VirtualBox and YDL for CUDA
Posted: 08 Aug 2012, 20:49Hi everyone,
I run YDL for CUDA using VirtualBox - which is a virtualizer like VMWare or Hyper-V. It basically allows me to run YDL for CUDA in a session on another host computer.
Part of the power of VirtualBox is the integration component. It has a guest additions module that links the guest OS (i.e. YDL for CUDA) into your host OS.
It works pretty well, but certain things have never worked. It turns out that the Makefile for the guest additions actually didn't recognize YDL for CUDA's "uname -r" string. This is the kernel ID.
I have a Makefile that will work - I've attached it to this post. You need to drop it in this directory: /usr/src/vboxguest-4.1.18/vboxsf
Unfortunately, I don't know how to tell you to compile with it. I have to kick off the guest additions installation, then dynamically drop this file in place before it compiles with it. I'm sure there's an official VirtualBox procedure for this - but I don't know what it is. I'll ask them on their forums.
Cheers,
Paul
I run YDL for CUDA using VirtualBox - which is a virtualizer like VMWare or Hyper-V. It basically allows me to run YDL for CUDA in a session on another host computer.
Part of the power of VirtualBox is the integration component. It has a guest additions module that links the guest OS (i.e. YDL for CUDA) into your host OS.
It works pretty well, but certain things have never worked. It turns out that the Makefile for the guest additions actually didn't recognize YDL for CUDA's "uname -r" string. This is the kernel ID.
I have a Makefile that will work - I've attached it to this post. You need to drop it in this directory: /usr/src/vboxguest-4.1.18/vboxsf
Unfortunately, I don't know how to tell you to compile with it. I have to kick off the guest additions installation, then dynamically drop this file in place before it compiles with it. I'm sure there's an official VirtualBox procedure for this - but I don't know what it is. I'll ask them on their forums.
Cheers,
Paul