Chinese summary of the Vega Prime distributed module

I. Application of Dr
1. First establish DR and Application Program
2. Dr creation includes setting the system Master/Slave and determining the virtual pipelines owned by the system
3. To implement an application working in Dr mode, you must add
Channel, pipelines, windows, pipeline ID
2. Establish Dr
Initialization requirements:
· The system is connected by a network
· The same version of Vega Prime is installed on each system.
· The master must have the permission to run the dr.
· Dr Establishment
· Enable MPI agent on each system
· Enable drutil on the Master System
· Check the configuration to ensure that the role allocation of the MPI agent is correct. If the role allocation of the relevant system is incorrect, modify it in the MPI agent.
· Assign a virtual monitor to each MPI agent.
· To verify the authenticity of the MPI agent configuration, perform the following operations: Test-> choose virtual monitors.
· Verify that slaver can find the Master application: Select "test path to application ".
· Enter "path to master's application" and press the "Start" button to start the test.
· If everything works properly, you can see the virtual monitor X on the blue screen. If there is a problem "Test Results", you will be given a question and modify the comment.
3. Create an application
VP supports active preview in the lynx prime panel to create a single-channel Dr application for Dr test.
1. Create a pipeline for each system
2. Give each pipeline a "pipeline ID"
3. Create a window for each pipeline and associate it with the Pipeline
4. Create a channel for each window and link it to the window
5. Add all channels to the observer.
Iv. Run Dr applications
5. Create Your Database
A typical application must load a lot of data, such as textures, models, and terrain. The operating system must be able to reference data correctly. There are two types of data to be obtained: local data, or remote access.
A. Local retrieval:
Each system stores a copy of data. The disadvantage of this method is that each copy needs to be updated when data is updated.
B. shared data: Shared data is stored in a central location and transmitted by the network to other systems. The advantage is that data is only updated once, and the disadvantage is that data acquisition is slow, you can map a network drive in the window to transmit data.

6. How does Dr work?
the DR system consists of one or more applications, one of which works on the master node, and the other works on the slave. The registry entry is used to locate the role of the application, if no setting is provided at the entrance, it does not matter which of the master and slave applications start to work first.
7. configuration, registry, and environment variable
configuration is a set of settings. Dr applies it to manage all its attribute settings (virtual pipeline, Master/slaver, in the registry, registry is a group of files containing the settings, and $ (mpi_install_default) \ config \ vegaprime \ REGISTRY \ agent exists. VRF. this file can be modified through drutil and MPI agent.
environment variables: Be careful when using environment variables. In all configure attribute settings of DR, you need to set the following environment variable attributes:
dr_role: "slave", or "master"
dr_group: Exp: All systems set to "lab1" work in the "lab1" Distribution Group.
Configuration: Set the environment variable name to the configuration name your application will use.
dr_ip_address: used to modify the IP address of the NIC used for Dr communication.
dr_app_root: set this environment as the drive letter (for example, "d:"). When slave looks for the master application, replace the path setting of the master with the environment variable.
the virtual pipelines is the abstraction of the physical pipeline (graphics card). It allows you to add, modify, and delete pipelines without modifying the physical pipeline.

8. How to enable and disable Dr
Dr Kai
A. The Master has virtual pipleline, and a pipeline is not included in the master application.
B. One of the systems is specified as slave.
Dr Guan
A. The master does not contain VDP definitions.
B. The master occupies all virtual pipelines in the application.
Turn Dr on
Create a master with VDP
Create an slave (s) containing virtual pipeline)
Run these two applications on master and slave.
9. Tools
Distributed rendering utilities is a tool used to establish, configure, and test MPI agents. Its window consists of the following parts.
MPI agents: each row in this window represents a running MPI agents, and the relevant attributes in this row can be modified through setup.

Tools Section
Setup: 1. Select the MPI agents object
2. Modify the attributes in setup.
3. Press the "Apply" key to update it.
Configuration:
On the left side of this tab, It is the configuration section, which can be built, renamed, and deleted.
* Configuration marked as [Default] is usually used. When an application is enabled, this default configuration will be used unless multiple configurations are included at the same time, you can also select a new configuration to replace the currently used default configuration.
Settings: each configuration contains related setting. The setting settings include the following parts:
Role: assign a role to an application. In slave, a new configuration must be created and assigned to slave. Therefore, a slave system contains at least two configurations, "Default master configuration ", and "Slave congfigurtion", slave MPI agent will use slave Configuration
Appliction root: sets the directory where the slave execution file is located. By default, slave downloads the master application according to the path where the master application is located, therefore, you must change the root directory of the path to the path where the slave application is located.
GROUP: used to restrict communication groups.
IP Address: set this attribute to, and Dr will use the first IP address it finds on the system. Generally, you can use this default setting, if multiple NICs are used in an application, you must specify the IP address before configuring the NIC.
Virtual pipeline: add, modify, and set virtual pipeline test. This tab is used to verify Dr settings and provides verification information to prompt possible problems.
1. Confirm the virtual pipeline you want to test.
2. verify whether the Server Load balancer can find the Master application. Select "test master application exists on slaves", and enter the path to master's applicaton and press start to start the test.
The processes tab is used to observe and stop remote processes.