On 10/05/2011 03:31 PM, DRC wrote:
On 10/5/11 2:23 PM, Eric Appleman wrote:
Pretty much. VirtualGL is very nice, but we at the Bumblebee team are
looking for more direct transport methods. Perhaps something that blends
the best ideas of VirtualGL, Xpra, and XShm.
Seems like a long way around the block to support a very specific
hardware configuration. Can someone please explain what the purpose of
these laptops is, why they're designed that way, and why VirtualGL seems
to be the only solution to using H/W-accelerated 3D on them? The
Bumblebee Project came completely out of left field, and it's curious to
me that it seems to have about 10x the social media following as the
software it's based on.
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
VirtualGL-Users mailing list
VirtualGL-Users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/virtualgl-users
An Nvidia Optimus laptop is a laptop that has two GPUs: The dedicated
Nvidia GPU that has no physical output or LVDS connection and the Intel
GPU (sometimes on the CPU die) that displays the desktop.
The Nvidia GPU is typically wired through the Intel GPU and is only
activated when needed (games, intense multimedia, etc).
Here's a whitepaper:
http://www.nvidia.com/object/LO_optimus_whitepapers.html
As I mentioned before, VirtualGL is a very nice solution for hardware
accelerated transportation. But we trade a fair amount of 3D performance
and hardware feature exposure (NV17, VDPAU, etc) by using it.
VirtualGL:
|$ optirun glxspheres
Polygons in scene: 62464
Visual ID of window: 0x21
Context is Direct
OpenGL Renderer: Gallium 0.4 on NVA8
30.888309 frames/sec - 27.538780 Mpixels/sec|
Hybrid-Windump or Xpra:
|DISPLAY=:8 glxspheres
Polygons in scene: 62464
Visual ID of window: 0x11c
Context is Direct
OpenGL Renderer: Gallium 0.4 on NVA8
147.114528 frames/sec - 164.179813 Mpixels/sec|
We looked into using hybrid-windump [1] to overcome those limitations
and found the solution to be more suitable for our needs. There was one
problem though. The window code was lacking and one would have to know
the window ID hex in order to place the window on the Intel X server or
place the entire root window of the Nvidia card.
We could write our own software solution from scratch, but then we would
encounter the same problems of hybrid-windump and perhaps duplicate
work. On the hardware side of things, we're simply not capable. We'll
defer to the Nouveau, Wayland, and X Hotplugging Redesign developers to
get the necessary bits in order in the long term. I know David Airlied
is trying to merge his PRIME and Nvidia PCOPY work into Nouveau and
future kernels, but the last few months have been quiet and his work is
agnostic to the Nvidia binary drivers.
VirtualGL and Xpra are attractive solutions because they are feature
complete for manipulating non-root windows. However, the transport
methods assume a remote machine, not a second X server local to the machine.
[1] https://github.com/harp1n/hybrid-windump
- Eric
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2dcopy1
_______________________________________________
VirtualGL-Users mailing list
VirtualGL-Users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/virtualgl-users
