Geekbench, a widely recognized benchmark for comparing CPU and GPU performance, has been updated to version 7. The new release aims to better represent the demands of modern applications by introducing new workload tests and a redesigned multi-core benchmark.
The updated benchmarks operate on larger data sets and include new media workloads such as encoding screen-sharing video with the AV1 codec, compressing music and speech with the Opus audio codec, and decoding audio and video while using the Whisper speech recognition model to generate live captions.
The multi-core tests have been redesigned to better reflect how real applications behave. Instead of forcing a series of exclusively multi-threaded tasks, which can produce a distorted view of performance, the Geekbench 7 tests include only tests in the multi-threaded benchmark if they are highly multi-threaded in real life.
The GPU benchmarks have also been updated to focus on the content creation and machine learning uses of GPUs, which dominate GPU compute applications. These tests include tasks such as tracking faces in video, upscaling images using machine learning algorithms, or blurring the background of a video chat.
It is worth noting that scores are not comparable between versions. Early tests on an M5 MacBook Air yielded a single-core CPU score of 3,608 and a multi-core score of 17,470 using Geekbench 7, compared to 4,164 and 16,995 using Geekbench 6.6.
Primate Labs has released Geekbench 7 for Mac, and an iPhone and iPad version will be available in the App Store soon. Users can download the Mac version from the Primate Labs website.
