Re: [PATCH v1 1/2] cpufreq: ACPI: Extend frequency tables to cover boost frequencies

From: Giovanni Gherdovich
Date: Fri Feb 05 2021 - 02:07:36 EST


On Thu, 2021-02-04 at 18:25 +0100, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
>
> A severe performance regression on AMD EPYC processors when using
> the schedutil scaling governor was discovered by Phoronix.com and
> attributed to the following commits:
>
> 41ea667227ba ("x86, sched: Calculate frequency invariance for
> AMD systems")
>
> 976df7e5730e ("x86, sched: Use midpoint of max_boost and max_P
> for frequency invariance on AMD EPYC")
>
> [snip]
>
> Fixes: 41ea667227ba ("x86, sched: Calculate frequency invariance for AMD systems")
> Fixes: 976df7e5730e ("x86, sched: Use midpoint of max_boost and max_P for frequency invariance on AMD EPYC")
> Fixes: db865272d9c4 ("cpufreq: Avoid configuring old governors as default with intel_pstate")
> Link: https://www.phoronix.com/scan.php?page=article&item=linux511-amd-schedutil&num=1
> Link: https://lore.kernel.org/linux-pm/20210203135321.12253-2-ggherdovich@xxxxxxx/
> Reported-by: Michael Larabel <Michael@xxxxxxxxxxxx>
> Diagnosed-by: Giovanni Gherdovich <ggherdovich@xxxxxxx>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
> ---
> drivers/cpufreq/acpi-cpufreq.c | 107 ++++++++++++++++++++++++++++++++++++-----
> 1 file changed, 95 insertions(+), 12 deletions(-)
>
> Index: linux-pm/drivers/cpufreq/acpi-cpufreq.c
> ===================================================================
> --- linux-pm.orig/drivers/cpufreq/acpi-cpufreq.c
> +++ linux-pm/drivers/cpufreq/acpi-cpufreq.c
> [...]

Tested-by: Giovanni Gherdovich <ggherdovich@xxxxxxx>
Reviewed-by: Giovanni Gherdovich <ggherdovich@xxxxxxx>

Note there is also the Tested-by: Michael, from the other thread
https://lore.kernel.org/lkml/5ea06dbe-255c-3d22-b9bd-6e627c5f94af@xxxxxxxxxxxx/

I tested this patch with the "NASA Parallel Benchmarks" from [link below], the
results confirms that the 5.10 performance is recovered:


Ratios of completion times, lower is better (5.10 is the baseline)

5.10 5.11-rc6 5.11-rc6-ggherdov 5.11-rc6-rafael
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Integer Sort 1.00 1.21 0.91 0.93
Embarrassingly Parallel 1.00 1.60 1.00 1.00
Discrete FFT 1.00 1.68 0.67 0.67


CPU : MODEL : 2x AMD EPYC 7742
FREQUENCY TABLE : P2: 1.50 GHz
P1: 2.00 GHz
P0: 2.25 GHz
MAX BOOST : 3.40 GHz

[link] https://www.nas.nasa.gov/publications/npb.html

Thanks,
Giovanni