Running The AMD "ABBA" Ryzen 3000 Boost Fix Under Linux With 140 Tests

Written by Michael Larabel in Processors on 19 September 2019 at 09:00 AM EDT.

When looking at the geometric mean of the 140 tests carried out, there was just under a 1% difference overall. All the results plus peak CPU frequency averages for each test below.

But of all the tests, in several of them there was a noticeable performance improvement under Linux with the Ryzen 9 3900X:

In only a few cases were there performance regressions, some of which were for traditionally noisy applications.

Of the nine and a half hours it took roughly each time to run all of these tests, lastly is the box plot of the peak frequencies recorded throughout that entire process. The average of the peak CPU frequency was 1.7% higher with the latest update using AGESA 1.0.0.3 ABBA. The peak CPU frequency recorded on the Ryzen 9 3900X also rose from 4597MHz to 4626MHz. Tests on other motherboards and Zen 2 CPUs are still ongoing.

Completely unrelated to the boost frequencies but reminded of the situation when checking the latest AGESA, there still is the potential for more performance out of a future firmware / CPU microcode update. Even with this latest revision, AMD is still defaulting to always-on RSB filling as part of the Spectre Variant Two mitigations for Ryzen 3000 series processors. Technically this only needs to be conditional filling of the return stack buffer. Intel is carrying out conditional RSB filling and EPYC 7002 series processors with an early firmware/microcode update switched from always-on to conditional RSB filling, matching the behavior of Intel in this regard. The Ryzen 3000 series processors meanwhile with the current mode is still always-on. Should Ryzen 3000 series switch to conditional, this should provide a small but measurable improvement in affected workloads. I raised this with AMD back at launch and they were evaluating but so far I haven't heard if/when they will make that change.

If you enjoy my daily and relentless Linux benchmarking and contributions to open-source benchmarking in general, consider showing your support by joining Phoronix Premium or making a PayPal tip.

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.


Related Articles
About The Author
Michael Larabel

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 20,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter, LinkedIn, or contacted via MichaelLarabel.com.