Linux 6.16's New "X86_NATIVE_CPU" Option Enhances I/O & Some Graphics/Gaming Workloads
In over 100 benchmarks tested on both kernel builds, overall there was little difference out of the X86_NATIVE_CPU kernel build on Linux 6.16 with GCC 14.2 on this AMD Strix Halo laptop (HP ZBook Ultra G1a).
But there were some exceptions where some differences were observed:
In some of the kernel micro-benchmarks there were improvements, especially for some of the synthetic I/O benchmarks. In some of the lightweight graphics/gaming tests were also 3~5% improvements when running on the X86_NATIVE_CPU kernel build. Those lightweight gaming tests were one of the main real-world areas observing a benefit from CONFIG_X86_NATIVE_CPU besides the I/O improvements.
Importantly the CONFIG_X86_NATIVE_CPU kernel build hadn't resulted in any measurable difference to the AMD Zen 5 CPU power consumption.
That's the initial data on Linux 6.16 with CONFIG_X86_NATIVE_CPU. I will be exploring the "-march=native" kernel tuning on some other hardware platforms to look for any other performance differences as well as if GCC 15 or LLVM/Clang is providing any better improvements with this new Linux kernel build option.
If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.