[HowTo] Faster application experience on hybrid-intel-nvidia

Difficulty: ★☆☆☆☆

Tutorial for Intel & NVIDIA GPU (Optimus laptop)

LibreOffice - fully working openCL
You need install the cuda package, it’s a hefty 5,4GB large!*.
When you install the cuda package, openCL will be fully usable in every application.(LibreOffice, Darktable, RawTherapee, etc.)

To install this package use:

pamac install cuda

Gaming experience (fps improvements)
Install the vulkan-intel and vulkan-mesa-layers packages. I’ve got more fps and stability in all games that I have (Trackmania, Star Trek online, Xonotic, Quake, etc.), now everything is smooth with a stable FPS.

To install these packages use:

pamac install vulkan-mesa-layers
pamac install vulkan-intel

All packages in this tutorial can also be installes through the official Manjaro repositories. (See screenshots below)


Note: I’ve never been able to play Trackmania on max graphic settings with my laptop before, but now it’s possible! (Better experience than Windows 10)

7 Likes

Thank u

1 Like

Anyone knows what does this package bring for the Vulkan/Mesa stacks?

After install this package, Intel GPU is capable fully working Vulkan (without this package only openGL). Why optimus laptop get more fps and better stability? I’m not sure, but i think Intel gpu can handle environment better now.
Discrate GPU (nvidia) must push data flow through integrated gpu (intel). I think if vulkan is fully working on the intel gpu, it helps nvidia gpu reach better raw performance. There is a screenshot with solutions for grahpics cards in a laptops.(I have laptop with MUXless)

2 Likes
  1. Thank you for editing your original and replying to questions.

  2. The above could be placed in your original post as well under a section:


    Why do this?


  1. I’ve edited your [HowTo] a bit to include standard Manjaro tools (pamac) instead of Arch (pacman) and added a difficulty as is standard for all our [HowTo]s.
  2. :+1: Thank you for sharing back with the community :+1: (Original post deleted)

:wink:

1 Like

Which means it also is only effective for things that use vulkan.
(it wont matter for, example, your browser or tetris)
And only for these hybrid setups, as its already included in nvidia-utils.
(if this were amd+nvidia, it would be amdvlk instead of vulkan-intel)
Also dont forget about 32bit versions (steam probably cares)

lib32-vulkan-intel  #for intel
lib32-amdvlk  #for amd
lib32-nvidia-utils  #for nvidia
lib32-vulkan-mesa-layers #this is the mesa-vulkan layers, for everybody, an optional dep of mesa

See more here:
https://wiki.archlinux.org/index.php/Vulkan

3 Likes

hi, thanks for this short tutorial

it seems standard nVidia package like Prime to be used with hybrid Intel GPU via e.g. nVidia-455xx driver includes Cuda and maybe openCL also, but I’m not sure, I usually check things via Blender if some is feature detected and activated

what about packages like
opencl-nvidia-455xx
or
opencl-nvidia-full-vulkan-all 455.46.02-1 ?

are they needed for anything else to activate some additional features to speed up things?