AMD

Improve performance and idle power consumption

Bug #1941893 reported by Mario Limonciello
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
AMD
Fix Released
Undecided
Alex Hung
linux-oem-5.14 (Ubuntu)
Invalid
Undecided
Unassigned
Focal
Fix Released
Undecided
Unassigned

Bug Description

[Impact]

  This can improve AMD performance for all Zen or newer CPU which support C3 shares cache.

[Fix]

  Its not necessary to flush the caches in software before entering C3.

  Empirical measurements show:
   40-45% improvement in speedometer score.
   ~15% improvement in idle power.

  The patch is in linux-next (tag next-20211007) and can be found @ https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/arch/x86/kernel/acpi/cstate.c?id=a8fb40966f19ff81520d9ccf8f7e2b95201368b8

[Test]

  This is requested by AMD.

[Where problems could occur]

  Low risk. This only works on AMD Zen CPUs that supports this features. AMD tested it and provided estimated measurement results.

==== Original descriptions ====

A patch queued for 5.15 to change the C3 entry will improve AMD performance.
https://lore.kernel.org/patchwork/patch/1478776/

Empirical measurements show:
 40-45% improvement in speedometer score.
 ~15% improvement in idle power.

Can you please include in 5.13-oem and 5.14-oem kernels?

Alex Hung (alexhung)
Changed in amd:
assignee: nobody → Alex Hung (alexhung)
Revision history for this message
Alex Hung (alexhung) wrote :

I tried this on a Ryzen 5 2500U but couldn't see any differences. That was the only AMD system I can have physical access at the moment.

Which CPU or platform is supported by this patch?

Revision history for this message
Mario Limonciello (superm1) wrote :

What test did you use?
It should be pronounced/notable on Ryzen 5000, particularly using speedometer for the test.

Revision history for this message
Mario Limonciello (superm1) wrote :

Also can you confirm the platform that you're using has ACPI C3 exported? If it doesn't you'll need a platform with C3.

Revision history for this message
Alex Hung (alexhung) wrote :

I only measured idle power.

Thanks. I will try to find Ryzen 5000 for testing.

Revision history for this message
Alex Hung (alexhung) wrote :

Status updates:

This patch is not in mainline as 5.15rc2.

The maintainer listed his reasons here: https://lore.kernel<email address hidden>/

Revision history for this message
Mario Limonciello (superm1) wrote :

It was resubmitted here with a guard to only run on Zen+ CPUs.
https://<email address hidden>/

Revision history for this message
Mario Limonciello (superm1) wrote :
Alex Hung (alexhung)
information type: Proprietary → Public
Alex Hung (alexhung)
description: updated
Revision history for this message
Alex Hung (alexhung) wrote :
Timo Aaltonen (tjaalton)
Changed in linux-oem-5.14 (Ubuntu):
status: New → Invalid
Changed in linux-oem-5.14 (Ubuntu Focal):
status: New → Fix Committed
Revision history for this message
Ubuntu Kernel Bot (ubuntu-kernel-bot) wrote :

This bug is awaiting verification that the linux-oem-5.14/5.14.0-1006.6 kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-focal' to 'verification-done-focal'. If the problem still exists, change the tag 'verification-needed-focal' to 'verification-failed-focal'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

tags: added: verification-needed-focal
Alex Hung (alexhung)
tags: added: verification-done-focal
removed: verification-needed-focal
Changed in amd:
status: New → Fix Committed
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package linux-oem-5.14 - 5.14.0-1007.7

---------------
linux-oem-5.14 (5.14.0-1007.7) focal; urgency=medium

  * focal/linux-oem-5.14: 5.14.0-1007.7 -proposed tracker (LP: #1949349)

  * Intel I225-IT ethernet controller: igc: probe of 0000:02:00.0 failed with
    error -1 (LP: #1945576)
    - igc: Remove _I_PHY_ID checking
    - igc: Remove phy->type checking

  * Fix Screen freeze after resume from suspend with iGPU [1002:6987]
    (LP: #1949050)
    - drm/amdgpu: reenable BACO support for 699F:C7 polaris12 SKU
    - drm/amdgpu: add missing cleanups for Polaris12 UVD/VCE on suspend
    - drm/amdgpu: Fix crash on device remove/driver unload

  * Fail to detect audio output from external monitor (LP: #1948767)
    - ALSA: hda: intel: Allow repeatedly probing on codec configuration errors

  * Fix System hangs on black screen when reboot (LP: #1949321)
    - SAUCE: drm/i915: Don't request GMBUS to generate irqs when called while irqs
      are off
    - SAUCE: drm/i915/hdmi: Turn DP++ TMDS output buffers back on in
      encoder->shutdown()

 -- Timo Aaltonen <email address hidden> Mon, 01 Nov 2021 12:38:15 +0200

Changed in linux-oem-5.14 (Ubuntu Focal):
status: Fix Committed → Fix Released
Alex Hung (alexhung)
Changed in amd:
status: Fix Committed → Fix Released
Alex Hung (alexhung)
Changed in amd:
status: Fix Released → In Progress
Alex Hung (alexhung)
Changed in amd:
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.