Cloud PC service system with independent graphics card

Abstract

The invention relates to a cloud PC service system with an independent graphics card. The system comprises a power supply module, a management module and a plurality of computing nodes which are arranged in a machine frame, the management module is connected with the power supply module and each computing node through an I2C interface so as to read the working state of the power supply module and each computing node and report the working state to a cloud PC management platform through a network interface, and each computing node comprises a display card, a computing board and a coding board. The graphics card is connected with the computing board through a PCIe extension line, the coding board is provided with at least two DP interfaces and at least one HDMI interface, and the coding board is connected with the graphics card through the two DP interfaces or the HDMI interface. The overall performance can be improved.

Description

Technical Field

[0001] The invention relates to a cloud PC service system with an independent graphics card, and belongs to the field of computers. Background Art

[0002] Cloud PC server is a virtualized desktop solution built on cloud computing technology. Its core concept is to migrate traditional local PC functions to the cloud. Users can access the virtual desktop in the cloud through a network connection to achieve efficient office and computing needs anytime and anywhere. However, existing cloud desktops mainly use virtual graphics cards. When multiple virtual machines share the same physical graphics card resources, resource competition is prone to occur, resulting in performance degradation. In addition, PCFarm, which provides underlying hardware support for cloud PC servers, uses graphics card streaming to transmit images in its traditional architecture. When multiple users share the same graphics card resources, resource competition is also prone to occur, resulting in performance fluctuations. Summary of the invention

[0003] The present invention provides a cloud PC service system with an independent graphics card, aiming to solve at least one of the technical problems existing in the prior art. To this end, the cloud PC service system proposed by the present invention can effectively improve the overall performance.

[0004] The technical solution of the present invention relates to a cloud PC service system, including: a cloud PC service system, including a power module, a management module and multiple computing nodes arranged in a machine frame; the management module is connected to the power module and each of the computing nodes through an I2C interface to read the working status of the power module and each of the computing nodes, and report to a cloud PC management platform through a network interface; the computing node includes a graphics card, a computing board and an encoding board; the graphics card is connected to the computing board through a PCIe extension cable; the encoding board is provided with at least 2 DP interfaces and at least 1 HDMI interface, and the encoding board is connected to the graphics card through 2 DP interfaces or 1 HDMI interface.

[0005] Furthermore, a chassis management software is also provided, and the architecture of the chassis management software is provided with an application layer, an interface layer and a system layer; the application layer includes a GPU computing node for slot number detection, power on and off status detection, temperature detection, in-place detection, power on and off control and resource usage detection, as well as a chassis ID reporting and abnormal alarm management module, a fan module and a power supply module; the interface layer includes an i2c_client port for accessing the i2c interface to communicate with the I2C of each node EC microcontroller, and a Socket_api port for communicating with the cloud PC management platform through the udp interface; the system layer includes a linux operating system.

[0006] Furthermore, the software architecture of the encoding board includes an encoding card system layer, an encoding card interface layer and an encoding card application layer; the encoding card interface layer includes a USBoIP interface library and an embedded media library; the encoding card application layer mainly provides the ability of PC to access the cloud through the GPU computing node-encoding card software architecture, and its functions include user management, computing resource management, network and data security, and single-node monitoring and operation and maintenance.

[0007] Furthermore, the image is transmitted by high-definition low-delay encoding, wherein additional chroma contour information is introduced in the encoding process of the YUV420 format for layered encoding, the image in the YUV420 format is used as the basic coding layer, and the U and V contour information are encoded as the enhancement layer; during decoding, the high-quality chroma component is reconstructed by the above information to approximate the effect of YUV444; wherein, in the encoding and transmission of the base layer, the YUV420 base image is passed to the encoder for standard YUV420 encoding, and the PTS is written, and the encoded video stream is stored or transmitted in a conventional manner; wherein, in the encoding and transmission of the enhancement layer, the YUV420 chroma contour image is passed to the encoder for standard YUV420 encoding, and the PTS is written, and the encoded video stream is stored or transmitted in a conventional manner; wherein, in the decoding process, the decoding of the enhancement layer uses a standard decoder to decode the enhancement layer video stream to obtain a YUV420 chroma contour image; the base layer decoding uses a standard decoder to decode the base layer video stream to obtain a YUV420 base image, and according to the PTS value, the enhancement frame corresponding to the current base frame is found, and the corresponding chroma contour information is extracted.

[0008] Further, the YUV transmission includes: S410, YUV444 is downsampled to YUV420, the image in YUV420 format is encoded as the basic coding layer, the chroma profile information is synchronously extracted from the YUV444 format, the chroma profile information is combined into an image in YUV420 format, and encoded as an enhancement layer; PTS is used between layers for time synchronization; S420, the base layer and the enhancement layer are respectively sent to the decoding end as independent coding streams; S430, the decoding end receives a base layer code stream and multiple enhancement layer code streams, and decodes them; wherein, the base layer code stream is decoded to generate a YUV420 base image, and the enhancement layer code stream is decoded to generate a corresponding YUV420 color enhanced image; S440, the base image uses PTS to find the corresponding color enhanced image, and synthesizes the final YUV444 image.

[0009] Furthermore, VOP overlay is used to process single-screen display scenarios, dual-screen different display scenarios and dual-screen same display scenarios; Weston and VPX transmit images to different Planes through DRM, and then VOP transmits them to different display interfaces; the GPU of FPGA or SOC collects images from the interface, overlays them and transmits them to the display; Among them, VOP adopts a unified display architecture, shares the Plane resources inside VOP, and supports layer overlay of Planes.

[0010] Furthermore, USB data transmission adopts the bridge USBIP method; among them, when the USBIP client on the cloud PC side needs to connect to the 3240 port of the USBIP server on the decoding side: the USBIP client first connects to the local 3240 port, the transmission agent on the cloud PC side will establish a dedicated virtual channel with the transmission agent on the encoding side, and the transmission agent on the encoding side is connected to the 3240 port on the decoding side.

[0011] Furthermore, the network ports of the encoding board and the management module are both used to connect to the user terminal access network, and the network port of the computing board is used to connect to the business data network.

[0012] Furthermore, it also includes a backplane for connecting the power module, the management module and multiple computing nodes, and the backplane is arranged on the top of the frame.

[0013] Furthermore, the power board of the computing node is provided with a Standby power supply for supplying power to the encoding card, so as to maintain the power output when shutting down.

[0014] The beneficial effects of the present invention are as follows: The cloud PC service system with an independent graphics card of the present invention adopts an independent graphics card, and uses an encoding card to transmit images in a high-definition and low-latency encoding manner, which can effectively improve the overall performance. Network security isolation is provided, and the cloud PC server can include two isolated networks, and data cannot be transmitted between the two networks. VOP overlay processing is adopted to realize the output of 2 VOPs corresponding to three scenes, thereby reducing the user terminal cost. USB data transmission adopts the bridge USBIP mode, and the USB Device side of the encoding board appears as a customized USB data receiving / transmitting device. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of the hardware architecture of the cloud PC service system according to an embodiment of the present invention.

[0016] Figure 2 is a schematic diagram of the structure of a computing node according to an embodiment of the present invention.

[0017] Figure 3 is a schematic diagram of the connection relationship of computing nodes according to an embodiment of the present invention.

[0018] Figure 4 is a schematic diagram of a coding card according to an embodiment of the present invention.

[0019] Figure 5 Schematic diagram of component connections of a cloud PC service system according to an embodiment of the present invention.

[0020] Figure 6 Schematic diagram of network connection of a cloud PC service system according to an embodiment of the present invention.

[0021] Figure 7 is a schematic diagram of a backplane interface according to an embodiment of the present invention.

[0022] Figure 8 Schematic diagram of chassis management software architecture of a cloud PC service system according to an embodiment of the present invention.

[0023] Fig. 9 2 is a schematic diagram of connecting a power board according to an embodiment of the present invention.

[0024] Fig.10 is a schematic diagram of fan module connections according to an embodiment of the present invention.

[0025] Fig.11 Schematic diagram of the coding card structure and heat dissipation arrangement according to an embodiment of the present invention.

[0026] Fig.12 It is a schematic diagram of the GPU computing node-encoding card software architecture according to an embodiment of the present invention.

[0027] Fig.13 It is a functional logic diagram of GPU computing node-encoding card software according to an embodiment of the present invention.

[0028] Fig.14 It is a management flow chart of a cloud PC management platform and a server according to an embodiment of the present invention.

[0029] Fig.15 The present invention is a flowchart of a user logging in and accessing a desktop according to an embodiment of the present invention.

[0030] Fig.16 4 is a YUV transmission flow chart according to an embodiment of the present invention.

[0031] Fig.17 Schematic diagram of a VOP overlay mechanism according to an embodiment of the present invention.

[0032] Fig.18 is a schematic diagram of a bridge USBIP according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that, unless otherwise specified, when a feature is referred to as "fixed" or "connected" to another feature, it can be directly fixed or connected to another feature, or it can be indirectly fixed or connected to another feature. The singular forms of "a", "said" and "the" used herein are also intended to include plural forms, unless the context clearly indicates other meanings. In addition, unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art. The terms used in this specification are only for describing specific embodiments, not for limiting the present invention. The term "and / or" used herein includes any combination of one or more related listed items. It should be understood that although the terms first, second, third, etc. may be used in the present disclosure to describe various elements, these elements should not be limited to these terms. These terms are only used to distinguish elements of the same type from each other. For example, without departing from the scope of the present disclosure, the first element may also be referred to as the second element, and similarly, the second element may also be referred to as the first element. The use of any and all examples, or exemplary language ("such as," "eg," etc.) provided herein, is intended merely to better illuminate embodiments of the invention and does not pose a limitation on the scope of the invention unless otherwise claimed.

[0034] Reference Figures 1 to 18 In some embodiments, the cloud PC service system according to the technical solution of the present invention includes a power module, a management module and multiple computing nodes arranged in a machine frame; the management module is connected to the power module and each computing node through an I2C interface to read the working status of the power module and each computing node, and report to the cloud PC management platform through a network interface; the computing node includes a graphics card, a computing board and an encoding board; the graphics card is connected to the computing board through a PCIe extension cable; the encoding board is provided with at least 2 DP interfaces and at least 1 HDMI interface, and the encoding board is connected to the graphics card through 2 DP interfaces or 1 HDMI interface.

[0035] Specifically, the cloud PC service system of the present invention is provided with a plurality of computing nodes, and the computing nodes include a graphics card, a computing board, a coding board and a power board, that is, the graphics card, the computing board, the coding board and the power board can constitute a computing node. Figure 2 and Figure 3, because most of the current graphics card output ports are in the form of 3 DP + 1 HDMI combination, the encoding card of the present invention is provided with 2 DP interfaces and 1 HDMI interface, which can realize the 3-choose-2 mode, thereby being compatible with the graphics card interfaces on the market. Furthermore, the encoding board can adopt different types of encoding and decoding main processors. Furthermore, the graphics card is connected to the computing board PCIe slot via a PCIe extension cable, making the installation flexible. Furthermore, the computing board of the present invention can adopt a universal motherboard, which can be quickly implemented through the "mini ITX motherboard" + "encoding card" solution, and the CPU and memory stick can arbitrarily select the mainstream configuration on the market, and the corresponding motherboard can be selected.

[0036] Furthermore, the independent graphics card used by the cloud PC service system of the present invention can be a general commercial graphics card, or an A2000 professional graphics card, or a low-end graphics card such as RTX3060, or a high-end graphics card such as RTX4090. Furthermore, the cloud PC service system of the present invention is compatible with existing graphics cards on the market. The present invention uses standard graphics card access to provide users with GPU computing capabilities through high-performance GPU computing to meet the needs of high-demand scenarios such as AI reasoning and graphics rendering. An independent graphics card is provided, which is suitable for product design fields and industrial design fields that require the use of tools such as 3D modeling and CAD, as well as meeting business needs such as digital special effects, animation, and film and television production that rely on GPU computing power for rendering and image processing, and supports medical imaging technologies such as CT and MRI that require rapid image processing or three-dimensional reconstruction with the help of GPU computing power, and in scenes that require a large amount of parallel computing such as AI reasoning, data analysis, or neural networks, GPUs can significantly improve training speed and algorithm efficiency.

[0037] In some embodiments, the cloud PC service system of the present invention further includes a power module, a management module and a backplane, and the connection between the power module, the management module and the computing node is realized through multiple interfaces provided on the backplane. Among them, the hardware such as the power module, the backplane, the computing node and the management module are all provided in the machine frame.

[0038] In some embodiments, see Figure 2 and Figure 5 The 2.5G network ports of the encoding board and the management module are used to connect to the user terminal access network, and the 2.5G network port of the computing board is used to connect to the business data network. The network connection of the present invention adopts the method of physical separation of the terminal access network and the user data network. The server frame does not have a switching function, and the computing board and the encoding card are directly connected to the switch. Specifically, see Figure 6 , the management module is connected to the terminal access network switch, the encoding board is connected to the terminal access network switch, and the terminal access network switch is connected to the cloud PC management platform. The computing board is connected to the business data network switch, and the business data network switch is connected to the cloud PC management platform. Further, the computing board of the embodiment of the present invention can use a computing board that supports dual 2.5G network ports.

[0039] In some embodiments, see Figure 5 The management module connects the power module and each computing node through the I2C interface, reads the working status of each computing node and power module, such as power on / off, power supply, temperature, fan speed, etc., and reports to the cloud PC management platform through the network interface. The management module connects to the cloud PC management platform through the external 1000M network to report the real-time status of the high-performance GPU server.

[0040] In some embodiments, the cloud PC service system of the present invention may adopt the form of 2U2 nodes, 4U4 nodes, 5U4 nodes, 6U6 nodes or 6U8 nodes. The technical solutions of each form of nodes are the same, but there are differences in structural size. Accordingly, the frame supports 2U2 nodes, 4U4 nodes, 5U4 nodes, 6U6 nodes and 8U8 nodes, and the encoding card can support the above-mentioned frames.

[0041] In one application implementation, the cloud PC service system of the present invention uses a 5U4 frame, and the backplane is installed on the top of the frame, so that the entire frame is ventilated from front to back, and the heat dissipation can be controlled by each blade. Figure 7 , applied to the backplane of the 5U4 chassis, multiple blade interfaces (such as blade interface 1, blade interface 2, blade interface 3 and blade interface 4) are set on one side of the backplane, and the management card interface and multiple power interfaces (such as power interface 1, power interface 2, power interface 3 and power interface 4) are set on the other side of the backplane. Further, there are four power interfaces and two of them are set on both sides of the management card interface. Among them, the blade interface is used to connect the computing board, the management card interface is used to connect the management module, and the power interface is used to connect the power module.

[0042] In some embodiments, the cloud PC service system of the present invention is provided with a power supply module, which can adopt the CPRS standard server power supply, which has the characteristics of high power density, high efficiency, modular design and flexible redundant configuration, and can support hot plugging, intelligent management and a wide input voltage range, meeting the high reliability and high performance requirements of key equipment such as high-performance GPU servers.

[0043] Furthermore, the power module adopts a standard CRPS (Common Redundant Power Supply) module, and the frame supports hot-swap and flexible configuration of 4 power modules, of which the 2U frame only supports dual power configuration. Specifically, the CRPS module adopted by the power module of the present invention is compactly designed and can provide high power output in a limited space. It supports hot-swap, which is convenient for replacing or adding power modules during system operation, improving maintenance efficiency and system availability, and supports multiple modules in parallel to achieve power redundancy (N+1 or N+N redundant configuration). It supports 1+1, 2+1 and 3+1 power supply current sharing hot standby. Even if a single module fails, it will not affect the operation of the system. It can meet 7x24 hours of high-load operation, and supports 80 PLUS certification, such as Platinum or Titanium. The conversion efficiency is usually higher than 94%-96%, and supports a wide range of input voltages (usually 90-264V AC), which can adapt to the power supply environment in different regions, and supports PMBus (Power ManagementBus) protocol, through remote monitoring and control of the operating status of the power supply, such as voltage, current, temperature, etc., and the use of a unified physical size, such as the CRPS specification defined by Intel, it is easy to integrate into standardized chassis and racks, and adopts efficient heat dissipation design and intelligent fan control to automatically adjust the fan speed according to load and temperature, reduce noise and extend service life.

[0044] In an application embodiment, the power consumption is the highest when the RTX4090 graphics card and the i9-14900K CPU are configured at the highest level. The present invention configures 4 1000W power supplies and adopts a 3+1 backup mode to achieve a maximum power of 4000W, thereby meeting the power output. Furthermore, when other graphics cards and CPUs are selected, the power supply can be selected according to the actual power consumption. The present invention preferably uses a single 800W power supply because it has the most optional brands and the cheapest price.

[0045] In some embodiments, see Figure 8 The chassis management software of the cloud PC service system of the present invention is configured with an application layer, an interface layer and a system layer. The application layer includes a GPU computing node, a fan module, a power module and a management module, the interface layer includes an i2c_client port and a socket_api port, and the system layer includes a linux operating system.

[0046] In an application embodiment, the GPU computing node in the application layer is provided with slot number detection, power-on and power-off status detection, temperature detection, in-position detection, power-on and power-off control, and resource usage detection. In the slot number detection, the management module obtains the slot number of each node by communicating with the EC microcontroller i2c of each node. In the power-on and power-off status detection, the management module obtains the power-on and power-off status of each node by communicating with the EC microcontroller i2c of each node, and determines whether the node is powered. If it is low level, it is powered on, otherwise it is powered off. In the temperature detection, the management module obtains the temperature information of each node by communicating with the EC microcontroller i2c of each node. In the in-position detection, the management module obtains the in-position information of each node by communicating with the EC microcontroller i2c of each node. If it is plugged in, it is in place, otherwise it is not in place. In the power-on and power-off control, the management module controls the IO output high and low levels or i2c to notify each node to power on and off. In the resource usage detection, the management module reads the resources of each computing node including the number of CPU cores, memory capacity, and storage capacity by communicating with the EC microcontroller i2c of each node.

[0047] Furthermore, the power module in the application layer is provided with power status detection, and the management module reads the input and output current, voltage, input and output current, voltage status, etc. of the power module through i2c communication with the EC microcontroller of each node. Furthermore, the management module in the application layer is provided with chassis ID reporting and abnormal alarm. In the chassis ID reporting, the chipid of the management module is used as the unique identifier of the chassis, and udp unicast is reported to the cloud PC management platform to distinguish the chassis. In the abnormal alarm, the fan alarm, temperature alarm, and power alarm information sent by the management module socket are sent to the cloud PC management platform.

[0048] In an application embodiment, the i2c_client port of the interface layer mainly encapsulates the interface for accessing i2c and communicating with the I2C of each node EC microcontroller. The socket_api port of the interface layer mainly encapsulates the interface for communicating with the cloud PC management platform through the udp interface.

[0049] In an application embodiment, in the system layer, the management module uses the Linux system, mainly opening two threads, which are used to receive cloud PC management platform messages via UDP and parse the messages, and to report data to the cloud PC management platform via UDP when data changes are detected through i2c polling.

[0050] In some embodiments, see Figure 2 and Figure 5 Each computing node of the present invention is a complete and independent computing unit, which can be installed with computing boards and graphics cards of different models, and has strong compatibility. Its blade supports 3-slot wide graphics cards, such as RTX4090, etc. Figure 2 and Figure 3Specifically, the power board in the computing node is used to provide 12V power to the graphics card, computing board, encoding card and fan module. The graphics card is connected to the computing board through the PCle×16 interface, and sends the output signal to the encoding card through the HDMI interface and the DP interface. As the core of the entire system, the computing board is connected to the graphics card through the PCle×4 interface, the encoding card's audio input and output port (Line in / out), two USB2.0 interfaces and the CPIO interface for determining the power on / off status, and the external network through the 1000Base-T interface.

[0051] In some embodiments, see Fig. 9 The power board of the computing node of the present invention is used to power the computing board, and an additional 12V / 2A Standby power output is added to power the encoding card. When the computer is turned off, the power supply also maintains output.

[0052] In some embodiments, the computing node is further provided with a fan module, which includes a CPU fan, a node fan, and a graphics card fan. Fig.10 The motherboard uses a standard 2U server side-blowing radiator, the frame uses two side-blowing fans, the graphics card has its own fan, and the blade reserves enough space for the graphics card to dissipate heat. Furthermore, the fan module uses two 8038 fans as node fans, and the wind direction flows from the front panel to the back panel. Combined with the position of the memory bar on the computing board, the ventilation effect can be improved.

[0053] In some embodiments, the coding board of the computing node of the present invention includes a mounting bracket, a coding card PCBA, a coding card heat sink, a fan and a thermal pad, see Fig.11 The whole machine adopts the forced air cooling method with front air and rear air outlet. The heat sink of the encoder card is in contact with the high-power chip with a thermal pad (silicone pad), which conducts the heat to the teeth of the heat sink of the encoder card. A fan is installed on the heat sink to increase heat dissipation and ventilation. At the same time, it is carried away by the airflow of the chassis, effectively improving the heat dissipation capacity of the product.

[0054] See also Figure 4 , unlike most graphics cards on the market that have 3 DP+1 HDMI output ports, the encoding card of the present invention is designed with 2 DP ports and 1 HDMI port, a 3-choose-2 mode, to be compatible with graphics card interfaces on the market. In some specific embodiments, the codec can support one HDMI direct input and support video capture, and expand two DP interfaces to adapt to the mainstream graphics card output ports on the market, so that 2 of the 3 video input interfaces can be used. The encoding board can convert the I2S digital port into analog audio input and output, and convert the RGMII port into an electrical port input.

[0055] In some embodiments, see Figure 4 and Fig.12The software architecture of the encoding board of the present invention includes an encoding card system layer, an encoding card interface layer and an encoding card application layer. The encoding card interface layer includes a USBoIP interface library and an embedded media library. The encoding card application layer mainly provides the ability of PC to access the cloud through the GPU computing node-encoding card software architecture. Its functions include user management, computing resource management, network and data security, and single-node monitoring and operation and maintenance.

[0056] Among them, the USBoIP interface library of the encoding card interface layer is provided to the application layer. When the cloud PC computing node needs to communicate with a remote USB device, it will send a request to the GPU computing node-encoding card. The GPU computing node-encoding card forwards it to the USB device connected to the cloud PC user terminal and returns the response data to the cloud PC computing node. In this way, communication is achieved just like using a locally connected USB device. The embedded media library of the encoding card interface layer is also provided to the application layer in the form of an interface library, which compresses and encodes the video signal and decompresses and decodes it to achieve the transmission and playback of video data.

[0057] In some application examples, see Fig.13 The software architecture of GPU computing node-coding card of the coding card application layer is based on modular and layered architecture, aiming to ensure the efficiency, security, maintainability and scalability of the system. The software consists of four parts: user management, computing resource management, network and data security, and system monitoring and operation. Among them, the interface management function of the coding card application layer relies on the display framework of Qt5.15.2 on Windows / Linux / macOS platform to render display and management. The communication management of the coding card interface layer is used to maintain the connection between the terminal and the desktop service, maintain the heartbeat, and ensure the transmission of signaling data. The user management of the coding card interface layer is responsible for the unified management of domain accounts and cloud PC accounts as well as the access and security control of users. The computing resource management of the coding card interface layer allocates blade resources and controls them in a networked manner. The network and data security functions of the coding card interface layer meet the enterprise network isolation requirements through network isolation and access permission control, and the data is safe and controllable. Through data transmission encryption technology, it is ensured that user network data will not be stolen or tampered with, further ensuring the security of data. The single-node monitoring and operation and maintenance function of the coding card interface layer is used to monitor the image acquisition output status of the current computing node, and can query and control the version of the GPU computing node-coding card software.

[0058] Furthermore, in the user management of the coding card application layer, unified account management is implemented, which uses AD domain unified identity authentication technology. The cloud PC user account and the computing node operating system account use the same account. After the user logs in to the cloud PC account, he can access all bound computing node hosts without having to enter the account password again. In the domain account management, the management function of the domain account is added, including creation, modification, deletion and other operations.

[0059] Furthermore, in the computing resource management of the encoding card application layer, flexible configuration of computing nodes can be achieved. It provides a variety of CPU motherboards, and users can flexibly configure them as needed and realize network control. The network encoding of computing nodes can be realized through the embedded encoding module, including video, audio and other functions, and the cloud PC server can be accessed through the Internet and the corresponding modules can be controlled accordingly.

[0060] Furthermore, in the network and data security of the encoding card application layer, user access permission control is set up. Its cloud PC management server provides user account access permission control. Only verified users can connect to access the computing nodes. Network security isolation is also set up. The cloud PC server can include two isolated networks, one data center network, connected to the CPU, and the other is a terminal access network, connected to the encoding card. The two networks can be physically isolated, and data cannot be transmitted between the two networks. Data transmission encryption is also set up, using SSL / TLS data transmission encryption to protect all data transmission between the cloud PC server and the cloud PC user terminal / cloud PC soft terminal.

[0061] Furthermore, in the single-point monitoring and operation and maintenance of the encoding card application layer, single-node monitoring can be performed, the VGA image output status of the current computing node can be queried, and the VGA image output status can be synchronously set to the video acquisition status light, so as to display the video acquisition status in real time and perform software upgrades. It supports online upgrades of encoding card software, kernel, media programs and other software through the cloud PC management platform, supports local upgrades of encoding card software, kernel, media programs and other software, and supports local upgrades of individual modules of encoding card software, kernel and media programs.

[0062] In some embodiments, the data interaction of the encoding board includes: S110, the computing node (encoding card) reports the status to the cloud PC management platform after startup; S120, the cloud PC user terminal / soft terminal connects and sends a login request to the cloud PC management platform; S130, the cloud PC user terminal / soft terminal selects and connects the encoding card, and the cloud PC management platform forwards the connection request; S140, the computing node (encoding card) establishes a media channel with the cloud PC user terminal / soft terminal; S150, the computing node (encoding card) establishes a USBIP channel with the cloud PC user terminal.

[0063] In some embodiments, see Figure 2The Cloud PC Management Platform communicates with the chassis management software of the GPU version server through the UDP protocol to implement functions such as slot management, fan control, power management, chassis information collection, and alarm information processing. Specifically, the chassis management software obtains the motherboard information collected by the embedded controller (EC) through the I2C protocol, including data such as CPU, GPU, memory, hard disk status, and motherboard serial number and MAC address. At the same time, it also collects operating information such as power parameters and fan speed. Through the embedded controller, the chassis management software can control the power on and off operation of the motherboard, and the EC has control over the power button. See Fig.14 The cloud PC management platform of the present invention provides comprehensive management functions for high-performance GPU servers, including automatic identification of GPU devices and presentation on the platform; support for remote power management (power on, power off, forced shutdown, restart); implementation of CPU, memory, and disk status resource monitoring; and real-time monitoring of graphics card parameters (such as GPU model, etc.) to ensure efficient operation and management of the equipment.

[0064] See also Fig.14, the interaction process between the cloud PC management platform and the GPU version server includes the following steps: S201, the high-performance GPU server chassis management software sends a chassis registration message (chassis ID, chassis type) to the cloud PC management platform; S202, after receiving the registration, the cloud PC management platform returns a registration success message to the high-performance GPU server chassis management software; S203, the high-performance GPU server chassis management software sends a chassis heartbeat packet message to the cloud PC management platform; S204, the cloud PC management platform responds to the heartbeat packet message to the high-performance GPU server chassis management software; if the cloud PC management platform does not receive the high-performance GPU server chassis heartbeat packet, it will mark the chassis as powered off; S205, after monitoring the chassis board information, board in-place status, computing board power on and off status and other information, the high-performance GPU server chassis management software actively reports the chassis device status information (slot usage, board information, power on and off status, etc.) to the cloud PC management platform; S206, the cloud PC management platform can send a heartbeat packet to the high-performance GPU server chassis management software through the control interface The GPU server initiates controls such as fan speed, chassis power on and off, and computing node startup, shutdown, and restart; S207, the cloud PC management platform sends alarm threshold settings to the high-performance GPU server chassis management software; S208, when the high-performance GPU server chassis management software detects abnormalities such as power status failure, high chassis temperature, and fan failure, it actively reports alarm information to the cloud PC management platform; S209, after receiving the request, each cloud PC server chassis returns the fan speed information to the cloud PC management platform; S210, after receiving the reported alarm information, the cloud PC management platform parses the data and pushes it to the operation and maintenance management interface; S211, when the operation and maintenance administrator needs to locate the high-performance GPU server chassis, select the high-performance GPU server chassis in the operation and maintenance management interface and click to turn on the indicator light; S212, the cloud PC management platform sends a location positioning command to the selected high-performance GPU server chassis; S213, after receiving the positioning command sent by the cloud PC management platform, the high-performance GPU server chassis management software turns on the flashing indicator light.

[0065] In some embodiments, see Fig.15 In the user login access and desktop control process of the cloud PC service system of the present invention, the user can send a login request to the cloud PC management platform, connect to the computing node, establish an audio and video stream with the computing node, and then the cloud PC user terminal / soft terminal transmits the keyboard and mouse events to the virtual keyboard and mouse device according to the user operation, and the computing node (CPU) responds to the keyboard and mouse events, thereby achieving the desktop remote control effect. The process of cloud PC user login access to the desktop is divided into three stages: the cloud PC management platform login stage, the desktop list display stage and the desktop control stage.

[0066] Specifically, during the login to the cloud PC management platform, the user uses the cloud PC user terminal to turn on the computer or open the cloud PC soft terminal, enters the user account and password to log in, and the login process first establishes a connection with the cloud PC management platform and passes the token authentication. After the authentication is successful, the user can log in, otherwise an error code is returned. During the desktop list display stage, after the user successfully logs in, the cloud PC user terminal software will automatically query the user's assigned desktop list. After the desktop list is returned, subsequent business processing is performed based on the number of desktop lists. Among them, when there is no desktop, the cloud PC user terminal displays that the current account has no desktop resources assigned; when there is only one desktop, it automatically triggers the request to connect to the desktop action; when there are two or more desktops, the desktop list is displayed for the user to choose which desktop to connect to. The desktop control process includes: S310, the cloud PC user terminal initiates a request to the cloud PC management platform to connect to the desktop; S320, when the cloud PC management platform receives the request to connect to the desktop, it queries the information and status of the requested desktop, and forwards the request connection information to the computing node after the desktop is online; S330, the computing node accepts the request to connect and triggers the multimedia service; S340, the cloud PC management platform registers the connection relationship of the desktop request, and forwards the received return information to the cloud PC user terminal; S350, the cloud PC user terminal receives the response information of the connection, and also triggers the multimedia service to realize the remote control of the desktop.

[0067] In some embodiments, the cloud PC service system of the present invention supports multi-machine dual-screen mode. After logging into the cloud PC user terminal interface, the user can configure two displays to display the screens of different computing nodes, and support keyboard and mouse cross-screen operations to improve multi-tasking capabilities and work efficiency. At the same time, it supports one-way copying of clipboard and files (security and data copy direction are controlled by a one-way isolation network gate) to ensure the safe and reliable transmission of data between different computing nodes. Specifically, the cloud PC multi-machine dual-screen mode refers to the related technology of distributed keyboard and mouse cross-screen, and maps multiple hosts and multiple screens through a screen area. Among them, after connecting to a certain desktop, moving the mouse and keyboard input will send corresponding events to the corresponding computing node; when the mouse moves to the left and right edges of the screen, the currently controlled IP will be modified, and the mouse starting point position will be mirrored (that is, the mouse is transformed from the right edge to the left edge, or the left edge is transformed to the right edge); after the transformation is completed, the mouse starts to move within the current screen range from the starting point, and the mouse will be sent to the new computing node. Similarly, the keyboard input will also be sent to the computing node, and complete keyboard and mouse control will be achieved at this time. It should be noted that in the scenario of mapping multiple computing nodes on a screen area and multiple screens, only absolute coordinates can be used at present. However, there are differences in data length and format between absolute coordinates and relative coordinates. Therefore, the computing node (encoding card) needs to add report information of absolute coordinates when creating a virtual keyboard and mouse device.

[0068] In some embodiments, the cloud PC service system of the present invention uses an encoding card to transmit images in a high-definition and low-latency encoding manner. Specifically, additional chroma profile information is introduced in the YUV420 format encoding process, and layered encoding is performed. The YUV420 format image is used as the basic encoding layer, and the U and V profile information is used as the enhancement layer encoding. During decoding, this information is used to reconstruct higher quality chroma components, approaching the effect of YUV444. See Fig.16 The YUV transmission step at least includes: S410, YUV444 downsampling to YUV420, encoding the image in YUV420 format as the basic coding layer, synchronously extracting chroma profile information from the YUV444 format, combining the chroma profile information into an image in YUV420 format, and encoding it as an enhancement layer; wherein, PTS is used between layers for time synchronization; S420, the base layer and the enhancement layer are respectively sent to the decoding end as independent coding streams; S430, the decoding end receives a base layer code stream and multiple enhancement layer code streams, and decodes them; wherein, the base layer code stream is decoded to generate a YUV420 base image, and the enhancement layer code stream is decoded to generate a corresponding YUV420 color enhanced image; S440, the base image uses PTS to find the corresponding color enhanced image, and synthesizes the final YUV444 image.

[0069] The cloud PC service system of the present invention adopts a coding card to transmit images in a high-definition and low-latency encoding manner, and adopts YUV420 color space to approximate the image color effect of YUV444, thereby optimizing the encoding efficiency and reducing the complexity of encoding while maintaining the fluency of decoding; the traditional SVC encodes the same data of the same frame of image in different layers, while the present technology uses the same frame of image with different data layers, and the basic image itself is not layered, thereby maximizing the image clarity; the enhancement layer uses the contour information of the YUV444 image color, which is further quantized to greatly reduce the amount of information in the color space, and is converted to YUV420 compression, which significantly improves the compression ratio and reduces the transmission bandwidth of the enhancement layer; different layers of code streams are transmitted independently without any coupling. Therefore, different layers can use different QoS transmission strategies. Usually, the basic layer can choose a strategy with better QoS for transmission. According to bandwidth conditions, the enhancement layer code stream can be selectively encoded and transmitted, the basic layer can be decoded and displayed separately, or the enhancement layer can be selected for color enhancement and converted to YUV444 display. If the enhancement layer is selected for color enhancement, the U / V contour information can be used separately or simultaneously according to the color to be enhanced or bandwidth conditions. When in use, the client's media uses low latency (multi-band) high fidelity, while the soft terminal's media uses low bandwidth (full screen) high fidelity, so that customers can get the best high-fidelity image experience in different bandwidth environments and different terminals.

[0070] In an application embodiment, the image YUV444 to image YUV420 processing of the present invention is obtained by the following calculation:

[0071] The Y component remains unchanged:

[0072]

[0073] Where (x, y) is the pixel coordinate in the image, Represents the pixel coordinates of the YUV420 image, Represents the pixel coordinates of the YUV444 image.

[0074] Downsampling of U and V components (calculating an average value for each 2x2 block of pixels):

[0075]

[0076]

[0077] Where ((i, j)) is the pixel coordinate of the YUV420 image, ((2i, 2j)), ((2i+1, 2j)), ((2i, 2j+1)), ((2i+1, 2j+1)) are the four pixel coordinates corresponding to the YUV444 image.

[0078] In an application embodiment, the acquisition of chromaticity profile information of the present invention includes the steps of:

[0079] S510, calculate chromaticity profile information. Specifically, the difference between the original chromaticity component of YUV444 and the chromaticity component of YUV420 obtained by downsampling is used as the chromaticity profile information to be calculated, and is defined as UResidual and VResidual respectively. For each pixel block, the chromaticity profile between the U and V components of YUV444 and the U and V components obtained by downsampling needs to be calculated, and the calculation is as follows:

[0080]

[0081]

[0082] Wherein, (U_{}(i, j)) is the corresponding YUV420 U component obtained by downsampling, and (V_{}(i, j)) is the corresponding YUV420 V component obtained by downsampling.

[0083] S520, chromaticity profile information quantization, which is expressed as follows:

[0084]

[0085]

[0086] S530, converting chromaticity profile information to YUV420. Specifically, the chromaticity profile data is converted to YUV420, and grayscale image encoding of width and height may be used, or image encoding of width and height of 2*width / 3 and 2*height / 3 may be used.

[0087] In an application embodiment, the present invention encodes and transmits the base layer and the enhancement layer respectively. Specifically, in the encoding and transmission of the base layer, the YUV420 base image is passed to the encoder for standard YUV420 encoding and written into PTS, and the encoded video stream is stored or transmitted in a conventional manner. In the encoding and transmission of the enhancement layer, the YUV420 chroma profile image is passed to the encoder for standard YUV420 encoding and written into PTS, and the encoded video stream is stored or transmitted in a conventional manner.

[0088] In one application embodiment, in the decoding stage of the present invention, the decoding of the enhancement layer uses a standard decoder to decode the enhancement layer video stream to obtain a YUV420 chroma contour image, while the decoding of the base layer uses a standard decoder to decode the base layer video stream to obtain a YUV420 base image, and according to the PTS value, finds the enhancement frame corresponding to the current base frame, and extracts the corresponding chroma contour information.

[0089] Furthermore, in the process of reconstructing the YUV444 image, contrary to calculating the chroma profile information, the U and V components in the YUV444 image obtained by upsampling the chroma profile information are used, and the restored U and V components are combined with the Y component to reconstruct a high-quality image close to the YUV444 format. When converting from YUV420 back to YUV444, the U and V components need to be interpolated and reconstructed. It can be understood that the common bilinear interpolation method can be used to restore the U and V components with higher resolution.

[0090] Specifically, for each pixel of U and V, interpolation calculation is performed using the surrounding U and V values.

[0091] Among them, the calculation formula of U component is as follows:

[0092]

[0093]

[0094]

[0095] .

[0096] Among them, the calculation formula of U component is as follows:

[0097]

[0098]

[0099]

[0100] .

[0101] In some embodiments, the cloud PC service system uses VOP (Video Output Processor) overlay processing to replace traditional logical overlay processing and reduce user terminal costs. For example, two VOPs are output on RK3588 / RK3576, corresponding to three scenarios (single-screen display, dual-screen different display, dual-screen same display).

[0102] Specifically, in a single-screen display scenario, that is, only the main screen or the secondary screen outputs a scenario, the main screen uses HDMI to output a 1080P@60Hz RGB image, including Layer 0 (Primary) for basic display, Layer 2 (UI) for user interface display, and Layer 3 (Cursor) for the mouse cursor. It displays or hides the user interface by controlling the enabled state of Layer 2; the secondary screen uses MIPI to output a 1080P@60Hz RGB image, including Layer 0 (Primary) for basic display, Layer 2 (UI) for user interface display, and Layer 3 (Cursor) for the mouse cursor. It displays or hides the user interface by controlling the enabled state of Layer 2.

[0103] In the dual-screen different display scenario, that is, dual-screen different display, a main screen and a secondary screen are defined. The main screen uses HDMI to output 1080P@60Hz RGB images, including Layer 0 (Primary) for basic display, Layer 2 (UI) for user interface display, and Layer 3 (Cursor) for the mouse cursor; the secondary screen uses MIPI to output 1080P@60Hz RGB images, including Layer 0 (Primary) for basic display, and Layer 2 (Cursor) for the mouse cursor.

[0104] In the dual-screen co-display scenario, that is, dual-screen co-display, the basic images and mouse of the two screens display the same content, the primary screen uses HDMI to output 1080P@60Hz RGB images, including Layer 0 (Primary) for basic display, Layer 2 (UI) for user interface display, and Layer 3 (Cursor) for the mouse cursor; the secondary screen uses MIPI to output 1080P@60Hz RGB images, including Layer 0 (Primary) for basic display, and Layer 2 (Cursor) for the mouse cursor.

[0105] In an application embodiment, see Fig.17 In the overlay mechanism adopted by the present invention, Weston and VPX transmit images to different Planes through DRM, and then VOP transmits them to different display interfaces. The GPU of FPGA or SOC collects images from the interface, overlays them, and then transmits them to the display. The new VOP of the present invention adopts a unified display architecture, shares the Plane resources inside the VOP, and supports layer overlay of the Plane. Among them, the property settings of the Plane include: specifying which CRTCs the Plane is shared with; in the Plane type setting, 0 is Overlay, 1 is Primary, and 2 is Cursor; the overlay level of the Plane can be modified in real time, ranging from 0 to 7, the smaller the value, the earlier the overlay; specifying the color transparency of the Plane can be modified in real time, 0 means closed; specifying the transparency of the Plane can be modified in real time, ranging from 0 to 65535.

[0106] In an application embodiment, the present invention dynamically configures and implements three scenarios (single-screen display, dual-screen different display, dual-screen same display) on RK3588 / RK3576, using the Wayland / Weston framework to implement the three scenarios. The three scenarios can be dynamic, thereby facilitating dynamic settings at the application layer.

[0107] Specifically, in the login scene switching, different overlay scenes can be met by setting the properties of Plane in real time. Among them, when the system is turned on or logged in, the application is displayed, and the VPX video stream is not displayed. The property settings of Plane are shown in Table 1; when the login is completed, the application is not displayed, and the VPX video stream is displayed, then the level of Weston Plane is adjusted, and the property settings of Plane are shown in Table 2; when it is necessary to display the VPX video stream and the sidebar or pop-up box of the application at the same time, the transparent color of Weston Plane is adjusted, and the property settings of Plane are shown in Table 3; if a watermark needs to be displayed, the transparent color and transparency of WestonPlane are adjusted, and the property settings of Plane are shown in Table 4;

[0108] Plane zpos colorkey alpha Weston main screen Esmart0-win0 6 0 65535 Weston sub-screen Esmart2-win0 7 0 65535 VPX Main Screen Cluster0-win0 4 0 65535 VPX secondary screen Esmart3-win0 5 0 65535

[0109] Table 1

[0110] Plane zpos colorkey alpha Weston main screen Esmart0-win0 2 0 65535 Weston sub-screen Esmart2-win0 3 0 65535 VPX Main Screen Cluster0-win0 4 0 65535 VPX secondary screen Esmart3-win0 5 0 65535

[0111] Table 2

[0112] Plane zpos colorkey alpha Weston main screen Esmart0-win0 6 1234 65535 Weston sub-screen Esmart2-win0 7 1234 65535 VPX Main Screen Cluster0-win0 4 0 65535 VPX secondary screen Esmart3-win0 5 0 65535

[0113] Table 3

[0114] Plane zpos colorkey alpha Weston main screen Esmart0-win0 6 1234 3000 Weston sub-screen Esmart2-win0 7 1234 3000 VPX Main Screen Cluster0-win0 4 0 65535 VPX secondary screen Esmart3-win0 5 0 65535

[0115] Table 4

[0116] Specifically, in the use scenario switching, in the single-screen output scenario, the current application is developed using QT, and QT has the function of monitoring the number of screens. When the output screen sends changes: when there is only HDMI-A-1, the application requests VPX to display the main screen video on Cluster0-win0; when there is only HDMI-A-2, the application requests VPX to display the main screen video on Cluster0-win0; when there is no display output, the application pauses requesting the main screen video. In the dual-screen different display scenario, when QT monitors that there are two screens, the application requests VPX to display: the main screen video on Cluster0-win0; the secondary screen video on Esmart3-win0. In the dual-screen same display scenario, when the user selects this scenario, it is necessary to modify the environment variable to set weston to mirror mode and restart weston, and then the application only applies for the main screen video. When VPX is initialized, the fd of the same drm is bound to two fb_ids, and then the fb_id is bound to Cluster0-win0 and Esmart3-win0.

[0117] In some embodiments, the USB data transmission of the cloud PC service system adopts the bridge USBIP mode, the encoding board and the cloud PC communicate with each other using the USB bus, and the USB Device side of the encoding board no longer appears as a USB network card, but as a custom USB data receiving / transmitting device. Figure 1, when the USBIP client on the cloud PC side needs to connect to the 3240 port of the USBIP server on the decoding side: the USBIP client first connects to the local 3240 port; the transmission agent on the cloud PC side and the transmission agent on the encoding side establish a dedicated virtual channel; the transmission agent on the encoding side connects to the 3240 port on the decoding side. The above-mentioned transmission agent on the encoding side and the transmission agent on the cloud PC side act as a bridge for the communication between the USBIP client on the cloud PC side and other network objects, where the transmission agent on the cloud PC side is a service program, which not only drives the custom USB Device device driver, but also acts as an agent for the network data reception and transmission between the USBIP client and other network objects, while the custom USBDevice driver on the encoding side is a character driver device, which is only responsible for USB data reception / transmission.

[0118] Specifically, the transmission agent on the encoding side reads the data packet sent from the USB bus through a custom USB Device driver, and then forwards it to the USBIP server on the decoding side or the local enc_client process through the network; similarly, after the USBIP server or enc_client receives the data packet, it also sends the data packet to the USB bus through the custom USB Device driver. In the above process, the USBIP server on the decoding side does not need to change the code; but the USBIP client on the cloud PC side needs to change the target address to the local IP address when connecting to the network; and the USBIP client on the cloud PC side can only act as a client and actively initiate a connection, but cannot monitor the connection.

[0119] The above is only a preferred embodiment of the present invention. The present invention is not limited to the above implementation. As long as the technical effect of the present invention is achieved by the same means, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of protection of the present invention. Within the scope of protection of the present invention, its technical scheme and / or implementation method may have various modifications and changes.

Claims

1. A cloud PC service system, characterized in that: It includes a power module, a management module and a plurality of computing nodes arranged in a machine frame; the management module is connected to the power module and each computing node through an I2C interface to read the working status of the power module and each computing node, and report to the cloud PC management platform through a network interface; The computing node includes a graphics card, a computing board and a coding board; the graphics card is connected to the computing board via a PCIe extension cable; the coding board is provided with at least 2 DP interfaces and at least 1 HDMI interface, and the coding board is connected to the graphics card via 2 DP interfaces or 1 HDMI interface.

2. The cloud PC service system according to claim 1, characterized in that: A chassis management software is also provided, wherein the architecture of the chassis management software is provided with an application layer, an interface layer and a system layer; The application layer includes a GPU computing node for slot number detection, power-on and power-off status detection, temperature detection, presence detection, power-on and power-off control, and resource usage detection, as well as a module for chassis ID reporting and abnormal alarm management, as well as a fan module and a power supply module; The interface layer includes an i2c_client port for accessing the i2c interface to communicate with the I2C of each node EC single-chip computer, and a Socket_api port for communicating with the cloud PC management platform via the udp interface; The system layer includes the Linux operating system.

3. The cloud PC service system according to claim 2, characterized in that: The software architecture of the encoding board includes an encoding card system layer, an encoding card interface layer and an encoding card application layer; the encoding card interface layer includes a USBoIP interface library and an embedded media library; the encoding card application layer mainly provides the ability for PC to access the cloud through the GPU computing node-encoding card software architecture, and its functions include user management, computing resource management, network and data security, and single-node monitoring and operation and maintenance.

4. The cloud PC service system according to claim 3, characterized in that: The image is transmitted by high-definition and low-delay coding. In the YUV420 format encoding process, additional chroma profile information is introduced for layered coding. The YUV420 format image is used as the basic coding layer, and the U and V profile information is used as the enhancement layer. During decoding, the high-quality chroma components are reconstructed through the above information to approach the effect of YUV444. Among them, in the encoding and transmission of the base layer, the YUV420 basic image is passed to the encoder for standard YUV420 encoding and written into PTS, and the encoded video stream is stored or transmitted in a conventional manner; in the encoding and transmission of the enhancement layer, the YUV420 chroma profile image is passed to the encoder for standard YUV420 encoding and written into PTS, and the encoded video stream is stored or transmitted in a conventional manner; Among them, during the decoding process, the decoding of the enhancement layer uses a standard decoder to decode the enhancement layer video stream to obtain a YUV420 chroma contour image; the base layer decoding uses a standard decoder to decode the base layer video stream to obtain a YUV420 base image, and according to the PTS value, find the enhancement frame corresponding to the current base frame, and extract the corresponding chroma contour information.

5. The cloud PC service system according to claim 4, characterized in that: The YUV transmission step at least includes: S410, YUV444 downsampled to YUV420, the image encoding in YUV420 format is used as the basic encoding layer, the chroma profile information is synchronously extracted from the YUV444 format, the chroma profile information is combined into an image in YUV420 format, and encoded as an enhancement layer; wherein, PTS is used for time synchronization between layers; S420, the base layer and the enhancement layer are respectively sent to the decoding end as independent coded streams; S430, the decoding end receives a base layer code stream and multiple enhancement layer code streams, and decodes them; wherein the base layer code stream is decoded to generate a YUV420 basic image, and the enhancement layer code stream is decoded to generate a corresponding YUV420 color enhanced image; S440, the base image uses PTS to find the corresponding color enhancement image and synthesize the final YUV444 image.

6. The cloud PC service system according to claim 2, characterized in that: VOP overlay is used to process single-screen display scenarios, dual-screen different display scenarios and dual-screen same display scenarios; Weston and VPX transmit images to different Planes through DRM, and then VOP transmits them to different display interfaces; the GPU of FPGA or SOC collects images from the interface, overlays them and transmits them to the display; VOP adopts a unified display architecture, shares the Plane resources inside VOP, and supports layer overlay on Plane.

7. The cloud PC service system according to claim 2, characterized in that: USB data transmission adopts the bridge USBIP method; among them, when the USBIP client on the cloud PC side needs to connect to the 3240 port of the USBIP server on the decoding side, the USBIP client first connects to the local 3240 port, and the transmission agent on the cloud PC side will establish a dedicated virtual channel with the transmission agent on the encoding side, and the transmission agent on the encoding side is connected to the 3240 port on the decoding side.

8. The cloud PC service system according to claim 1, characterized in that: The network ports of the encoding board and the management module are both used to connect to the user terminal access network, and the network port of the computing board is used to connect to the business data network.

9. The cloud PC service system according to claim 1, characterized in that: It also includes a backplane for connecting the power module, the management module and a plurality of computing nodes, and the backplane is arranged on the top of the frame.

10. The cloud PC service system according to claim 1, characterized in that: The power board of the computing node is provided with a Standby power supply for supplying power to the encoding card so as to maintain the power output when the computer is shut down.

Images