A network address configuration system and method for a computing power module cluster

Abstract

The application discloses a kind of network address configuration system and method of computing power module cluster, the method includes: management server writes the network address information input by user into the storage device in the i-th computing power module, to realize reading and writing operation to the storage device, complete the network address modification of the i-th computing power module;When detecting that CTL [i] Signal is low level, the i-th computing power module connects third storage passage and first storage passage;When detecting that PWR [i] Signal is high level, the i-th computing power module starts module power supply, at this time, the i-th computing power module starts running with new network address;The above steps are repeatedly executed, i.e. the network address configuration of user to all computing power modules can be realized.The application improves the efficiency of computing power module cluster network address configuration, and the method has very high reliability, and still can be carried out when computing power module is in shutdown state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network address configuration, and particularly relates to a network address configuration system and method for a computing power module cluster. BACKGROUND

[0002] A computing power module refers to a hardware module capable of providing computing power. Multiple computing power modules can be combined into a computing power module cluster. The computing power module cluster is deployed in the cloud and can provide online computing power resources for multiple users at the same time.

[0003] A single computing power module is usually composed of a core operation chip and various peripheral circuits. The core operation chip can be a CPU, a GPU, a DSP, an FPGA, an artificial intelligence dedicated chip, etc. The computing power module usually runs an operating system, such as Linux. The computing power module can also run some small real-time operating systems, such as RT-Thread, QNX, UCOS, etc.

[0004] The computing power module usually runs a network protocol stack to access the Ethernet. A network address is a logical address that a node on the Ethernet has in the network and can be used to address the node. An IP address is a way of addressing a host on the Ethernet. Each computer is assigned a logical address, so that the computer can be identified and information can be shared.

[0005] The computing power module is usually a small embedded computing module. Multiple computing power modules can be integrated into a chassis and managed by a server to form a computing power module cluster. The server that manages these computing power modules is called a management server.

[0006] A computing power module cluster usually includes multiple computing power modules. When the network address of the computing power module needs to be initialized or reconfigured, there are two traditional methods. The first method is to change the network address configuration file of the computing power module after remotely logging into the computing power module. However, this method has certain limitations. For example, the computing power module is in a shutdown state or does not support remote login (some computing power modules run some lightweight operating systems and do not have remote login functions), so this method cannot be used to change the network address. In addition, if the number of computing power modules is large, manually configuring the network address of the computing power module is extremely inefficient.

[0007] The second method is to run a service program in the computing power module and run a management program on the management server of the computing power module, the management program sends a command to the service program of the computing power module through the network, and the service program modifies according to the command. This method has certain problems in reliability: for example, when the service program in the computing power module is abnormally stopped and cannot be automatically restarted, the configuration cannot be performed; for another example, when the computing power module is in a shutdown state, this method cannot be used to change. In addition, this method requires that the network of the management server and the computing power module be in the same local area network, which has great limitations. SUMMARY

[0008] Therefore, the application provides a network address configuration system and method of a computing power module cluster to solve the above technical problems.

[0009] The application discloses a network address configuration system of a computing power module cluster, which comprises a management server, a control module and a computing power module cluster connected in sequence.

[0010] The management server is configured to send a control command to the control module according to a user's network address configuration requirement of the computing power module, so as to set a PWR[i] signal to a low level or a high level, so that the i th computing power module determines whether the power supply needs to be turned off; set a CTL[i] signal to a low level or a high level, so that the i th computing power module determines whether different storage paths need to be connected; and connect with a USB[i] signal, so that the i th computing power module is read and written to complete the network address configuration of the i th computing power module; wherein the PWR[i] signal, the CTL[i] signal and the USB[i] signal are i th control signals respectively, and the value range of i is 1 to N; N is a positive integer greater than 1.

[0011] The control module is configured to generate three groups of control signals according to the received control command and output to the N computing power modules; the number of each group of control signals is N, and the N control signals of each group correspond to the N computing power modules one by one; the three groups of control signals are respectively denoted as PWR[1:N] signals, CTL[1:N] signals and USB[1:N] signals.

[0012] Further, the management server uses only one USB to generate one USB signal.

[0013] The control module comprises a USB_HUB module, and a control unit and a gate connected with the USB_HUB module respectively.

[0014] The USB_HUB module is configured to receive the USB signal sent by the management server, and expand the USB signal into a control USB signal and a data USB signal, and input the control USB signal and the data USB signal into the control unit and the gate, respectively.

[0015] The control unit is configured to generate a PWR[i] signal, a CTL[i] signal and a USB[i] signal.

[0016] The gate is controlled by the control unit, and is configured to output the USB[i] signal to the i-th computing power module.

[0017] Further, each computing power module is composed of a computing power chip, a USB-storage reading and writing device, a path selector and a storage device.

[0018] The computing power chip is configured to perform a required computing task.

[0019] The storage device is configured to store a system image required when the computing power chip is running, and the system image includes a device initialization program, an operating system and a network address configuration file.

[0020] The USB-storage reading and writing device is configured to perform reading and writing operations on the storage device.

[0021] The path selector is configured to select whether the third storage path is connected to the first storage path or the second storage path according to the enablement of the CTL[1:N] signal.

[0022] Further, the first storage path is arranged between the computing power chip and the path selector, the second storage path is arranged between the USB-storage reading and writing device and the path selector, and the third storage path is arranged between the path selector and the storage device.

[0023] Further, the power supply of each computing power module is controlled by one of the PWR[1:N] signals, when one of the PWR[1:N] signals is at a high level, the computing power module corresponding to the PWR[1:N] signal is in a power-on state, and the power supply of the computing power module is turned on; when one of the PWR[1:N] signals is at a low level, the computing power module corresponding to the PWR[1:N] signal is in a power-off state, and the power supply of the computing power module is turned off.

[0024] The path selector in each computing power module is controlled by one of the CTL[1:N] signals, when one of the CTL[1:N] signals is at a high level, the third storage path and the second storage path are connected; when one of the CTL[1:N] signals is at a low level, the third storage path and the first storage path are connected.

[0025] When the third storage path and the second storage path are connected, the management server reads and writes the storage device through the USB signal; when the third storage path and the first storage path are connected, the computing power chip is used to load the system image in the storage device and normally start.

[0026] The application further discloses a network address configuration system for the computing power module cluster.

[0027] Step 1: when the PWR[i] signal is detected as low by the i-th computing power module, the power supply of the i-th computing power module is turned off;

[0028] Step 2: when the CTL[i] signal is detected as high by the i-th computing power module, the third storage path and the second storage path are connected;

[0029] Step 3: when the PWR[i] signal is detected as high by the i-th computing power module, the power supply of the computing power module is turned on;

[0030] Step 4: the management server writes the network address information input by the user into the storage device in the i-th computing power module, so as to read and write the storage device, and the network address modification of the i-th computing power module is completed;

[0031] Step 5: when the CTL[i] signal is detected as low by the i-th computing power module, the third storage path and the first storage path are connected;

[0032] Step 6: when the PWR[i] signal is detected as high by the i-th computing power module, the power supply of the i-th computing power module is turned on, and the i-th computing power module starts to run with the modified network address;

[0033] Step 7: steps 1 to 6 are repeatedly executed, so that the network address configuration of all computing power modules required by the user can be performed.

[0034] Further, before the step 1, the method further comprises:

[0035] The user determines the serial number of the computing power module to be modified as i, and determines the network address information to be modified; after the user inputs the serial number of the computing power module to be modified and the network address information to be modified into the management server, the management server sends a control command to the control module, and sets the PWR[i] signal to low.

[0036] Further, after the step 1 and before the step 2, the method further comprises:

[0037] The management server sends a control command to the control module, and sets the CTL[x] signal to high.

[0038] Further, after the step 2 and before the step 3, the method further comprises:

[0039] The management server sends a control command to the control module to enable the USB signal and the USB[i] signal;

[0040] The management server sends a control command to the control module to set the PWR[i] signal to high.

[0041] Further, after the step 4 and before the step 5, the method further comprises:

[0042] After the management server completes the modification of the address, the PWR[i] signal is set to low;

[0043] When the i-th computing power module detects that the PWR[i] signal is low, the power supply of the i-th computing power module is turned off;

[0044] The management server sends a control command to the control module to set the CTL[i] signal to low.

[0045] Due to the adoption of the above technical solutions, the present application has the following advantages: the management server only uses 1 USB interface, and can configure the network address of multiple computing power modules, thereby improving the efficiency of the network address configuration of the computing power module cluster, and the method has very high reliability and can still be performed when the computing power module is in a shutdown state. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0047] Figure 1 It is a whole schematic diagram of a network address configuration system of a computing power module cluster of the embodiment of the present application.

[0048] Figure 2 It is a schematic diagram of the composition of the control module of the embodiment of the present application.

[0049] Figure 3 It is a schematic diagram of the composition of the computing power module of the embodiment of the present application.

[0050] Figure 4 It is a flowchart of a network address configuration method of a computing power module cluster of the embodiment of the present application. DETAILED DESCRIPTION

[0051] With reference to the drawings and embodiments, it is obvious that the described embodiments are only a part of the embodiments of the present application, and are not all the embodiments. All other embodiments obtained by those skilled in the art shall belong to the scope of protection of the embodiments of the present application.

[0052] Referring to Figure 1 The present application provides an embodiment of a network address configuration system of a computing power module cluster, which comprises a management server, a control module and a computing power module cluster connected in sequence. Specifically,

[0053] The management server is used for process control and data control of network address configuration of the computing power module cluster.

[0054] The control module outputs specified CTL, PWR[1:N] signal values according to the control command of the management server, and selects the USB signal to different computing power modules.

[0055] The computing power module cluster is composed of N computing power modules 1-N.

[0056] The computing power module is composed of ARM, GPU, CPU and other chips as core chips, and is used for performing high-performance computing or performing artificial intelligence training and inference tasks.

[0057] The connection signal between the management server and the control module is a USB signal.

[0058] The connection signal between the control module and the computing power module includes CTL[1:N] signal, USB[1:N] signal and PWR[1:N] signal.

[0059] CTL[1:N] signal is a signal set of CTL[1-N], and there are N in total. Among them, CTL[1] signal is connected to computing power module [1], CTL[2] signal is connected to computing power module [2], and so on, CTL[N] signal is connected to computing power module N.

[0060] PWR[1:N] signal is a signal set of PWR[1-N], and there are N in total. Among them, PWR[1] signal is connected to computing power module [1], PWR[2] signal is connected to computing power module [2], and so on, PWR[N] signal is connected to computing power module N.

[0061] USB[1:N] signal is a signal set of USB[1-N], and there are N in total. Among them, USB[1] signal is connected to computing power module [1], USB[2] signal is connected to computing power module [2], and so on, USB[N] signal is connected to computing power module N.

[0062] The management server should generally have the following features:

[0063] 1. User can use the management server to record, modify, configure the network address of all computing power modules.

[0064] 2. The management server can control the control module by sending control commands through the USB signal, including: controlling the high and low levels of the PWR[1:N] signal of the computing power module 1-N; controlling the high and low levels of the CTL signal of the computing power module 1-N; selecting the USB signal to the computing power module 1-N.

[0065] 3. When the control module selects the USB signal to any signal path in the USB[1] signal-USB[N] signal, the management server can read and write the storage device mounted on the USB link.

[0066] Referring to Figure 2 , the control module should generally have the following features:

[0067] 1. It usually contains a USB_HUB module inside, which expands the external USB signal into control USB signal and data USB signal. The control USB signal is connected to the control unit. The data USB signal is connected to the selector, which is controlled by the control unit and can be selected to any USB signal path in the USB[1] signal-USB[N] signal.

[0068] 2. The control module is connected to the management server through the USB, receives the command of the management server, and controls the output level of the CTL[1:N] signal and the PWR[1:N] signal.

[0069] 3. The control module receives the command of the management server and can select the data USB signal to any signal path in the USB[1] signal-USB[N] signal. Since the data USB signal is expanded from the external USB signal, it is equivalent to selecting the external USB signal to any signal path in the USB[1] signal-USB[N] signal.

[0070] Referring to Figure 3 , the computing power module is composed of computing power chips, USB-storage device read-write equipment, path selectors, and storage devices inside, and includes PWR, USB, and CTL outside.

[0071] The computing power chip is used to perform computing tasks, such as scientific computing, artificial intelligence model reasoning, etc.

[0072] The storage device is used to store the system image required by the computing power chip during operation, which includes device initialization program, operating system, network address configuration file, etc.

[0073] The USB-storage reading and writing device is used for reading and writing the storage device, and the computing power cluster management server can be connected to the module through the USB bus, so as to read and write the storage device.

[0074] The path selector selects the third storage path to be connected to the first storage path or the second storage path according to the enablement of the CTL signal.

[0075] The features that the computing power module should generally have include:

[0076] 1. The power supply of the computing power module is controlled by the PWR[1:N] signal, and when the PWR[1:N] signal is high, the computing power module is in a power-on state; and when the PWR[1:N] signal is low, the computing power module is in a power-off state.

[0077] 2. The path selector in the computing power module is controlled by the CTL signal, and when the CTL signal is high, the third storage path and the second storage path are connected. When the CTL signal is low, the third storage path and the first storage path are connected.

[0078] 3. When the third storage path and the second storage path are connected, the external management server can read and write the storage device through the USB signal. When the third storage path and the first storage path are connected, the computing power chip can load the system image in the storage device and start normally.

[0079] Referring to Figure 4 , the present application provides an embodiment of a network address configuration method of a computing power module cluster. It should be noted that the embodiment and the above-mentioned embodiment are mutually referenced.

[0080] The method in the embodiment includes:

[0081] S1: When the PWR[i] signal is detected to be low by the i-th computing power module, the power supply of the i-th computing power module is turned off;

[0082] S2: When the CTL[i] signal is detected to be high by the i-th computing power module, the third storage path and the second storage path are connected;

[0083] S3: When the PWR[i] signal is detected to be high by the i-th computing power module, the power supply of the computing power module is turned on;

[0084] S4: The management server writes the network address information input by a user into the storage device in the i-th computing power module, so as to read and write the storage device, and completes the network address modification of the i-th computing power module;

[0085] S5: When the CTL[i] signal is detected to be low by the i-th computing power module, the third storage path and the first storage path are connected.

[0086] S6: When the i-th computing power module detects that the PWR[i] signal is high, turn on the power supply of the i-th computing power module, at this time the i-th computing power module starts running with the modified network address;

[0087] S7: Repeat S1-S6, that is, the network address configuration of all computing power modules required by the user can be performed.

[0088] Before S1, the method further comprises:

[0089] The user determines the serial number of the computing power module to be modified as i, and determines the network address information to be modified; after the user inputs the serial number of the i-th computing power module and the network address information to be modified into the management server, the management server sends a control command to the control module to set the PWR[i] signal to low. For example, for the computing power module using Ethernet as the network protocol, the network address information includes MAC address, IP address, subnet mask, gateway address, DNS address and the like.

[0090] After S1 and before S2, the method further comprises:

[0091] The management server sends a control command to the control module to set the CTL[x] signal to high.

[0092] After S2 and before S3, the method further comprises:

[0093] The management server sends a control command to the control module to gate the USB signal and the USB[i] signal;

[0094] The management server sends a control command to the control module to set the PWR[i] signal to high.

[0095] After S4 and before S5, the method further comprises:

[0096] After the management server completes the address modification, the PWR[i] signal is set to low;

[0097] When the i-th computing power module detects that the PWR[i] signal is low, turn off the power supply of the i-th computing power module;

[0098] The management server sends a control command to the control module to set the CTL[i] signal to low.

[0099] It is to be explained that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit it, and although the present application has been described in detail with reference to the above embodiments, those ordinarily skilled in the art should understand that the specific embodiments of the present application can be modified or equivalently replaced without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A method for configuring network addresses in a computing module cluster, characterized in that, The network address configuration system of the computing power module cluster comprises a management server, a control module and a computing power module cluster connected in sequence; the computing power module cluster comprises N computing power modules; the management server is configured to issue a control command to the control module according to a user's network address configuration requirement for the computing power module, so as to set a PWR[i] signal to low or high, so that the i-th computing power module determines whether to turn off its power supply; so as to set a CTL[i] signal to low or high, so that the i-th computing power module determines whether to connect different storage paths; so as to connect with a USB[i] signal, so as to read and write the i-th computing power module, and complete the network address configuration of the i-th computing power module; wherein the PWR[i] signal, the CTL[i] signal and the USB[i] signal are respectively the i-th control signal, and the value range of i is 1 to N; N is a positive integer greater than 1; the control module is configured to generate three groups of control signals according to the received control command, and output to N computing power modules; the number of each group of control signals is N, and the N control signals of each group correspond to the N computing power modules one by one; the three groups of control signals are respectively denoted as PWR[1:N] signal, CTL[1:N] signal and USB[1:N] signal; The method comprises: The user determines that the computing power module serial number to be modified is i, and determines the network address information to be modified; after the user inputs the computing power module serial number to be modified and the network address information to be modified into the management server, the management server sends a control command to the control module to set the PWR[i] signal to low; Step 1: when the i-th computing power module detects that the PWR[i] signal is low, turn off the power supply of the i-th computing power module; Step 2: when the i-th computing power module detects that the CTL[i] signal is high, connect the third storage path and the second storage path; Step 3: when the i-th computing power module detects that the PWR[i] signal is high, turn on the power supply of the computing power module; Step 4: the management server writes the network address information input by the user into the storage device in the i-th computing power module, so as to read and write the storage device, and complete the network address modification of the i-th computing power module; Step 5: when the i-th computing power module detects that the CTL[i] signal is low, connect the third storage path and the first storage path; Step 6: when the i-th computing power module detects that the PWR[i] signal is high, turn on the power supply of the i-th computing power module, and at this time, the i-th computing power module starts running with the modified network address; Step 7: repeat steps 1 to 6 to configure the network address of all computing power modules required by the user.

2. The method of claim 1, wherein, After step 1 and before step 2, the method further comprises: The management server sends a control command to the control module to set the CTL[x] signal to high.

3. The method of claim 1, wherein, After step 2 and before step 3, the method further comprises: The management server sends a control command to the control module to select the USB signal and the USB[i] signal; The management server sends a control command to the control module to set the PWR[i] signal to high.

4. The method of claim 1, wherein, After the step 4 and before the step 5, the method further comprises: After the management server completes the address modification, the PWR[i] signal is set to low; When the i-th computing power module detects that the PWR[i] signal is low, the power supply of the i-th computing power module is turned off; The management server sends a control command to the control module to set the CTL[i] signal to low.

5. The method of claim 1, wherein, The management server uses only one USB to generate a USB signal; The control module comprises a USB_HUB module, and a control unit and a gate connected to the USB_HUB module, respectively; The USB_HUB module is configured to receive the USB signal sent by the management server, and expand the USB signal into a control USB signal and a data USB signal, which are input into the control unit and the gate, respectively; The control unit is configured to generate the PWR[i] signal, the CTL[i] signal and the USB[i] signal; The gate is controlled by the control unit and is configured to output the USB[i] signal to the i-th computing power module.

6. The method of claim 1, wherein, Each computing power module is composed of a computing power chip, a USB-storage reading and writing device, a path selector and a storage device; The computing power chip is configured to perform a required computing task; The storage device is configured to store a system image required by the computing power chip during operation; The system image comprises a device initialization program, an operating system and a network address configuration file; The USB-storage reading and writing device is configured to perform reading and writing operations on the storage device; The path selector is configured to select whether the third storage path is connected to the first storage path or the second storage path according to the enablement of the CTL[1:N] signal.

7. The method of claim 6, wherein, The first storage path is arranged between the computing power chip and the path selector, the second storage path is arranged between the USB-storage reading and writing device and the path selector, and the third storage path is arranged between the path selector and the storage device.

8. The method of claim 1, wherein, The power supply of each computing power module is controlled by one of the PWR[1:N] signals, and when one of the PWR[1:N] signals is high, the computing power module corresponding to the signal is in a powered-on state, and the power supply of the computing power module is turned on; when one of the PWR[1:N] signals is low, the computing power module corresponding to the signal is in a powered-off state, and the power supply of the computing power module is turned off; The path selector in each computing power module is controlled by one of the CTL[1:N] signals, and when one of the CTL[1:N] signals is high, the third storage path and the second storage path are connected; when one of the CTL[1:N] signals is low, the third storage path and the first storage path are connected; When the third storage path and the second storage path are connected, the management server reads and writes the storage device through the USB signal; when the third storage path and the first storage path are connected, the computing power chip is configured to load the system image in the storage device and start normally.

Images