[vfio-users] Quadro M1200 Mobile for CUDA

[Alex Williamson]

On Mon, 15 Oct 2018 18:00:57 +0200
Simon Brand <1992 at gmx.net> wrote:

> Good evening,
> 
> I have a Lenovo Thinkpad P51 and I want to pass the GPU (GM107GLM
> [Quadro M1200 Mobile]) to (linux) VM. I only want to run cuda on the
> GPU, so there is no output needed.

I don't really have any advice to give, but I will note that hybrid
laptop graphics are mostly an unsolved problem for device assignment,
afaik.  Last I looked at Optimus-based hybrid graphics, there are
numerous ACPI hooks and even SMI interfaces, presumably so we can't see
the secret sauce that makes it work.  Associating an Intel graphics
with the VM is probably barely scratching the surface of the
dependencies.  It seems like there have been some semi-success
stories with people using bumblebee-type approaches to make it work,
but I have no further details.  In general, I suspect it's easier to
make external GPU solutions work with laptops than it is to assign
hybrid graphics in many cases.  Thanks,

Alex