Para-virtualized domain, hull, and geometry shaders

A computer program product for use at a computer system, the computer program product for implementing a method for providing GPU-accelerated computing functionality to a guest application executing in a child partition of a para-virtualized execution environment, the computer program product comprising one or more hardware storage devices having stored thereon computer-executable instructions that, when executed at a processor, cause the computer system to perform the method, including the following:

• instantiate a virtual machine session, including instantiating a hypervisor that provides (i) a root partition having access to the physical GPU, and (ii) the child partition which executes the guest application;
• present a virtualized graphics processing unit (“vGPU”) to the guest application, the vGPU executing within the child partition, the vGPU including:
• a first user-mode driver (“UMD”), executing in a user mode of the child partition, that presents a first plurality of device driver interfaces (“DDIs”) of a first version of a rendering framework to the guest application, the first plurality of DDIs enabling the guest application to send commands to the vGPU for programming a GPU pipeline of the physical GPU to utilize one or more features unique to the first version of the rendering framework, including utilizing at least one of:
• a domain shader, a hull shader, or a geometric shader;
• a second UMD, executing in the user mode of the child partition, that presents a second plurality of DDIs of a different version of the rendering framework to the guest application, the second plurality of DDIs enabling the guest application to send commands to the vGPU for programming a GPU pipeline of the physical GPU to utilize one or more features unique to the different version of the rendering framework;
• a first kernel-mode driver (“KMD”) processing component that corresponds to the first UMD, that executes in a kernel mode of the child partition, and that receives a first command buffer from the first UMD and constructs a corresponding first direct memory access (DMA) buffer;
• and
• a second KMD processing component that corresponds to the second UMD, that executes in the kernel mode of the child partition, and that receives a second command buffer from the second UMD and constructs a corresponding second DMA buffer;
• receive, at a first render component executing within the root partition, the first DMA buffer from the first KMD processing component, the first DMA buffer including at least one first physical GPU-specific command for using a feature unique to the first version of the rendering framework including one or more of a domain shader, a hull shader, or a geometric shader at the physical GPU;
• receive, at a second render component executing within the root partition, the second DMA buffer from the second KMD processing component, the second DMA buffer including at least one second physical GPU-specific command for using a feature unique to the different version of the rendering framework;
• and
• schedule the at least one first physical GPU-specific command and the at least one second physical GPU-specific command for execution at the physical GPU.

A computer system, the computer system comprising:

• one or more processors;
• a graphics processing unit (“GPU”);
• system memory;
• and
• one or more computer-readable storage devices having stored thereon computer-executable instructions representing a virtualized graphics processing unit (“vGPU”) and a render component,
• wherein the vGPU executes within the child partition and includes:
• a first user-mode driver (“UMD”), executing in a user mode of the child partition, that presents a first plurality of device driver interfaces (“DDIs”) of a first version of a rendering framework to the guest application, the first plurality of DDIs enabling the guest application to send commands to the vGPU for programming a GPU pipeline of the physical GPU to utilize one or more features unique to the first version of the rendering framework, including utilizing at least one of:
• a domain shader, a hull shader, or a geometric shader;
• a second UMD, executing in the user mode of the child partition, that presents a second plurality of DDIs of a different version of the rendering framework to the guest application, the second plurality of DDIs enabling the guest application to send commands to the vGPU for programming a GPU pipeline of the physical GPU to utilize one or more features unique to the different version of the rendering framework;
• a first kernel-mode driver (“KMD”) processing component that corresponds to the first UMD, that executes in a kernel mode of the child partition, and that receives a first command buffer from the first UMD and constructs a corresponding first direct memory access (DMA) buffer;
• and
• a second KMD processing component that corresponds to the second UMD, that executes in the kernel mode of the child partition, and that receives a second command buffer from the second UMD and constructs a corresponding second DMA buffer;
• and
• wherein the render component executes in a root partition, and includes:
• a first renderer that receives the first DMA buffer from the first KMD processing component, the first DMA buffer including at least one first physical GPU-specific command for using a feature unique to the first version of the rendering framework including one or more of a domain shader, a hull shader, or a geometric shader at the physical GPU, and that schedules the at least one first physical GPU-specific command for execution at the physical GPU;
• and
• a second renderer that receives the second DMA buffer from the second KMD processing component, the second DMA buffer including at least one second physical GPU-specific command for using a feature unique to the different version of the rendering framework, and that schedules the at least one first physical GPU-specific command for execution at the physical GPU.