GPU virtualization method and device thereof

A virtualization and virtual machine technology, which is applied in multi-programming devices, program control devices, software emulation/interpretation/simulation, etc., can solve the problems of incomplete support and less implementation of different operating systems, and improve GPU utilization and efficiency. The effect of operating efficiency

Pending Publication Date: 2022-02-15
上海励驰半导体有限公司
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The current GPU virtualization technology is rarely implemented on heterogeneous multi-core chips in vehicles, and the support for different operating systems is not comprehensive.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • GPU virtualization method and device thereof
  • GPU virtualization method and device thereof
  • GPU virtualization method and device thereof

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0042] figure 1 For the flowchart of the GPU virtualization method according to the present invention, the following will refer to figure 1 , the GPU virtualization method of the present invention is described in detail.

[0043] First, in step 101, a host and multiple VM systems are created and managed through a hypervisor.

[0044] In the embodiment of the present invention, Hypervisor can create different VM systems by booting and loading different images, including linux, qnx, android and other types of linux systems, and install different vGPU driver versions in different VM systems to suit with different operating systems. The graphics and computing requirements of each system are converted into unified GPU tasks through the vGPU driver.

[0045] In step 102, the graphics and computing requirements in the VM system are converted into unified task commands through the virtual GPU and sent to the physical GPU.

[0046] In step 103, the physical GPU notifies the corresp...

Embodiment 2

[0048] figure 2 It is a flowchart of a GPU virtualization method supporting hardware virtualization according to the present invention, which will be referred to below figure 2 , the GPU virtualization method of the present invention is described in detail.

[0049] In step 201, a host (host) and multiple vm (Virtual Machine, virtual machine) systems are created and managed based on a hypervisor (management program).

[0050] In step 202, for a GPU that does not support hardware virtualization, a GPU manager is set in the host to be responsible for distributing tasks and accepting and forwarding irq (Interrupt ReQuest, interrupt request).

[0051] Preferably, the way the GPU is used in each vm can be configured through the hypervisor.

[0052] Further, two different GPU usage methods are supported in each vm:

[0053] method one:

[0054] exclusive mode, such as Figure 4 The usage in VM' shown in . When the hypervisor creates VM', configure GPU' as the GPU exclusively...

Embodiment 3

[0059] image 3 It is a flowchart of a GPU virtualization method supporting hardware virtualization according to the present invention, which will be referred to below image 3 , the GPU virtualization method of the present invention is described in detail.

[0060] In an embodiment of the present invention, for a GPU supporting hardware virtualization, the physical GPU may provide each VM with independent registers and interrupt numbers.

[0061] In step 301, during the creation phase of the VM, the HOST helps to create the command buffer of the VM, and notifies the physical GPU to establish a connection between the vGPU in the VM and the physical GPU.

[0062] Preferably, the step of notifying the physical GPU may be specifically executed as:

[0063] HOST informs the physical GPU of the VM's ID, the address of the command buffer, the registers to be accessed, and the IRQ number to be monitored, so that the physical registers can bind these information.

[0064] In step 3...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to a GPU virtualization method, which comprises the following steps that a host and a plurality of VM systems are created and managed through a management program; graphics and calculation requirements in the VM system are converted into a unified task command through the virtual GPU, and the unified task command is sent to the physical GPU; and the physical GPU notifies the corresponding VM system after completing the task. According to the GPU virtualization method and the device thereof, GPU virtualization under various operating systems is supported, and the GPU utilization rate is increased.

Description

technical field [0001] The invention relates to the technical field of chip design, in particular to a GPU virtualization method and device on a vehicle-mounted heterogeneous multi-core processor chip. Background technique [0002] With the development of automotive electronics, driven by the needs of multimedia, connectivity, navigation, instrumentation, etc., more and more complex system-on-chip (SOC) has emerged, which integrates multiple different processor cores and various Device resources support more operating systems, and the number of GPUs in the system is usually limited. How to use limited GPU resources more efficiently and better support different operating systems has become an important issue in this type of chip. [0003] In recent years, GPU virtualization technology has made great progress in cloud computing and other fields. With the development of on-board chips, the demand for GPU virtualization is also increasing. The current GPU virtualization techno...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/455G06F9/48
CPCG06F9/45558G06F9/4812G06F2009/45575
Inventor 戴莉莉
Owner 上海励驰半导体有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap