GeekBench version 6.1 released, single-core score increased by 5%, multi-core score increased by 10%

According to news on June 9, the benchmarking platform GeekBench today released version 6.1 update. According to officials, this version of the update will more accurately reflect the multi-core performance of the device, while improving the single-core score and multi-core score. GeekBench 6.1 introduces Clang 16 factors across all platforms to improve the accuracy of performance measurements.

According to the editor’s understanding, the update content of GeekBench 6.1 includes the following aspects. First, this version incorporates Clang 16 parameters, which will help more accurately evaluate the multi-core performance of the device. Secondly, in order to reduce the difference in results caused by thermal underclocking of new smartphones such as the Galaxy S23 series, GeekBench 6.1 Extends the workload gap (i.e., the pause time between workloads) from two seconds to five seconds. In this way, the test results will be more stable and reliable.

Also, GeekBench 6.1 also adds support for Scalable Vector Extensions (SVE) instructions and AVX512-FP16 instructions, which will provide better performance in multiple image processing and machine learning tasks. For systems without floating point 16-bit (FP16) instructions, GeekBench 6.1 uses a fixed-point algorithm to better consider multiple image processing functions.

It is particularly worth mentioning that for 12-core and 16-core high-end desktop processors, such as AMD Ryzens, AMD Threadrippers and Intel Xeons, GeekBench 6.1 further improves the performance of these processors by enhancing multi-core performance for background blur and horizon detection workloads.

The above is the detailed content of GeekBench version 6.1 released, single-core score increased by 5%, multi-core score increased by 10%. For more information, please follow other related articles on the PHP Chinese website!