The latest Intel CPUs might be impressive, but they aren’t as impressive as some benchmarks suggest. A new tool from Intel is tampering with Geekbench 6 results, and giving the company’s latest CPUs a boosted score in some tests.
Primate Labs, the company behind the popular open-source CPU and GPU benchmarking software Geekbench, made a blog post today explaining that Intel’s new Binary Optimization Tool is creating false scores for Intel CPUs. The tool is designed to modify instruction sequences to improve performance, which would normally be helpful, but benchmarks are only useful if they measure the same exact workload on each attempt. Otherwise, the final scores aren’t comparable between different devices.
When Geekbench 6 is run through the Binary Optimization Tool, some workload scores increase by up to 40%, and overall scores increase by up to 8%. Those scores aren’t comparable to other devices running Geekbench, and they aren’t a true reflection of computing performance. The Intel Binary Optimization Tool only supports a few applications, and it only works on certain Core Ultra Series 3 processors (“Panther Lake”) and Core Ultra 200 Plus chips (“Arrow Lake Refresh”).
The blog post explained, “The techniques used are not publicly documented, and it is unclear how widely applicable these techniques are across different applications. The tool only supports a short list of applications, and Geekbench 6 is one of the few supported applications.”
Related
3 ways the Windows Task Manager is lying to you
Task Manager looks authoritative, but several of its numbers are misleading. Here are the ways Windows’ favorite diagnostic tool gets things wrong.
Unfortunately, this means that some Geekbench scores for those Core Ultra CPUs aren’t accurate, and Geekbench currently has “no way to detect if a Geekbench 6 result was run with or without the Binary Optimization Tool.” As a result, all benchmark results from CPUs that support the tool will have a message explaining that the result might be invalid. “It’s a fake!”
Hardware manufacturers finding ways to cheat on benchmarks is nothing new. The practice was especially common on smartphones, with Samsung, HTG, Sony, LG, and OnePlus all trying it at one point or another—usually by temporarily overclocking the device when a benchmarking tool was detected. Intel was using a compiler hack to artificially boost benchmark results for its Xeon processors for several years.
If you’re trying to compare CPUs, and there’s at least one newer Intel CPU in the mix, you’ll want to check that the test was completed without Intel’s tampering. Geekbench 6.3 is the only benchmarking tool currently supported, but Intel could add others to the list.
Source: Geekbench Blog
