Linux - hybrid graphics - nVidia GPU offload - slow conversion/rendering

Ideas for improvements and requests for new features in XnView MP

Moderators: XnTriq, helmut, xnview

xnuser
Posts: 3
Joined: Fri Mar 15, 2013 7:22 pm

Linux - hybrid graphics - nVidia GPU offload - slow conversion/rendering

Post by xnuser »

Hi.
If i'm running my Ubuntu in hybrid graphics mode (integrated Intel + descrete nVidia)

and if i want to launch XnView MP via nVidia graphic chip like this:

Code: Select all

export __NV_PRIME_RENDER_OFFLOAD=1 \
export __GLX_VENDOR_LIBRARY_NAME=nvidia \
export __NV_PRIME_RENDER_OFFLOAD_PROVIDER=NVIDIA-G0 \
export __VK_LAYER_NV_optimus=NVIDIA_only \
export LIBVA_DRIVER_NAME=nvidia \
export PW_PRIME_RENDER_OFFLOAD=1 \
/usr/bin/xnview
it looks like nVidia is not involved.
I think so because when I click the Export command, the preview progress bar (auto update) crawls very slowly, that is, the image is rendered as if it were not nVidia chip, but the built-in Intel graphics.
And when I click Save in the Export window, the image saves slowly again.

But sometimes, if i just run

Code: Select all

/usr/bin/xnview
i get the progress bar moving as fast as lightning. But this happens rarely and randomly.

So
How about adding information somewhere about which graphics card is actually doing the export/conversion/rendering?
I think it can help to find out if this issue related to Intel, if XnView MP cannot work correctly with hybrid graphics on Linux.