discrete card not powered down in Intel mode with legacy drivers

Bug #1912974 reported by Dan Robinson
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
nvidia-prime (Ubuntu)
New
Undecided
Unassigned

Bug Description

This is the same issue as originally reported in bug 1765363 but that report is old and has a lot of extra data on it so it was requested that a new bug be opened if we were still having the issue.

So... I am still having the issue, at least with the 390 series drivers (my laptop has a Quadro 5400M).

Switching to Intel mode via either the GUI in nvidia-settings or via prime-select successfully moves the rendering to the Intel card but does not power down the discrete card, as evidenced by either power consumption in powertop in excess of 20W, or by manually checking /proc/acpi/bbswitch (shows ON).

Manually echoing off into /proc/acpi/bbswitch does turn the card off as can be seen from the power consumption drop.

ProblemType: Bug
DistroRelease: Ubuntu 20.10
Package: nvidia-prime 0.8.15
ProcVersionSignature: Ubuntu 5.8.0-40.45-generic 5.8.18
Uname: Linux 5.8.0-40-generic x86_64
ApportVersion: 2.20.11-0ubuntu50.3
Architecture: amd64
CasperMD5CheckResult: skip
CurrentDesktop: ubuntu:GNOME
Date: Sun Jan 24 19:07:10 2021
Dependencies:

InstallationDate: Installed on 2021-01-24 (0 days ago)
InstallationMedia: Ubuntu 20.10 "Groovy Gorilla" - Release amd64 (20201022)
PackageArchitecture: all
ProcEnviron:
 TERM=xterm-256color
 PATH=(custom, no user)
 XDG_RUNTIME_DIR=<set>
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: nvidia-prime
UpgradeStatus: No upgrade log present (probably fresh install)

Revision history for this message
Dan Robinson (dlrobinson) wrote :
Revision history for this message
Dan Robinson (dlrobinson) wrote :

Well I played with this some more in Hirsute and I wonder if I may have jumped the gun.

If I give it a minute to idle the power usage drops down to about 11-12 watts, which is about 2W more than using bbswitch or disabling the card in BIOS.

Which suggests that power management is working and it's *probably* good enough since my battery is shot anyway; at least it's low enough to keep idle temps sane.

I'm thinking maybe we leave the bug open to see if the power management can be brought any closer, but if it can't I guess this could be closed too.

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.