[ad_1]
With a pending fix, the performance of Linux 5.11 AMD Zen 2 / Zen 3 looks very good when it comes to out-of-the-box performance when using Schedutil as it increasingly becomes the update regulator. scale to default CPU frequency on more distros / default cores. With the previously mentioned Linux 5.11 regression since the introduction of AMD CPU frequency invariance support, Schedutil’s performance from small Ryzen systems to large EPYC hardware looks pretty good. But what advantage remains over the optimal performance of scaling the processor frequency with the “performance” governor? Here is a look at those benchmarks on Ryzen and EPYC for Schedutil vs Performance on a patched Linux 5.11 kernel.
As a follow-up to last week’s tests on the patched Linux 5.11 kernel leading to better Schedutil behavior with AMD Zen 2 / Zen 3 hardware, this article provides benchmarks on how the patched Governor Schedutil compared to the switching. on the governor “performance” on the same kernel. Most Linux distributions still do not use the performance controller by default, but either using “ondemand” or newer kernels often use “schedutil” for AMD hardware with the CPUFreq driver.
When testing on an AMD Ryzen 9 5950X, there is still room to further tune Governor Schedutil for better performance:
While ignoring a few key exceptions for some faster running tests where the performance throttle benefits are more pronounced compared to Schedutil taking some (short) time to increase clock rates, for most loads working, the performance regulator produces up to a few percent faster performance. That’s not that big of a deal considering that this patched Linux 5.11 kernel for testing is far better off than the major regressions noted earlier in cycle 5.11 and the patch often leading to performance beyond that of stable Linux 5.10. .
218 different test cases were run on this Ryzen 9 5950X system by comparing the two CPUFreq governors. The side-by-side chart above simply shows the workloads with a measurable and statistically significant difference.
Taking the geometric mean of the 218 results, the performance governor was just over 1% faster than the default governor in Schedutil on Linux 5.11.
But now let’s look at the Schedutil vs Performance difference on AMD EPYC while also monitoring CPU power consumption as exposed by the AMD_Energy driver.
[ad_2]
Source link