On Mar 25, 2016, at 13:31, Brad Bozarth <prettygood(a)cs.stanford.edu> wrote:
On Fri, Mar 25, 2016 at 1:18 AM Jike Song <jike.song(a)intel.com>
Sorry, I have no idea about how the intel gfx driver in MacOS works.
To me it seems that MacOS is even not officially supported to be
running inside a VM, but I might have misunderstood the situation.
Well, we can run Mac OS X in KVM since years - and VMware does so, too.
As I asked the other month I’m also interested in this. There are many valid scenarios for
running OSX in a VM:
- driver development
- testing different versions for QA and support
Yes, it's not officially supported, but can work - let's
assume the MacOS driver does nothing special, like the Windows or Linux driver used to
before Intel changed them to support guest GVT. In Chapter four of Intel's "Linux
GPU Driver Developer's Guide," it says "However with the partitioning an
unmodified i915 driver would assume a smaller graphics memory starting from address ZERO,
then requires vGPU emulation module to translate the graphics address between guest view
and host view, for all registers and command opcodes which contain a graphics memory
I'm just wondering if Intel's linux driver currently has the capability to do
this translation, or if we'd need to add that.
I have the impression right now the Intel vGPU model only works with modified guest
I would also be inserted in running unmodified guests (such as OSX/Darwin).
Any guidance from the Intel vGPU experts what to modify would be welcome ;-)!
Even if this includes a performance hit: running at 50% of hw performance is still
enormously faster than SW blitting, .. and some apps requiring GL not starting, ...
ExactCODE GmbH, Lietzenburger Str. 42, DE-10789 Berlin