vSphere 5.1 GPU acceleration now supported on XenDesktop with HDX 3D Pro and XenApp with HDX 3D
VMware introduced GPU hardware acceleration support for its latest hypervisor release of vSphere 5.1. XenDesktop 5.6 FP1 and XenApp 6.5 FP1 have tested successfully with the new GPU acceleration capabilities on vSphere 5.1 and Citrix officially supports these configurations.
vSphere 5.1 introduced two GPU acceleration methods: Direct graphics acceleration (vDGA) and Shared graphics acceleration (vSGA). vDGA is similar to XenServer’s GPU pass-through method where each graphics processor supports a single VM to enable high performance rendering for designers and engineers that require a dedicated GPU or for GPU sharing with multi-user XenApp VMs. vSGA is a light-weight implementation of GPU sharing designed for business graphics such as Microsoft Office and Windows Aero graphics and may not be suitable for demanding 3D professional graphics applications such as CAD, geographical information systems and medical imaging. vSGA is based on API Intercept, a software implementation which has the following limitations:
API Intercept is limited to older versions of OpenGL and DirectX
API Intercept may not perform well with large 3D models since the graphics commands have to be sent from the user session to Session 0 which controls the GPU
To address the GPU sharing requirement for OpenGL- based 3D professional graphics apps, XenApp recently released an OpenGL GPU sharing add-on that overcomes the performance limitations of the API intercept method by directly talking to the GPU video driver. This new capability augments the DirectX GPU sharing feature introduced in XenApp 6.0. To learn more about the OpenGL GPU Sharing Add-On for XenApp, click here.