Labs and was first introduced as a preview version in 2021 under the name “Geekbench ML”. However, the name has been updated to Geekbench AI to better express that the functions offered by the software are artificial intelligence-oriented. This tool is designed to evaluate how hardware performs on different AI tasks and can evaluate both accuracy and speed. With Geekbench AI, users can see in detail how effective their devices are in artificial intelligence-related tasks. The performance metrics offered by Geekbench AI fall into three main categories: full precision, half precision and quantized results. These measurements are important to understand how accurately the device performs workloads. Additionally, Primate Labs states that these scores evaluate not only speed and performance, but also accuracy. These accuracy measurements are especially critical for understanding the quality of output of AI applications. This new test…
The post Geekbench AI, a performance testing tool specific to artificial intelligence workloads, has been published appeared first on Teknoblog.
Source link: https://www.teknoblog.com/geekbench-ai-cihaz-performans-testi/