Switch from the apt bindings to the apt_pkg bindings for the ubuntu-drivers tool.
This significantly improves performance (2,673s vs 13,751s , on my dated Intel Ivy Bridge i7-3770K CPU).
[Attach sbuild log]
[Performance data]
system_driver_packages() performance for a lot of modaliases ... [0.94 s] ok
system_driver_packages() performance for a lot of modaliases ... [0.96 s] ok
Old detection code:
:~$ time ubuntu-drivers list
nvidia-driver-510, (kernel modules provided by linux-modules-nvidia-510-generic-hwe-20.04)
nvidia-driver-470, (kernel modules provided by linux-modules-nvidia-470-generic-hwe-20.04)
nvidia-driver-470-server, (kernel modules provided by linux-modules-nvidia-470-server-generic-hwe-20.04)
real 0m13,751s
user 0m13,480s
sys 0m0,258s
:~$ time ubuntu-drivers list
nvidia-driver-510, (kernel modules provided by linux-modules-nvidia-510-generic-hwe-20.04)
nvidia-driver-470-server, (kernel modules provided by linux-modules-nvidia-470-server-generic-hwe-20.04)
nvidia-driver-470, (kernel modules provided by linux-modules-nvidia-470-generic-hwe-20.04)
real 0m13,785s
user 0m13,578s
sys 0m0,196s
New detection code:
:~$ time ubuntu-drivers list
nvidia-driver-470, (kernel modules provided by linux-modules-nvidia-470-generic-hwe-20.04)
nvidia-driver-470-server, (kernel modules provided by linux-modules-nvidia-470-server-generic-hwe-20.04)
nvidia-driver-510, (kernel modules provided by linux-modules-nvidia-510-generic-hwe-20.04)
real 0m2,673s
user 0m1,749s
sys 0m0,914s
:~$ time ubuntu-drivers list
nvidia-driver-470-server, (kernel modules provided by linux-modules-nvidia-470-server-generic-hwe-20.04)
nvidia-driver-510, (kernel modules provided by linux-modules-nvidia-510-generic-hwe-20.04)
nvidia-driver-470, (kernel modules provided by linux-modules-nvidia-470-generic-hwe-20.04)
real 0m2,645s
user 0m1,702s
sys 0m0,935s
[PR]
https://github.com/tseliot/ubuntu-drivers-common/pull/68
[PPA]
https://launchpad.net/~albertomilone/+archive/ubuntu/u-d-c-lp1963712
Hello Alberto! Thank you for filling in the FFe! Did you do a test-build of the change on some PPA that we could take a look at? And did you do a test install of the test packages so that you could provide some logs? The changes seem on one side simple, but touch a lot of the code, so I'm obviously a bit worried about any bumps. How noticeable is the performance improvement? Has this been measured? Just trying to assess the risk-to-merit ratio here ;)