Now, with the CPUFreq fix coming this week on Linux Git, the Linux 5.11 kernel in its near-final state is in good shape for AMD Zen 2/3 hardware from Ryzen laptops and desktops via EPYC servers. The Linux 5.11 development kernel has been regressed for most of the past two months, but now that the frequency invariance regression has been resolved, not only has the regression ended, but it is generally performing much better compared to previous versions of kernel.
Continuing in the benchmark matrix since December, when I first encountered the regression in Zen 2 / Zen 3 when using the default Schedutil governor, here are my final benchmarks looking at the performance of AMD Linux 5.11, now that the fix through the CPUFreq change mentioned earlier this week and stable Linux 5.11 is expected to be released on Sunday.
On the desktop side, I ran some new Linux kernel benchmarks with the Ryzen 9 5900X desktop at stock speeds, ASUS ROG CROSSHAIR VIII HERO motherboard, 2 x 8GB DDR4-3600 memory, 1 TB Samsung 980 PRO NVMe SSD and Radeon graphics RX 5600 XT. When installing Ubuntu 20.10, I tested Linux 5.9.16, Linux 5.10.12 and Linux 5.11 Git as the three most recent stable kernel series at the time. The Ubuntu Mainline Kernel PPA was used to obtain these vanilla kernel builds in an easily reproducible way. The default CPUFreq governor in recent AMD hardware kernels is Schedutil.
On the pages that follow the prominent individual benchmarks, we will jump straight to the geometric average of the 63 benchmarks executed in this round.
Linux 5.9 / 5.10 had almost the same overall performance, while with this effectively final state of Linux 5.11, performance increased 5.8% in 63 different test cases. Not bad for performance to have been lower in the last two months due to the regression, but luckily now everything is in good shape using the default Schedutil governor (as shown in previous tests, ondemand governor is also better with Linux 5.11).
The Linux 5.11 kernel came out first in this Ryen 9 5900X box with 60% of the benchmark. Where 5.9 / 5.10 won, it was often by narrow margins.
Or lastly, the Linux 5.11 Git kernel came in last for just 5 of the 63 benchmarks. So things are looking pretty good at this point, with Linux 5.11 offering a better performance experience than previous kernels …
I also ran new benchmarks on the AMD EPYC 7F72 2P server with Supermicro H11Dsi-NT v2.0 motherboard, 16 x 8GB DDR4-3200 memory and 1 TB Western Digital BLACK SN850 NVMe SSD. Ubuntu 20.10 was running on this AMD EPYC Zen 2 server while the Linux kernels 5.9.16, 5.10.13 and 5.11 Git were used as the most recent when starting these tests.
42 tests were performed on this EPYC server in the three most recent kernel series focusing more on server / workstation loads. With this set of benchmarks, Linux 5.11 based on the geometric average of all benchmarks came out 6.0% ahead … Very similar to 5.8% with the 5900X even with the difference in tests. However, it was the leader in 76% of the benchmarks compared to 5.10 / 5.9 and was in last place for only 11% of the tests.
Let’s take a closer look at the performance of the Ryzen 9 5900X and EPYC 7F72 2P with Linux 5.11, which should be the version of Ubuntu 21.04 this spring.