Tuxedo InfinityBook Pro 14 Gen 8 Intel Iris Xe + NVIDIA GeForce RTX 3050

I am, but I can’t see anything other than the section title and the footer…

That’s a pity. The computers look pretty good and I’d like to see Manjaro have the same level of performance on those machines as TUXEDO OS. Perhaps we can achieve this if the community continues to share their solutions as they find them.

I can say that My Tuxedo running Manjaro is better than using the Tuxedo OS.

I did choose my Tux based on the low power consumption when excluding the Nvidia GPU - also knowing that Nvidia and Linux is a sub optimal combo.

Just a thought
You are aware that to make the tuxedo drivers work - as they are based on dkms - you need headers for your kernel and dkms - right?

Honestly, I didn’t know. Unfortunately, my knowledge of Linux/Arch/Manjaro hasn’t reached that point yet. What should I do to install them or how should I proceed?

That’s precisely what I’d like to achieve.
The TUXEDO team got back to me, giving me some basic tips and saying that Manjaro is not an officially supported distribution. So if I want to stick with Manjaro, I’ll have to do it on my own.

Since we have the same computer, you wouldn’t be able to help me or point me to some sources of information where I can achieve the same level of performance as yours?
My only interest in the Nvidia graphics card is for prototyping machine learning models, so I can perfectly well have it switched off by default and only switch it on to run specific code.

can you check that onevpl is installed ?

pacman -Qs onevpl
pacman -Qs onevpl-intel-gp

OK, I found your post about kernels/headers/dkms. I installed the headers for my kernels (6.5, 6.6, and 6.7), and reinstalled TUXEDO Control Center and the TUXEDO drivers.
One difference stands out: Control Centre now shows more information about the CPU and other options have become available (keyboard back-light and battery charging options, for example).
However, I still can’t change the screen resolution.

onevpl yes, onevpl-intel-gpu no. I installed it.
Output of pacman -Qs onevpl:

local/libvpl 2.10.1-1
    Intel Video Processing Library
local/onevpl-intel-gpu 24.1.1-1
    oneVPL runtime implementation for Intel GPUs (Tiger Lake and newer)

I restarted the computer and, as before, I still can’t change the screen resolution.

I had to install dkms for the drivers to work.
What say

dkms status

Just a thought that came to me … last night

Usually the iGPU is used but as Tuxedo provides customized firmware - it is entirely possible that systems with dGPU has been configured to use that.

Do you have somewhere in the firmware an option to prioritise the primary GPU?

I created topic to collect what we know - [How To] Configure your Tuxedo Laptop

1 Like
tuxedo-drivers/4.0.3, 6.5.13-7-MANJARO, x86_64: installed
tuxedo-drivers/4.0.3, 6.6.10-1-MANJARO, x86_64: installed
tuxedo-drivers/4.0.3, 6.7.0-0-MANJARO, x86_64: installed

It seems they are installed.

I had a look and no, I don’t.

This is gold. :coin: Thanks a million!

I have followed the steps you described in your [How To]. Everything seems to be working now, except changing the resolution of the video card.

I had optimus-manager and optimus-manager-plasma installed. I removed them and defaulted to envycontrol. Now, the output of glxinfo | grep renderer is

    GLX_MESA_copy_sub_buffer, GLX_MESA_gl_interop, GLX_MESA_query_renderer, 
    GLX_MESA_copy_sub_buffer, GLX_MESA_gl_interop, GLX_MESA_query_renderer, 
Extended renderer info (GLX_MESA_query_renderer):
OpenGL renderer string: Mesa Intel(R) Graphics (RPL-P)

I have tried both “hybrid” (currently selected) and “integrated” in the Optimus GPU Switcher, but the results are the same, i.e., no way to change the resolution of the video card.

I am considering reinstalling Manjaro, but I would rather not do that and find a solution…

Hmm - I don’t know - perhaps the resolution is a limitation of the Omnia display.

I tried to change resolution got a flickering screen - fortunately Plasma reset after 15s of non-application.

You should be able to apply scaling though - again - I am at loss since your system is a dual-gpu and mine is Iris only.

In my mkinitcpio.conf I have added

MODULES=(i915 nvidia nvidia_modeset nvidia_uvm nvidia_drm)

you can try that

Unfortunately, it didn’t work.

This might seem to indicate that Manjaro cannot currently handle this type of system well. In fact, I found that I could not change the screen resolution when I booted Manjaro from the live USB stick before installing. At that point I was hoping to find out after the installation.
I’m afraid I’m going to have to go back to TUXEDO OS, at least for a while, until a solution to this problem is found.

Not the best solution in my case because I use an external monitor and things get huge there.

Just for information: I tried two other Arch-based distributions using their live USB keys - EndeavourOS and CauchyOS. Both had the same symptoms I reported here, i.e., I could not change the screen resolution. Also, neither was able to start X using the default drivers/configuration. To get to the Plasma desktop, I had to select the option corresponding to the Nvidia drivers, just like I did when I installed Manjaro.

So, it seems that Arch-based distributions are currently unable to handle hybrid Intel Iris Xe + Nvidia RTX 3500 systems out of the box. That’s a shame because, on the one hand, it is perfectly possible (as TUXEDO OS proves) and, on the other hand, although I’m not an Nvidia fan for the same reasons as most of Linux users, these cards have become important in the field of machine learning in recent years. Therefore, it would be nice to at least be able to use the Intel video card as the default, while switching on and using the Nvidia GPU only for specific tasks, such as training ML models. Hopefully this will soon be possible in the Arch world out of the box, or at least without too much fuss.

It is a mistake to buy a laptop with switchable GPU then use it for machine learning.

I don’t know how it will work but you could remove the switchable GPU drivers. Then install video-linux and after you get that working install video-nvidia.

This is likely to make your system use the iGPU and you can use the dGPU for the task you created it for.

But if I recall correct Nvidia has some built-in limitations when it comes to run software on the GPU because so many used the cards for crypto-mining instead of gaming.

Can you elaborate a bit on this, please? I’ve seen a lot of people do this, mostly Windows users. It is very common to use the GPU for prototyping models, before full training on clusters.

I actually once did as you do - not machine learning but coding and remote service - with a laptop for everything - I used Thinkpad T410 and T420 with a dock and two monitors - I had to add a cooling pad with a huge fan 12" or so to ventilate the device.

It was never the solution I hoped for thus my comment - my apologies if I appeared offensive - I am not.

So my comment is more like my personal opinion, like been there, done that - I mean 14" laptops - I don’t use for serious work for that I rely on a real desktop with an exorbitant amount of screen estate and a full keyboard.

Laptops - as powerful they may be - they tend to get awfully hot when put to real work.

I am crazy with hardware - just ask my wife.

I have a collection of laptops - these days they are mostly used for testing and when I occasionally leave the house - having something to keep me busy.

The suggestion I made about the driver thing - I believe there was a topic once where the poster used the dGPU only for CUDA related stuff - running everything else on the iGPU.

And I think the limitations with intensive GPU calculations are real.

I just recalled - I think @philm has a system with similar specs and a Nvidia 30xx.

Not at all! I was really curious as to why you thought it was a mistake. Thanks for your feedback.
You know, we do use bigger computers for the heavy work, but a laptop is very handy because it allows us to do prototyping (small things, just to check that the code is right and we’re going in the right direction before sending it for production) and we can take it with us whenever we travel :wink:

The T420 is still my favourite. I’m still using it, with Manjaro, and I’m very proud to have been able to do all my research on it for the last five years.

This was my original idea. The only thing stopping me is the fact that currently I can’t change the resolution of the screen.

1 Like

I admit it is strange.

I remember when I ordered this Tuxedo laptop back mid november 2023 there was a choice between the omnia 3K display and a 4K display.

If I recall that choice correct the omnia 3K display had a significantly lower power consumption but also - as I recall - some limitations with the display size - whereas the 4K was more flexible.

As I went for maximum battery - my primary usage - it should be able to last a full 8h work cycle - I did choose the omnia 3K.

I am fully aware that - for reasons yet unknown - the Ubuntu based Tuxedo OS seems to be more flexible with the Iris XE than Manjaro.

On a personal note - I am trying to figure out why …

Good luck with that. I can’t afford to wait any longer at the moment, but I will switch back to Manjaro as soon as a solution is found and I have some time to make the transition. Don’t forget to post the solution when/if you find it :wink:

I admit it will be a bit painful to leave a rolling release and go back to an Ubuntu-based distro… Although I also admit that TUXEDO has done a good job of tuning their OS for their machines.

It is apparently a general issue Can't change monitor resolution on Tuxedo InfinityBook Pro 14 - Gen7 / Laptop Issues / Arch Linux Forums

The internal notebook-screen is a 3K-Omnia-Display, which I strangely can’t change the screen resolution.

Precisely what I am experiencing.

Judging from the this above topic it appears Gnome is working - Plasma is not.

Possibly this is related to the version of Plasma used with Ubuntu.