JP5311473B2 - Computer system and re-installation method of CPU - Google Patents

Description

  The present invention relates to a computer system and a CPU reincorporation method, and more particularly to a computer system and a CPU reincorporation method in which, when a failure occurs in a CPU, the CPU is detached and then reincorporated.

  In a large-scale system such as a supercomputer, when a failure occurs in a CPU (Central Processing Unit), a process is performed in which the CPU is initialized after being disconnected from the system and incorporated into the system again.

  As related techniques, Patent Documents 1 to 5 disclose a method of re-installing a CPU when a CPU fault occurs and a technique for analyzing the contents of the fault.

Japanese Patent Laid-Open No. 02-129730 JP 06-051864 A JP 09-034852 A JP 09-128258 A Patent 2790204

  However, none of the related technologies described above discloses a failure recurrence prevention measure corresponding to the failure content of the CPU in which the failure has occurred. For this reason, there has been a problem that the same failure is likely to recur because the CPU has been initialized and re-installed without performing the failure recurrence prevention measure.

  The computer system according to the present invention is a computer system that performs re-installation after disconnecting the CPU when a failure occurs in the CPU, and analyzes the content of the failure that has occurred in the CPU. After performing the failure recurrence prevention treatment according to the analysis result, a diagnostic unit is provided for re-incorporating the CPU into the computer system.

  Further, the CPU re-installation method according to the present invention is a CPU re-installation method in which when a failure occurs in a CPU, the CPU is detached from the computer system and then re-installed. A step of analyzing a content of the failure, and a step of performing a failure recurrence prevention treatment according to the analysis result of the failure content.

  According to the present invention, when re-installing a CPU in which a failure has occurred, a computer system that reduces the possibility of re-occurrence of the re-integrated CPU by performing a failure relapse preventive measure according to the analysis result of the failure content. And a CPU re-installation method can be provided.

1 is a block diagram illustrating a configuration of a computer system according to a first embodiment. FIG. 6 is a flowchart showing an operation example of the computer system according to the first embodiment. FIG. 3 is a diagram for explaining a set voltage determination method according to the first embodiment.

Embodiment 1
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a computer system 1 according to the first embodiment. The computer system 1 includes a diagnostic processor 10, a power control unit 20, a clock control unit 30, a CPU 40_1 to CPU 40_n (hereinafter may be collectively referred to as a CPU 40), an MMU (Memory Management Unit) 70, a cooling device. 80.

  A diagnostic processor 10 as a diagnostic unit has embedded software 11. The diagnosis processor 10 diagnoses the CPU 40 and the MMU 70 through the diagnosis path. The power control unit 20 supplies power to the diagnostic processor 10, the CPU 40, the MMU 70, the cooling device 80, and the like. The clock control unit 30 controls a clock supplied to the diagnostic processor 10, the CPU 40, the MMU 70, the cooling device 80, and the like.

  The CPU 40_1 to CPU 40_n includes a temperature sensor 50_1 to a temperature sensor 50_n for measuring the temperature of each of the CPU 40_1 to CPU 40_n (hereinafter may be collectively referred to as the temperature sensor 50), and a BIST (Built In Self Test) to each of the CPU 40_1 to CPU 40_n. BIST 60_1 to BIST 60_n (hereinafter, may be collectively referred to as BIST 60) to be executed.

  As will be described later, the embedded software 11, when a failure occurs in the CPU 40, disconnects the CPU 40 from the system 1, analyzes the failure content, and performs failure recurrence prevention measures corresponding to the failure content. The embedded software 11 can change the voltage of the power supplied by the power control unit 20. Furthermore, the embedded software 11 can change the clock supplied by the clock control unit 30.

  For example, as a failure recurrence prevention measure, the embedded software 11 performs a measure of increasing the voltage of the power supply to be supplied to the CPU 40 when it is analyzed that the failure is an internal failure caused by an internal logic portion. As a result, it is possible to perform effective failure recurrence prevention treatment corresponding to the failure content.

  Here, if the voltage of the power supply to be supplied is increased too much, there is a possibility of causing a temperature failure in the CPU 40. For this reason, the embedded software 11 determines the temperature state by the temperature sensor 50 mounted on the CPU 40, and increases the voltage when the temperature has a margin with respect to a predetermined set value, and cools the temperature when there is no margin. By taking measures such as increasing the voltage after strengthening, it is possible to perform more appropriate failure recurrence prevention treatment. As will be described later, the optimum voltage value of the power supply to be supplied is determined in advance with respect to the relationship between voltage and temperature, and a value derived based on the result of the investigation is set.

  As another failure recurrence prevention measure by the built-in software 11, an effect corresponding to the content of the failure, such as reducing the voltage if the failure is caused by noise, or lowering the clock if the failure is caused by the interface, etc. Relapse prevention treatment can be performed.

  Furthermore, when performing the failure recurrence prevention treatment, the failure recurrence prevention treatment can be performed in consideration of the performance of the CPU 40 in accordance with a preset mode. Here, as modes, i) re-installation is performed without degrading the performance of the CPU 40 (separate when performance is degraded), ii) re-installation is performed even when performance is degraded, iii) immediate decoupling, etc. These modes are preset. The embedded software 11 performs a treatment by selecting one of these modes.

  Next, an example of the operation of the system will be specifically described with reference to the flowchart shown in FIG.

  First, when a failure occurs in the CPU 40, the embedded software 11 disconnects the CPU 40 from the system 1 when detecting a failure interrupt of the CPU 40 (step S101). Next, the embedded software 11 determines whether or not the corresponding CPU 40 can be reincorporated into the system 1 (step S102), and if reincorporation is possible, analyzes the failure content (step S103). If re-installation is not possible, the CPU 40 is disconnected from the system 1 (step S119), and the operation of the system 1 is continued.

  As a result of the analysis of the failure content in step 103, for example, if the failure is caused by a delay (Yes in step S104), the embedded software 11 executes BIST for the corresponding CPU 40 (step S105), and BIST The temperature at the time of execution is measured (step S106). Next, the embedded software 11 converts the measured temperature into a temperature at the time of execution of a high-load JOB (step S107), and determines whether or not the converted temperature has a margin (step S108).

  If there is a margin in temperature (Yes in step S108), the embedded software 11 increases the voltage of the power supplied to the CPU 40 as a failure recurrence prevention measure (step S109). If there is no room for the temperature, the embedded software 11 further determines whether or not the cooling of the CPU 40 can be enhanced (step S111). If the cooling is possible (Yes in step S111). ), After enhancing the cooling (step S112), the voltage is increased (step S113). Next, the embedded software 11 initializes and re-incorporates the CPU 40 (step S110), and continues the operation of the system 1.

  In addition, as a result of the analysis of the failure content in step 103, if the failure is not caused by delay (in the case of No in step S104), for example, if it is a failure caused by noise (Yes in step S114), The embedded software 11 reduces the voltage (step S115), and performs initialization and re-incorporation (step S110).

  As a result of the analysis of the failure content in step 103, if the failure is not a failure caused by delay (No in step S104), and if the failure is not caused by noise (No in step S114), the embedded software 11, for example, determines whether or not the failure is an interface failure (S116).

  In the case of an interface failure (Yes in step S116), the embedded software 11 further determines whether or not the set mode is, for example, a mode in which performance degradation of the system 1 is recognized (step S1). S117). If the mode is a mode in which performance degradation is recognized (Yes in step S117), the embedded software 11 reduces the clock of the CPU 40 (step S118), and performs initialization and re-incorporation (step S110). If there is no interface failure (No in step S116), or if the set mode is not a mode that recognizes the performance degradation of the system 1 (No in step S117), the clock is decreased. Without initialization, initialization and re-installation are performed (step S110).

  Next, a method for determining the voltage to be set when the voltage is increased as a failure recurrence prevention measure will be described. The voltage to be set is preferably set so that when the voltage is increased after the occurrence of a failure, the temperature at the time of executing a high-load job is equal to or lower than the temperature at which the temperature failure occurs. Further, since there is an individual difference in the temperature increase rate of each CPU 40, BIST is used as a method for estimating the temperature at the time of executing a high-load JOB.

  Specifically, first, the correlation between the voltage and the temperature of the CPU 40 during the BIST execution and the high-load JOB execution is examined in advance. The investigated correlation is held in a storage unit (not shown). Then, when a failure occurs, the embedded software 11 performs BIST on the CPU 40 disconnected from the system 1 and measures the temperature of the CPU 40. Further, the embedded software 11 converts the temperature measured at the time of executing the BIST into the temperature at the time of executing the high load job using the correlation that has been investigated and held in advance, and the converted temperature causes a temperature failure. Set the voltage so that there is no range. That is, when a failure occurs, the BIST is executed, the temperature at the time of executing the high load job is estimated from the temperature at the time of executing the BIST, and the set voltage is determined, so that the optimum voltage can be obtained without stopping the system 1. Can be set.

  With reference to FIG. 3, the setting voltage determination method in the case of raising a voltage as a failure recurrence prevention measure will be specifically described. FIG. 3 is a graph showing the correlation between the voltage and the temperature of the CPU 40, which has been investigated in advance, when the BIST is executed and when the high-load job is executed. FIG. 3A is a graph showing an example in which the temperature increase rate is A. FIG. FIG. 3B is a graph showing an example in which the temperature increase rate is B.

  In FIG. 3, it is preferable to set the voltage to be set so that the temperature at the time of execution of the high-load job is equal to or lower than the temperature Tmax at which a failure occurs. For example, when the measured temperature is the temperature T2 when the BIST is executed with the voltage V2, it is determined that the temperature increase rate of the CPU 40 is A from FIGS. 3 (a) and 3 (b). Can do. In this case, if the voltage to be set is equal to or higher than the voltage V4, it will exceed the temperature Tmax. Therefore, it is preferable to set the voltage to be lower than the voltage V4. Further, for example, when the measured temperature is the temperature T1 when the BIST is executed with the voltage V2, it is determined from FIGS. 3A and 3B that the temperature increase rate of the CPU 40 is B. be able to. In this case, the voltage V4 can be set as the set voltage.

  As described above, according to the present invention, when a failure occurs in the CPU 40, the content of the failure that has occurred in the CPU 40 is analyzed, and after the failure recurrence prevention treatment is performed according to the analysis result of the failure content, the computer Providing the diagnosis unit 10 that re-installs the CPU 40 in the system 1 can reduce the possibility of failure recurrence of the re-installed CPU 40.

  Note that the present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention.

1 computer system,
10 diagnostic processor,
11 Embedded software,
20 power control unit,
30 clock controller,
40_1-40_n CPU,
50_1 to 50_n temperature sensor,
60_1 to 60_n BIST,
70 MMU,
80 Cooling device

Claims (16)

When a failure occurs in a CPU (Central Processing Unit), it is a computer system that re-installs after disconnecting the CPU,
A computer system comprising: a diagnostic unit that analyzes the content of a failure that has occurred in the CPU and performs re-prevention of the failure according to the analysis result of the failure content, and then re-installs the CPU into the computer system. A power control unit for controlling the voltage of the power supplied to the CPU;
When the failure that has occurred in the CPU is caused by a delay in an internal logic portion of the computer system, the diagnostic unit increases the voltage of the power supply to be supplied to the CPU as the failure recurrence prevention measure. The computer system according to claim 1, wherein the computer system is characterized. A BIST unit that causes the CPU to execute a BIST (Built In Self Test);
A temperature measuring unit for measuring the temperature of the CPU,
The diagnosis unit, when a failure that occurs in the CPU is caused by a delay in an internal logic part of the computer system,
The computer according to claim 2, wherein the CPU is configured to execute a BIST to measure a temperature, and to set a voltage value for increasing a voltage of a power supply supplied to the CPU based on the measured temperature. system. In response to the voltage change of the power supplied to the CPU, the CPU temperature change at the time of the BIST execution and the CPU temperature change at the time of execution of the high load JOB are measured in advance, and the measured CPU Keep the correlation between the voltage change and temperature change of the power supply to supply,
The diagnosis unit, when a failure that occurs in the CPU is caused by a delay in an internal logic part of the computer system,
The CPU is caused to execute the BIST to measure the temperature, and the measured temperature at the time of the BIST execution is calculated using the correlation between the voltage change of the held power supply and the temperature change of the CPU at the time of execution of the high load JOB. A voltage value for increasing the voltage of the power source supplied to the CPU is set so that the converted temperature becomes a voltage value smaller than a predetermined temperature at which the failure occurs in the CPU. The computer system according to claim 3. A cooling device for cooling the CPU;
The diagnostic unit determines whether or not the measured temperature at the time of the BIST execution has a margin with respect to a predetermined set value when increasing the voltage of the power source supplied to the CPU. 5. The computer system according to claim 2, wherein after the cooling of the CPU is enhanced, a voltage of a power source supplied to the CPU is increased. A power control unit for controlling the voltage of the power supplied to the CPU;
The diagnosis unit, when a failure that has occurred in the CPU is caused by noise in the computer system, lowers the voltage of a power source that is supplied to the CPU as the failure recurrence prevention measure. Item 4. The computer system according to Item 1. A clock control unit for controlling a clock supplied to the CPU;
The said diagnostic part reduces the clock supplied to said CPU as said failure recurrence prevention measure, when the failure which generate | occur | produced in said CPU is the interface failure of said computer system. The computer system described. A mode determined in advance by a combination of the CPU performance degradation when the CPU is re-integrated and the CPU re-integration or disconnection treatment according to the performance degradation is preset,
The diagnosis unit selects whether or not to reduce a clock supplied to the CPU according to the set mode when a failure occurring in the CPU is an interface failure of the computer system. The computer system according to claim 7. A method of re-installing a CPU that performs re-installation after disconnecting the CPU from a computer system when a failure occurs in a CPU (Central Processing Unit).
Analyzing the content of a failure that has occurred in the CPU;
Performing a failure recurrence prevention treatment according to the analysis result of the failure content;
CPU re-installation method. When the failure occurred in the CPU is caused by a delay in the internal logic part of the computer system,
The method of re-installing a CPU according to claim 9, wherein the step of performing the failure recurrence prevention treatment includes a step of increasing a voltage of a power source supplied to the CPU. When the failure occurred in the CPU is caused by a delay in the internal logic part of the computer system,
The step of performing the failure recurrence prevention treatment comprises:
Causing the CPU to execute a BIST (Built In Self Test) and measuring the temperature;
The method of re-installing a CPU according to claim 10, further comprising: setting a voltage value when increasing a voltage of a power source supplied to the CPU based on the measured temperature. In response to the voltage change of the power supplied to the CPU, the CPU temperature change at the time of the BIST execution and the CPU temperature change at the time of execution of the high load JOB are measured in advance, and the measured CPU A step of maintaining a correlation between a voltage change and a temperature change of the power supply to be supplied;
When the failure occurred in the CPU is caused by a delay in the internal logic part of the computer system,
The step of performing the failure recurrence prevention treatment comprises:
Causing the CPU to perform a BIST and measuring the temperature;
Converting the measured temperature at the time of the BIST execution into the temperature of the CPU at the time of executing the high-load JOB using the correlation between the voltage change of the held power supply and the temperature change;
Setting a voltage value at the time of increasing the voltage of the power supply to be supplied to the CPU so that the converted temperature becomes a voltage value lower than a predetermined temperature at which a failure occurs in the CPU. The CPU re-embedding method according to claim 11. The step of performing the failure recurrence prevention treatment comprises:
Determining whether the measured temperature at the time of the BIST execution has a margin with respect to a predetermined set value when raising the voltage of the power source supplied to the CPU;
13. The method according to claim 10, further comprising: increasing the power supply voltage supplied to the CPU after enhancing the cooling of the CPU if there is no room as a result of the determination. The re-installation method of the CPU according to the item. When the failure that occurred in the CPU is caused by noise in the computer system,
The step of performing the failure recurrence prevention treatment comprises:
The method of re-installing a CPU according to claim 9, further comprising a step of reducing a voltage of a power source supplied to the CPU. When the failure that has occurred in the CPU is an interface failure in the computer system,
The step of performing the failure recurrence prevention treatment comprises:
The method of re-installing a CPU according to claim 9, further comprising a step of reducing a clock supplied to the CPU. The method further includes a step in which a mode determined by a combination of the performance degradation of the CPU when the CPU is re-assembled and the re-installation or separation processing of the CPU according to the performance degradation is preset,
When the failure that has occurred in the CPU is an interface failure in the computer system,
The step of performing the failure recurrence prevention treatment comprises:
The method of re-embedding a CPU according to claim 15, further comprising a step of selecting whether or not to reduce a clock supplied to the CPU according to the set mode.

Images