Benchmarks Of The Various Kernel Flavors Of Clear Linux
Last month Clear Linux rolled out new kernel options as bundles for those wanting to run a mainline/vanilla kernel build on this Intel open-source distribution without their extra patches applied as well as other options. Here are some benchmarks of those different kernel flavors available to Clear Linux users.
At the end of May I carried out some Clear Linux benchmarks of the different kernel bundles. The kernel results didn't end up being all that interesting and since then they switched over to Linux 5.1, but here's that data for those interested.
For those wondering what extra patches Clear Linux is currently carrying against their default kernel build, that list of work can be found via this GitHub repository. These patches include power management improvements, EXT4 tweaks, changes to the initialization/start-up to help with boot time, adding of WireGuard to their kernel, security patches, and other fixes/tweaks. Ultimately some of the work ends back into the upstream Linux kernel where relevant but some of the tweaks do not where it may not be appropriate or could risk regressing some users.
I ran benchmarks on the default/native kernel (5.0.18 at the time), kernel-pk that is using its iot-lts2018 kernel, the kernel-mainline-vanilla, kernel-native-current, and kernel-lts options.
In some cases, the older Linux 4.19 kernel was performing ahead of the newer 5.0~5.1 kernel options.
Granted in other cases the 2018 LTS kernel was the outlier on the slow end.
But overall not a whole lot of performance differences across dozens of tests.
Those wanting to look at all of the data in full can do so via this OpenBenchmarking.org result file.
At the end of May I carried out some Clear Linux benchmarks of the different kernel bundles. The kernel results didn't end up being all that interesting and since then they switched over to Linux 5.1, but here's that data for those interested.
For those wondering what extra patches Clear Linux is currently carrying against their default kernel build, that list of work can be found via this GitHub repository. These patches include power management improvements, EXT4 tweaks, changes to the initialization/start-up to help with boot time, adding of WireGuard to their kernel, security patches, and other fixes/tweaks. Ultimately some of the work ends back into the upstream Linux kernel where relevant but some of the tweaks do not where it may not be appropriate or could risk regressing some users.
I ran benchmarks on the default/native kernel (5.0.18 at the time), kernel-pk that is using its iot-lts2018 kernel, the kernel-mainline-vanilla, kernel-native-current, and kernel-lts options.
In some cases, the older Linux 4.19 kernel was performing ahead of the newer 5.0~5.1 kernel options.
Granted in other cases the 2018 LTS kernel was the outlier on the slow end.
But overall not a whole lot of performance differences across dozens of tests.
Those wanting to look at all of the data in full can do so via this OpenBenchmarking.org result file.
4 Comments