You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, laptops with NVIDIA Optimus needs disabling the integrated GPU, or otherwise NVIDIA proprietary driver may not work.
Since NVIDIA driver v435.17, Prime offload is supported, such that iGPU is used for display but optionally some intensive jobs could be done on NVIDIA driver, if the user explicitly choose to.
Main benefits
Adding instructions on how to set this up is mainly beneficial to laptop users who cannot disable iGPU in the BIOS. But it also serves the following use cases:
A few applications can use NVIDIA GPU(s) for (parallel) GPU computations. But as introduced in MATLAB's documentation:
In general you can achieve the best performance when your GPU is dedicated to computing. It is usually not practical to use the same GPU device for both computations and graphics, because of the amount of memory taken up for problems of reasonable size and the constant use of the device by the system for graphics. If possible, obtain a separate device for graphics.
Therefore, it's common for data practitioners to connect display to iGPU. With offloading support, they could fully utilise gpu performance for computations, and meanwhile allow other applications to be rendered by NVIDIA GPU as well.
iGPU in general consumes less battery, only offloading selected tasks to NVIDIA GPU prolongs battery life.
Additional notes
NVIDIA's documentation says this functionality requires several patches not made into Xorg-server's release as time of writing. But I found they are included in xorg-server-1.20.6. So the documentation should notify the user to use a recent version of Clear Linux and NVIDIA driver.
An example as for setting up PRIME offloading with GDM could be found in Manjaro's forum.
The documentation could optionally refer users to NVIDIA's documentation for using environment variables to run a program with NVIDIA gpu. Alternatively, download the nvidia-prime script from Arch Linux's nvidia-prime package, and install it somewhere in the PATH ( but not in /usr/bin ).
The text was updated successfully, but these errors were encountered:
Description
Currently, laptops with NVIDIA Optimus needs disabling the integrated GPU, or otherwise NVIDIA proprietary driver may not work.
Since NVIDIA driver
v435.17
, Prime offload is supported, such that iGPU is used for display but optionally some intensive jobs could be done on NVIDIA driver, if the user explicitly choose to.Main benefits
Adding instructions on how to set this up is mainly beneficial to laptop users who cannot disable iGPU in the BIOS. But it also serves the following use cases:
A few applications can use NVIDIA GPU(s) for (parallel) GPU computations. But as introduced in MATLAB's documentation:
Therefore, it's common for data practitioners to connect display to iGPU. With offloading support, they could fully utilise gpu performance for computations, and meanwhile allow other applications to be rendered by NVIDIA GPU as well.
iGPU in general consumes less battery, only offloading selected tasks to NVIDIA GPU prolongs battery life.
Additional notes
Xorg-server
's release as time of writing. But I found they are included in xorg-server-1.20.6. So the documentation should notify the user to use a recent version ofClear Linux
and NVIDIA driver.GDM
could be found in Manjaro's forum.nvidia-prime
script from Arch Linux's nvidia-prime package, and install it somewhere in thePATH
( but not in/usr/bin
).The text was updated successfully, but these errors were encountered: