Closed noxifoxi closed 2 months ago
Strange. It's working as expected for me, on an RX 6600 XT.
Show me the output for lshw -c video 2>/dev/null | grep -i nvidia
on both the host and in the container
I think the script is having an issue excluding inactive GPUs.
I have an NVIDIA GPU installed for VM passthrough, the GPU is not used at all by the host OS as indicated by driver=vfio-pci
under configuration.
lshw -c video
*-display
description: VGA compatible controller
product: Navi 21 [Radeon RX 6800/6800 XT / 6900 XT]
vendor: Advanced Micro Devices, Inc. [AMD/ATI]
physical id: 0
bus info: pci@0000:03:00.0
logical name: /dev/fb0
version: c1
width: 64 bits
clock: 33MHz
capabilities: vga_controller bus_master cap_list rom fb
configuration: depth=32 driver=amdgpu latency=0 resolution=2560,1440
resources: iomemory:f80-f7f iomemory:fc0-fbf irq:142 memory:f800000000-fbffffffff memory:fc00000000-fc0fffffff ioport:f000(size=256) memory:f5f00000-f5ffffff memory:f6000000-f601ffff
*-display
description: VGA compatible controller
product: GP106 [GeForce GTX 1060 6GB]
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:0e:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: vga_controller cap_list rom
configuration: driver=vfio-pci latency=0
resources: iomemory:fc0-fbf iomemory:fc0-fbf irq:255 memory:f4000000-f4ffffff memory:fc20000000-fc2fffffff memory:fc30000000-fc31ffffff ioport:d000(size=128) memory:f5000000-f507ffff
*-display
description: VGA compatible controller
product: Raphael
vendor: Advanced Micro Devices, Inc. [AMD/ATI]
physical id: 0
bus info: pci@0000:18:00.0
version: c2
width: 64 bits
clock: 33MHz
capabilities: vga_controller bus_master cap_list
configuration: driver=amdgpu latency=0
resources: iomemory:fc0-fbf iomemory:fc0-fbf irq:106 memory:fc40000000-fc4fffffff memory:fc50000000-fc501fffff ioport:e000(size=256) memory:f5e00000-f5e7ffff
lshw -c video 2>/dev/null | grep -i nvidia
for good measure:
Host:
vendor: NVIDIA Corporation
Container:
vendor: NVIDIA Corporation
The script basically just uses the line I requested you to run for its GPU detection, so more niche situations like secondary GPUs for VM passthrough aren't currently accounted for.
Specifically, the setup.sh
script checks whether or not an Nvidia GPU is present (as, if an Nvidia GPU is there, it's usually likely that it's the primary GPU) then applies Nvidia-specific options if it is. The container also does a GPU check to apply relevant workarounds, but it checks for Nvidia first for the aforementioned reason.
tl;dr: Whatever fix is done for this will need to be done in both setup.sh
and /etc/profile.d/davinci.sh
. e.g. checking for driver=<driver-name>
rather than checking vendor name like we do now
In the meantime, your workaround of using run-davinci
should suffice; if you need to run any of the other programs included with Resolve, you can also edit /etc/profile.d/davinci.sh
in your container and remove the Nvidia GPU check for now.
Describe the bug
The script recognises my AMD GPU as an NVIDIA GPU, which I assume is what is causing the problem.
When launching resolve normally I get the error described here: #74
I had to change the
.desktop
entry toin order to force rusticl and for it to work.
Installation method used
setup.sh
To Reproduce
Expected behavior
The script should detect the GPU correctly and set rusticl as the default for AMD GPUs so there is no tinkering required afterwards.
System information (please complete the following)