diff options
Diffstat (limited to 'core')
| -rw-r--r-- | core/modules/qemu/runvirt-plugin-qemu/src/main/java/org/openslx/runvirt/plugin/qemu/App.java | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/core/modules/qemu/runvirt-plugin-qemu/src/main/java/org/openslx/runvirt/plugin/qemu/App.java b/core/modules/qemu/runvirt-plugin-qemu/src/main/java/org/openslx/runvirt/plugin/qemu/App.java index 6626f5fc..1566accc 100644 --- a/core/modules/qemu/runvirt-plugin-qemu/src/main/java/org/openslx/runvirt/plugin/qemu/App.java +++ b/core/modules/qemu/runvirt-plugin-qemu/src/main/java/org/openslx/runvirt/plugin/qemu/App.java @@ -188,9 +188,15 @@ public class App transformationManager.register( new TransformationSpecificQemuGraphics( hypervisorQemu ), true ); transformationManager.register( new TransformationSpecificQemuSerialDevices( hypervisorQemu ), true ); transformationManager.register( new TransformationSpecificQemuMdevPassthroughIntel( hypervisorQemu ), false ); + // For now, do not use looking-glass, ever. Windows can handle rendering on the NVIDIA GPU and shoving the + // frames onto the QXL device just fine, which might not be as low-latency as looking-glass, but is WAY + // less fragile, doesn't require the looking-glass service in the VM, and doesn't require anything to + // be connected to the nvidia card. So long, looking-glass! + /* String os = config.getLibOsInfoOsId(); lookingGlass = cmdLn.isNvidiaGpuPassthroughEnabled() && ( "http://microsoft.com/win/10".equals( os ) || "http://microsoft.com/win/11".equals( os ) ); + */ transformationManager.register( new TransformationSpecificQemuPciPassthrough( hypervisorQemu, lookingGlass ), false ); } |