August 15, 2018 4:19 AM, tu...@posteo.de wrote: > On 08/14 11:16, Nikos Chantziaras wrote: > >> On 14/08/18 13:35, tu...@posteo.de wrote: >> Hi, >> >> after upgrading to nvidia-drivers-396.51 no CUDA devices were found. >> Last version, which works for me is nvidia-drivers-396.24-r1. >> >> Do you have the "uvm" USE flag set? It might be required for CUDA, but it's >> disabled by default (perhaps wrongly, because USE flags should follow >> upstream defaults unless there's a reason not to.) > > Yes it is: > > (this is the version, which is currentlu still working > Installed versions: 396.24-r1(0/396)^md(08:31:04 PM 08/14/2018)(X driver kms > static-libs tools uvm > -acpi -compat -gtk3 -multilib -pax_kernel -wayland ABI_MIPS="-n32 -n64 -o32" > ABI_PPC="-32 -64" > ABI_S390="-32 -64" ABI_X86="64 -32 -x32" KERNEL="linux -FreeBSD") > > and set via /etc/portage/package.use > > # required by app-admin/conky-1.10.6-r1::gentoo[nvidia,X] > # required by @selected > # required by @world (argument) >> =x11-drivers/nvidia-drivers-378.13 static-libs uvm > > What else could be the reason for the problem? > How can I fix it?
Can you also show content of modprobe.d file ? Did you read the whole wiki page ? Did you check for MSI interrupts ? https://wiki.gentoo.org/wiki/NVidia/nvidia-drivers#Driver_fails_to_initialize_when_MSI_interrupts_are_enabled Regards, -- Corentin “Nado” Pazdera
