You are not logged in.
I'll check on powertop when I can and let you guys know, I was getting about 11W with the script method before with my 4k touchscreen model with the touch screen enabled.
According to the reddit thread you've linked to in your previous post, the power management is implemented only for the "Turing or later" cards. This means that with this method, the dGPU is going to be always powered on for XPS 9570.
Offline
oddsocks wrote:I'll check on powertop when I can and let you guys know, I was getting about 11W with the script method before with my 4k touchscreen model with the touch screen enabled.
According to the reddit thread you've linked to in your previous post, the power management is implemented only for the "Turing or later" cards. This means that with this method, the dGPU is going to be always powered on for XPS 9570.
Yep you guys were right. My power consumption is reported at about 13 - 16W depending on usage, which is a little bit worse than with the other two methods.
For me power consumption was not a major priority, and is not really a deal breaker for me, so I didn't look into it a great deal.
I'll probably continue to use PRIME offloading method.
Obviously not for everyone.
Thanks for pointing that out, yaworski and LazyLucretia.
Offline
I've played around today, inspired by new development. In the end, I've switched to optimus-manager. The fact that I need to log out to switch the GPU is not ideal, but at least I'm sure that NVidia is powered off when I'm in Intel mode.
Offline
If they didn't do anything with power management, as soon as the Nvidia driver is loaded, idle power consumption almost doubles in case of my 9570. So it would probably still require some module loading/unloading for optimal power usage.
https://www.reddit.com/r/linux/comments … dium=web2x According to this answer on reddit, X server needs to be started with nvidia module loaded. So, load/unload tricks we have used so far might not work. Additional power management features are only available on Turing GPU's and it's still experimental, according to release notes.
Offline