WO2021250737A1 - 情報処理システム及び情報処理システムの制御方法 - Google Patents
情報処理システム及び情報処理システムの制御方法 Download PDFInfo
- Publication number
- WO2021250737A1 WO2021250737A1 PCT/JP2020/022517 JP2020022517W WO2021250737A1 WO 2021250737 A1 WO2021250737 A1 WO 2021250737A1 JP 2020022517 W JP2020022517 W JP 2020022517W WO 2021250737 A1 WO2021250737 A1 WO 2021250737A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- performance
- operating frequency
- execution block
- processor
- storage device
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/324—Power saving characterised by the action undertaken by lowering clock frequency
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3206—Monitoring of events, devices or parameters that trigger a change in power modality
- G06F1/3215—Monitoring of peripheral devices
- G06F1/3225—Monitoring of peripheral devices of memory devices
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3206—Monitoring of events, devices or parameters that trigger a change in power modality
- G06F1/3228—Monitoring task completion, e.g. by use of idle timers, stop commands or wait commands
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/3243—Power saving in microcontroller unit
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/325—Power saving in peripheral device
- G06F1/3275—Power saving in memory, e.g. RAM, cache
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/329—Power saving characterised by the action undertaken by task scheduling
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/3296—Power saving characterised by the action undertaken by lowering the supply or operating voltage
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5094—Allocation of resources, e.g. of the central processing unit [CPU] where the allocation takes into account power or heat criteria
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Definitions
- This disclosure relates to an information processing system and a control method of the information processing system.
- a processor having a dynamic voltage frequency control (DVFS; Dynamic Voltage and Frequency Scaling) function which is one of the mechanisms for reducing power consumption.
- the DVFS function is realized by a power saving mechanism that gives the processor several kinds of operating frequencies and operating voltages and changes the operating frequency and operating voltage of the processor according to the load condition of the processor.
- Patent Document 1 proposes a method of controlling to reduce the computing power when the memory bandwidth is dominant in terms of performance, based on the statistical information on the memory performance.
- Patent Document 2 compares the calculation amount of the central processing unit (CPU; Central Processing Unit) with the access amount to the cache memory, and proposes a method of enabling the power saving mechanism of the processor when the latter is dominant. do.
- CPU Central Processing Unit
- Patent Document 1 has a problem that since statistical information on memory access is used only inside the processor, it is not possible to perform power performance control with high accuracy adapted to the algorithm of the arithmetic application. be. Further, this method has a problem that the power saving control is delayed because the calculation strength of the calculation application is not used, and the frequency of the processor remains low, particularly when high calculation performance is required. Further, in this method, since the operating frequency of the processor and the instruction issuance width are only controlled, the on / off control of the multi-core and the operating frequency of the main storage device are not controlled, and sufficient power saving control is performed. There is a problem that it cannot be done.
- This disclosure was made in view of these problems. It is an object of the present disclosure to enable performance power control adapted to an algorithm of an arithmetic application. The present disclosure also aims to prevent delays in performance power control.
- This disclosure relates to an information processing system.
- the information processing system includes an execution block calculation strength data area, a roofline model data storage unit, a calculation strength data acquisition unit, and a performance power control unit.
- the execution block calculation strength data area holds the calculation strength data of each execution block constituting the calculation application operating in the operating environment of the processor equipped with the power saving mechanism and the computer system equipped with the main storage device.
- the roofline model data storage unit holds a roofline model corresponding to the operating frequency and number of cores of the processor and the operating frequency of the main storage device.
- the calculation strength data acquisition unit acquires the calculation strength data of each execution block from the execution block calculation strength data area.
- the performance power control unit controls the operating frequency and number of cores of the processor and the operating frequency of the main memory based on the roofline model and the calculation strength data of each execution block.
- This disclosure is also directed to the control method of the information processing system.
- performance power control is performed based on the calculation strength data of each execution block constituting the calculation application. This enables performance power control adapted to the algorithm of the arithmetic application. In addition, performance power control is performed in a feedforward manner based on predefined calculation intensity data. This makes it possible to prevent a delay in performance power control.
- FIG. 1 It is a block diagram schematically illustrating the hardware configuration of the information processing system of Embodiment 1. It is a block diagram schematically illustrating the functional configuration of the information processing system of Embodiment 1. It is a flowchart which illustrates the operation flow of the system basic software installed in the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the example of the roofline model held in the roofline storage part provided in the information processing system of Embodiment 1.
- FIG. 1 It is a figure which illustrates the relationship between the selectable operating frequency and the bandwidth of the main storage device which constitutes the roofline model held in the roofline storage part provided in the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the example of the information held in the execution block calculation strength data area provided in the information processing system of Embodiment 1.
- FIG. It is a flowchart which illustrates the operation flow of the performance power determination unit provided in the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the example of the policy of the power saving control when the execution block is memory intensive, which is performed by the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the example of the policy of the power saving control when the execution block is arithmetic intensive, which is performed by the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the example of the overhead time required to perform each control in the information processing system of Embodiment 1.
- FIG. It is a figure which illustrates the operation procedure of the power control latency data unit and the performance power command unit provided in the information processing system of the first embodiment.
- It is a flowchart which illustrates the operation flow of the performance power determination unit provided in the information processing system of Embodiment 2.
- FIG. 1 It is a figure which illustrates the example of the policy of the power saving control when the execution block is memory intensive, which is performed by the information processing system of Embodiment 2. It is a figure which illustrates the example of the policy of the power saving control when the execution block is arithmetic intensive, which is performed by the information processing system of Embodiment 2. FIG. It is a figure which illustrates the example of the policy of the power saving control when the execution block is arithmetic intensive, which is performed by the information processing system of Embodiment 2. FIG.
- FIG. 1 is a block diagram schematically showing a hardware configuration of the information processing system of the first embodiment.
- the information processing system 1000 of the first embodiment includes a computer system 10.
- the computer system 10 includes a processor 11, a main storage device 12, and an auxiliary storage device 13.
- the processor 11 includes a central processing unit (CPU; Central Processing Unit), a graphics processing unit (GPU; Graphics Processing Unit), a digital signal processor (DSP; Digital Signal Processor), and the like.
- the processor 11 includes a power saving mechanism. The power saving mechanism dynamically changes the operating frequency and / or the number of cores of the processor 11.
- the main storage device 12 is a random access memory (RAM; RandomAccessMemory) or the like.
- the auxiliary storage device 13 is a hard disk drive, a solid state drive, a RAM disk, or the like.
- FIG. 2 is a block diagram schematically illustrating the functional configuration of the information processing system of the first embodiment.
- the system basic software 1100 and the arithmetic application 1200 are installed in the information processing system 1000.
- the system basic software 1100 and the calculation application 1200 operate in the operating environment of the computer system 10.
- the system basic software 1100 may be an operating system.
- the algorithm is an algorithm or the like that controls a self-driving car that is executed at a fixed cycle.
- the information processing system 1000 includes a roofline model data storage unit 1110, an operating environment acquisition unit 1120, a calculation strength data acquisition unit 1130, and a performance power control unit 1140. These elements are configured by the processor 1 executing the system basic software 1100 loaded from the auxiliary storage device 13 to the main storage device 12.
- the roofline model data storage unit 1110 holds performance information related to the computer system 10.
- the operating environment acquisition unit 1120 acquires the current operating environment of the computer system 10.
- the calculation strength data acquisition unit 1130 acquires the calculation strength data of each execution block constituting the calculation application 1200 from the execution block calculation strength data area 1230 described below.
- the performance power control unit 1140 performs performance power control based on the held performance information and the acquired calculation strength data of each execution block.
- the performance information about the held computer system 10 includes a roofline model corresponding to the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12.
- the acquired current operating environment of the computer system 10 includes the current operating frequency and the number of cores of the processor 11 and the current operating frequency of the main storage device 12.
- performing performance power control based on the performance information and the calculation strength data of each execution block is the operating frequency and the number of cores of the processor 11 based on the roofline model included in the performance information and the calculation strength data of each execution block. It also includes controlling the operating frequency of the main storage device 12.
- Using the current operating environment of the computer system 10 includes using the current operating frequency and number of cores of the processor 11 and the current operating frequency of the main memory 12 included in the current operating environment of the computer system 10.
- the performance power control unit 1140 includes a performance power determination unit 1141, an execution time measurement unit 1142, a power control latency data unit 1143, and a performance power command unit 1144.
- the performance power determination unit 1141 determines the performance power control policy from the roofline model to be held and the calculation strength data of each execution block.
- the execution time measurement unit 1142 measures the execution time of each execution block.
- the power control latency data unit 1143 determines whether or not to cause the performance power command unit 1144 to perform the performance power control from the overhead time required when the performance power control is performed by the performance power command unit 1144.
- the performance power command unit 1144 outputs a control command according to the determined performance power control policy.
- the performance power command unit 1144 outputs a control command when it is determined by the power control latency data unit 1143 that the performance power control unit 1144 is to perform the performance power control.
- the determined performance power control policy includes the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12.
- following the determined performance power control policy includes following the operating frequency and number of cores of the processor 11 and the operating frequency of the main storage device 12 included in the determined performance power control policy.
- the output of the control command is performed in order to control the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12.
- the information processing system 1000 includes a program area 1210, a data area 1220, and an execution block calculation intensity data area 1230. These elements are secured in at least one of the main storage device 12 and the auxiliary storage device 13.
- the program area 1210 holds the programs constituting the arithmetic application 1200.
- the data area 1220 holds variables, arrays, etc. that make up the arithmetic application 1200.
- the execution block calculation strength data area 1230 holds the calculation strength data of each execution block constituting the calculation application 1200 and the deadline time of each execution block.
- the deadline time of each execution block indicates the time when the processing of each execution block must be completed.
- performance power control is performed based on the calculation intensity data of each execution block constituting the calculation application 1200. This enables performance power control adapted to the algorithm of the arithmetic application 1200.
- the performance power control is performed in a feedforward manner based on the predetermined calculation intensity data. This makes it possible to prevent a delay in performance power control.
- the operating frequency of the main storage device 12 is controlled. As a result, it is possible to prevent the main storage device 12 from consuming more power than necessary.
- FIG. 3 is a flowchart illustrating the operation flow of the system basic software installed in the information processing system of the first embodiment.
- the system basic software 1100 executes steps S100 to S105 shown in FIG.
- step S100 the operating environment acquisition unit 1120 acquires the current operating environment of the computer system 10. At that time, the operating environment acquisition unit 1120 acquires the current operating frequency and the number of cores of the processor 11 and the current operating frequency of the main storage device 12.
- the operating environment acquisition unit 1120 selects a roofline model corresponding to the acquired current operating environment of the computer system 10.
- steps S100 and S101 it becomes possible to refer to the roofline model corresponding to the current operating environment of the computer system 10.
- the calculation strength data acquisition unit 1130 acquires the calculation strength data of the execution block to be executed next.
- the performance power control unit 1140 collates the selected roofline model with the acquired calculation intensity data of the execution block. Further, the performance power control unit 1140 selects the operating environment of the computer system 10. At that time, the performance power control unit 1140 selects the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12.
- step S104 when the performance power control unit 1140 changes the operating environment of the computer system 10 from the current operating environment to the operating environment selected in step S103, the execution time of the execution block is the deadline time due to the control delay. To determine whether or not to exceed.
- the control delay is generated by the overhead time that occurs when the operating environment of the computer system 10 is changed from the current operating environment to the selected operating environment.
- the performance power control unit 1140 determines that the execution time of the execution block exceeds the deadline time, the performance power control unit 1140 ends the operation without executing step S105. On the other hand, if the performance power control unit 1140 determines that the execution time of the execution block does not exceed the deadline time, the performance power control unit 1140 executes step S105 and then ends the operation.
- step S105 the performance power control unit 1140 controls the performance power.
- the performance power control unit 1140 sets the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12 to those selected.
- FIG. 4 is a diagram illustrating an example of a roofline model held in a roofline storage unit provided in the information processing system of the first embodiment.
- the calculation strength is taken on the horizontal axis.
- the performance of floating point arithmetic is taken on the vertical axis.
- the roofline model defines an upper limit for the performance of floating-point arithmetic with respect to the arithmetic strength for each of the selectable arithmetic performance of the processor 11 and each of the selectable memory performance of the main storage device 12.
- the roofline model may specify an upper limit for performance other than the performance of floating point arithmetic.
- the arithmetic performance of the processor 11 is a combination of the operating frequency of the processor 11 and the number of cores.
- the memory performance of the main storage device 12 is the operating frequency of the main storage device 12 or the like.
- the roofline data corresponding to the operating frequency of the main storage device 12 can be referred to.
- the roofline model is the processor 11 with selectable operating frequencies "2.6 GHz”, “2.4 GHz”, “1.8 GHz” and "1.0 GHz” respectively and the main storage device 12.
- the upper limit of the performance of floating-point arithmetic with respect to the arithmetic strength is specified.
- the roofline model from the calculation strength of the execution block constituting the calculation application 1200, either the calculation performance of the processor 11 or the memory performance of the main storage device 12 in the performance of floating-point arithmetic when the execution block is executed. Can be visually determined whether is dominant. Details of the roofline model are described in Samuel Williams, Andrew Waterman and David Patterson, "Roofline: An Insightful Visual Performance Model for Floating-Point Programs and Multicore, (2009)".
- FIG. 5 shows a combination of selectable operating frequencies and number of cores of a processor and an upper limit of the performance of floating-point arithmetic, which constitutes a roofline model held in the roofline storage unit provided in the information processing system of the first embodiment. It is a figure which illustrates the relationship with a value.
- the roofline model defines an upper limit of the performance of floating-point arithmetic with respect to the arithmetic intensity for each of the selectable arithmetic performances of the processor 11.
- the upper limit of the performance of the floating-point arithmetic does not depend on the arithmetic strength.
- the upper limit of the floating-point arithmetic performance with respect to the arithmetic strength for each of the selectable arithmetic performances of the processor 11 can be specified. For example, depending on the relationship between the combination of the selectable operating frequency and the number of cores of the processor 11 and the upper limit of the performance of floating-point arithmetic shown in FIG. 5, the combination of the selectable operating frequency and the number of cores of the processor 11 For each, the upper limit of the performance of floating-point arithmetic can be specified for the arithmetic strength.
- FIG. 6 is a diagram illustrating the relationship between the selectable operating frequency and the bandwidth of the main storage device constituting the roofline model held in the roofline storage unit provided in the information processing system of the first embodiment. be.
- the roofline model defines an upper limit of the performance of floating-point arithmetic with respect to the arithmetic strength for each of the selectable memory performances of the main storage device 12.
- the bandwidth of the main storage device 12 has a one-to-one relationship with the operating frequency of the main storage device 12. Therefore, for each of the selectable bandwidths, the upper limit of the performance of floating-point arithmetic with respect to the arithmetic strength is defined, and the relationship between the selectable operating frequency of the main memory and the bandwidth shown in FIG. 6 is prepared. By doing so, it is possible to specify the upper limit of the performance of floating-point arithmetic with respect to the arithmetic intensity for each of the selectable operating frequencies of the main storage device 12.
- FIG. 7 is a diagram illustrating an example of information held in the execution block calculation strength data area provided in the information processing system of the first embodiment.
- the execution block calculation strength data area 1230 holds the execution address of each execution block, the calculation strength data of each execution block, and the deadline time of each execution block.
- a file containing information that can specify an execution block, calculation intensity data, and deadline time pairing data is created in advance.
- the file is created as a file separate from the source code file of the arithmetic application 1200.
- the information that can specify the execution block is the name of the function corresponding to the execution block or the like.
- compilation is performed to create an executable file of the arithmetic application 1200 from the source code file of the arithmetic application 1200 and the created file.
- the executable and linkable format (ELF; Executable and Linkable Format) is adopted in the created executable file
- the section dedicated to the calculation strength data of each execution block is executed in the execution format file. It may be newly established as. In this case, the information of the newly established section is added to the ELF header and the section header.
- the corresponding machine language part in the program area 1210 is specified from the information that can specify the execution block, and an instruction to generate a software interrupt is inserted in the specified machine language part.
- the instruction that generates a software interrupt is an INT3 instruction or the like when the processor 11 is an x86-based processor.
- An instruction that generates a software interrupt can replace the first byte of the original instruction as a breakpoint. Further, the execution address of the specified machine language portion is acquired, and the acquired execution address is added to the execution block calculation strength data area 1230.
- an interrupt handler that executes a series of processes included in the performance power control unit 1140 is registered in the corresponding interrupt number in the interrupt descriptor table.
- a software interrupt occurs every time each block is reached.
- Loading and execution of the arithmetic application 1200 is initiated by the exec memory in a UNIX® environment. For example, if the instruction that causes the software interrupt is the INT3 instruction, the SIGTRAP signal is notified to the system basic software 1100.
- the system basic software 1100 an interrupt handler registered in advance in the interrupt descriptor table is activated in conjunction with the occurrence of a software interrupt, and a series of processes included in the performance power control unit 1140 are executed. To.
- the calculation strength data acquisition unit 1130 acquires the calculation strength data of each execution block and the deadline time of each execution block based on the execution address of each execution block. At that time, the calculation strength data acquisition unit 1130 identifies an execution block corresponding to the address loaded in the main storage device 12 from the currently executed address, and the calculation strength data of the specified execution block and the said execution block. Get the deadline time of the execution block. Further, the calculated intensity data acquisition unit 1130 passes the acquired calculated intensity data of each execution block and the deadline time of each execution block to the performance power determination unit 1141.
- FIG. 8 is a flowchart illustrating the operation flow of the performance power determination unit provided in the information processing system of the first embodiment.
- the performance power control unit 1140 receives the roofline model corresponding to the current operating environment from the roofline model data storage unit 1110, and the calculation strength data and deadline time of the execution block to be executed next from the calculation strength data acquisition unit 1130. After receiving the above, steps S200 to S207 shown in FIG. 8 are executed.
- step S200 the performance power determination unit 1141 plots the calculated intensity data of the received execution block on the received roofline model. Further, the performance power determination unit 1141 collates the roofline model with the calculation strength data of the execution block.
- the performance power determination unit 1141 determines whether or not the execution block is memory intensive.
- the performance power determination unit 1141 determines whether the memory performance of the main storage device 12 or the calculation performance of the processor 11 is the rate-determining factor in terms of the performance of the calculation application 1200, and the memory performance of the main storage device 12 is the rate-determining factor. If it is determined that there is, it is determined that the execution block is memory intensive, and if it is determined that the arithmetic performance of the processor 11 is a rate-determining factor, it is determined that the execution block is not memory intensive, that is, it is arithmetic intensive. do.
- steps S202 to S204 are executed. If it is determined that the execution block is not memory intensive, steps S205 to S207 are executed.
- step S202 the performance power determination unit 1141 raises the operating frequency of the main storage device 12. At that time, the performance power determination unit 1141 selects an operating frequency higher than the current operating frequency of the main storage device 12 from the selectable operating frequencies of the main storage device 12 held in the roofline model data storage unit 1110. ..
- the performance power determination unit 1141 updates the roofline model. At that time, the performance power determination unit 1141 updates the roofline model based on the operating frequency of the selected main storage device 12.
- the performance power determination unit 1141 determines the operating frequency and / or the operating frequency of the processor 11 so that the discontinuity between the slope portion of the roofline model and the flat portion of the roofline model is located on the calculated strength. Reduce the number of cores. At that time, the performance power determination unit 1141 is based on the current operating frequency and / or the number of cores of the processor 11 from the selectable operating frequency and / or the number of cores of the processor 11 held in the roofline model data storage unit 1110. Select a small operating frequency and / or number of cores.
- the gradient portion of the roofline model exists in the range of calculation strength in which the memory performance of the main storage device 12 is a rate-determining factor.
- the flat portion of the roofline model exists in the range of the calculation strength in which the calculation performance of the processor 11 is a rate-determining factor.
- step S205 the performance power determination unit 1141 raises the operating frequency and / or the number of cores of the processor 11. At that time, the performance power determination unit 1141 is based on the current operating frequency and / or the number of cores of the processor 11 from the selectable operating frequency and / or the number of cores of the processor 11 held in the roofline model data storage unit 1110. Select a large operating frequency and / or number of cores.
- the performance power determination unit 1141 updates the roofline model. At that time, the performance power determination unit 1141 updates the roofline model based on the operating frequency and / or the number of cores of the selected processor 11.
- the performance power determination unit 1141 determines the operating frequency of the main storage device 12 so that the discontinuity between the slope portion of the roofline model and the flat portion of the roofline model is located on the calculated strength. Lower. At that time, the performance power determination unit 1141 selects an operating frequency smaller than the current operating frequency of the main storage device 12 from the selectable operating frequencies of the main storage device 12 held in the roofline model data storage unit 1110. ..
- FIG. 9 is a diagram illustrating an example of a power saving control policy when the execution block is memory intensive, which is performed by the information processing system of the first embodiment.
- the performance when the execution block is executed with respect to the memory performance of the current main storage device 12 and the arithmetic performance of the processor 11 shown by the broken line It is decided to increase the memory performance of the main storage device 12 which is a hindrance to the memory performance of the main storage device 12 illustrated by the solid line gradient portion, and the performance requirement is satisfied. Further, it is determined to reduce the arithmetic performance of the processor 11 to the arithmetic performance of the processor 11 illustrated by the solid flat portion so that the discontinuity between the gradient portion and the flat portion is located on the arithmetic strength, thereby saving power. Is planned.
- the memory performance of the main storage device 12 and the calculation performance of the processor 11 are shifted so that the memory performance of the main storage device 12 and the calculation performance of the processor 11 shift to the memory performance of the main storage device 12 and the calculation performance of the processor 11 shown by the solid line.
- the calculation performance of is selected.
- FIG. 10 is a diagram illustrating an example of a power saving control policy when the execution block is arithmetic intensive, which is performed by the information processing system of the first embodiment.
- the performance when the execution block is executed with respect to the memory performance of the current main storage device 12 and the arithmetic performance of the processor 11 shown by the broken line It is determined that the arithmetic performance of the processor 11 that is a hindrance is increased to the arithmetic performance of the processor 11 illustrated by the solid line flat portion, and the performance requirement is satisfied. Further, it is decided to reduce the memory performance of the main storage device 12 to the memory performance of the main storage device 12 illustrated by the solid line gradient portion so that the discontinuity point between the gradient portion and the flat portion is located on the calculation strength. It is possible to save power.
- the memory performance of the main storage device 12 and the calculation performance of the processor 11 are shifted so that the memory performance of the main storage device 12 and the calculation performance of the processor 11 shift to the memory performance of the main storage device 12 and the calculation performance of the processor 11 shown by the solid line.
- the calculation performance of is selected.
- power saving can be performed while satisfying the required performance requirements.
- FIG. 11 is a diagram illustrating an example of an overhead time required to perform each control in the information processing system of the first embodiment.
- the overhead time required to perform each control shown in FIG. 11 is predefined.
- the overhead time required to perform each of the controls includes the operating frequency of the processor 11, the ON / OFF of the core of the processor 11, and the overhead time required to control the operating frequency of the main storage device 12.
- FIG. 12 is a diagram illustrating an operation procedure of a power control latency data unit and a performance power command unit provided in the information processing system of the first embodiment.
- the processing related to the performance power control by the system basic software 1100 is executed before each execution block constituting the arithmetic application is executed by the software interrupt.
- the execution time measuring unit 1142 can measure the execution time of each execution block by acquiring the current time before and after the process. ..
- the power control latency data unit 1143 holds the execution time of each measured execution block. Further, the power control latency data unit 1143 determines whether or not to perform performance power control from the measured execution time of each execution block and the overhead time required to perform each control shown in FIG. 11. do.
- the total of the execution time and the overhead time of the execution block in the previous cycle is the deadline time of the execution block acquired by the calculation intensity data acquisition unit 1130.
- an instruction for performing performance power control for the execution block is output to the performance power command unit 1144.
- the power control latency data unit 1143 outputs an instruction for not performing performance power control for the execution block to the performance power command unit 1144.
- the performance power control is performed from the roofline model corresponding to the current operating environment of the computer system 10 based only on the calculation strength data of each execution block constituting the calculation application 1200.
- the operating environment is the operating frequency and the number of cores of the processor 11 and the operating frequency of the main storage device 12.
- the actual performance when the arithmetic application 1200 is executed does not necessarily match the limit performance of the computer system 10 shown by the roofline model.
- the actual calculation performance when the calculation application 1200 is executed is used to have higher accuracy. Performance power control is realized.
- the calculated performance used is referred to as "actual calculation performance”.
- the actual calculation performance of each execution block is such that the total number of floating-point operations specified from the calculation strength data of each execution block acquired by the calculation strength data acquisition unit 1130 is held by the power control latency data unit 1143 in each execution block. It can be calculated by dividing by the execution time of.
- FIG. 13 is a flowchart illustrating the operation flow of the performance power determination unit provided in the information processing system of the second embodiment.
- the performance power control unit 1140 executes steps S300 to S309 shown in FIG.
- step S300 the performance power determination unit 1141 plots the calculated intensity data of the received execution block on the received roofline model. Further, the performance power determination unit 1141 collates the roofline model with the calculation strength data of the execution block.
- the performance power determination unit 1141 determines whether or not the execution block is memory intensive.
- steps S302 to S305 are executed. If it is determined that the execution block is not memory intensive, steps S306 to S309 are executed.
- step S302 the performance power determination unit 1141 determines whether or not the actual calculation performance of the execution block has reached the peak performance of the memory performance of the main storage device 12 in the current operating environment.
- steps S303 to S305 are executed. If it is determined that the actual calculation performance of the execution block has not reached the peak performance of the memory performance, step S305 is executed.
- step S303 the performance power determination unit 1141 raises the operating frequency of the main storage device 12. At that time, the performance power determination unit 1141 selects an operating frequency higher than the current operating frequency of the main storage device 12 from the selectable operating frequencies of the main storage device 12 held in the roofline model data storage unit 1110. ..
- the performance power determination unit 1141 updates the roofline model. At that time, the performance power determination unit 1141 updates the roofline model based on the operating frequency of the selected main storage device 12.
- the performance power determination unit 1141 determines the operating frequency and / or the operating frequency of the processor 11 so that the discontinuity between the slope portion of the roofline model and the flat portion of the roofline model is located on the calculated strength. Reduce the number of cores. At that time, the performance power determination unit 1141 is based on the current operating frequency and / or the number of cores of the processor 11 from the selectable operating frequency and / or the number of cores of the processor 11 held in the roofline model data storage unit 1110. Select a small operating frequency and / or number of cores.
- steps S302 to S305 if the actual calculation performance of the execution block does not reach the peak performance of the memory performance of the main storage device 12, the current operating environment is mainly stored with respect to the operating frequency of the main storage device 12. It is determined that the memory performance requirements of the device 12 are met and the selection is not made.
- step S306 the performance power determination unit 1141 determines whether or not the actual calculation performance of the execution block has reached the peak performance of the calculation performance of the processor 11 in the current operating environment.
- steps S307 to S309 are executed. If it is determined that the actual calculation performance of the execution block has not reached the peak performance of the calculation performance of the processor 11, step S309 is executed.
- step S307 the performance power determination unit 1141 raises the operating frequency and / or the number of cores of the processor 11. At that time, the performance power determination unit 1141 is based on the current operating frequency and / or the number of cores of the processor 11 from the selectable operating frequency and / or the number of cores of the processor 11 held in the roofline model data storage unit 1110. Select a large operating frequency and / or number of cores.
- the performance power determination unit 1141 updates the roofline model. At that time, the performance power determination unit 1141 updates the roofline model based on the operating frequency and / or the number of cores of the selected processor 11.
- the performance power determination unit 1141 determines the operating frequency of the main storage device 12 so that the discontinuity between the slope portion of the roofline model and the flat portion of the roofline model is located on the calculated strength. Lower. At that time, the performance power determination unit 1141 selects an operating frequency smaller than the current operating frequency of the main storage device 12 from the selectable operating frequencies of the main storage device 12 held in the roofline model data storage unit 1110. ..
- steps S306 to S309 if the actual calculation performance of the execution block does not reach the peak performance of the calculation performance of the processor 11, the current operating environment of the processor 11 is the operating frequency and the number of cores of the processor 11. It is determined that the requirements for computing performance are met and the selection is not made.
- FIGS. 14 and 15 are diagrams illustrating an example of a power saving control policy when the execution block is memory intensive, which is performed by the information processing system of the second embodiment.
- the actual calculation performance of the execution block reaches the peak performance of the memory performance of the main storage device 12 in the current operating environment. Therefore, the memory performance of the main storage device 12, which is a hindrance to the performance when the execution block is executed, is raised to the memory performance shown by the solid line gradient portion, and the performance requirement is satisfied. Further, the calculation performance of the processor 11 is reduced to the calculation performance shown by the solid line flat portion so that the discontinuity point between the gradient portion and the flat portion is located on the calculation strength, and power saving is achieved. As a result, the memory performance of the main storage device 12 and the arithmetic performance of the processor 11 shift to those shown by the solid line.
- the actual calculation performance of the execution block does not reach the peak performance of the memory performance of the main storage device 12 in the current operating environment. Therefore, the memory performance of the main storage device 12 that does not hinder the performance when the execution block is executed is maintained. Further, the calculation performance of the processor 11 is reduced to the calculation performance of the processor 11 illustrated by the solid line flat portion so that the discontinuity point between the gradient portion and the flat portion is located on the calculation strength, thereby saving power. Be done. As a result, the memory performance of the main storage device 12 and the arithmetic performance of the processor 11 shift to those shown by the solid line.
- 16 and 17 are diagrams illustrating an example of a power saving control policy when the execution block is arithmetic intensive, which is performed by the information processing system of the second embodiment.
- the actual calculation performance of the execution block reaches the peak performance of the calculation performance of the processor 11 in the current operating environment. Therefore, the arithmetic performance of the processor 11, which is a hindrance to the performance when the execution block is executed, is raised to the arithmetic performance of the processor 11 illustrated by the solid line flat portion, and the performance requirement is satisfied. Further, the memory performance of the main storage device 12 is reduced to the memory performance of the main storage device 12 illustrated by the solid line gradient portion so that the discontinuity point between the gradient portion and the flat portion is located on the calculation strength. Electricity will be achieved. As a result, the memory performance of the main storage device 12 and the arithmetic performance of the processor 11 shift to those shown by the solid line.
- the actual calculation performance of the execution block does not reach the peak performance of the calculation performance of the processor 11 in the current operating environment. Therefore, the arithmetic performance of the processor 11 that does not hinder the performance when the execution block is executed is maintained. Further, the memory performance of the main storage device 12 is reduced to the memory performance of the main storage device 12 illustrated by the solid line gradient portion so that the discontinuity point between the gradient portion and the flat portion is located on the calculation strength. Electricity will be achieved. As a result, the memory performance of the main storage device 12 and the arithmetic performance of the processor 11 shift to those shown by the solid line.
- Computer system 11 Processor, 12 Main memory device, 13 Auxiliary storage device, 1000 Information processing system, 1100 System basic software, 1200 Calculation application, 1110 Roofline model data storage unit, 1120 Operating environment acquisition unit, 1130 Calculation strength data acquisition Unit 1140 Performance power control unit 1141 Performance power judgment unit 1142 Execution time measurement unit 1143 Power control latency data unit 1144 Performance power command unit 1210 Program area 1220 Data area 1230 Execution block calculation strength data area.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Power Sources (AREA)
Abstract
Description
図1は、実施の形態1の情報処理システムのハードウェア構成を模式的に図示するブロック図である。
以下では、実施の形態2が実施の形態1と異なる点が説明される。説明されない点については、実施の形態1において採用された構成が実施の形態2においても採用される。
Claims (10)
- 省電力機構を備えるプロセッサ及び主記憶装置を備える計算機システムの動作環境で動作する演算アプリケーションを構成する各実行ブロックの演算強度データを保持する実行ブロック演算強度データ領域と、
前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数に対応するルーフラインモデルを保持するルーフラインモデルデータ記憶部と、
前記実行ブロック演算強度データ領域から前記各実行ブロックの演算強度データを取得する演算強度データ取得部と、
前記ルーフラインモデル及び前記各実行ブロックの演算強度データに基づいて前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数の制御を行う性能電力制御部と、
を備える情報処理システム。 - 前記プロセッサの現在の動作周波数及びコア数並びに前記主記憶装置の現在の動作周波数を取得する動作環境取得部を備える
請求項1の情報処理システム。 - 前記実行ブロック演算強度データ領域は、前記各実行ブロックの実行アドレス、前記各実行ブロックの演算強度データ、及び前記各実行ブロックの処理を終了しなければならない時間を示す前記各実行ブロックのデッドライン時間を保持する
請求項1又は2の情報処理システム。 - 前記演算強度データ取得部は、前記各実行ブロックの実行アドレスに基づいて前記各実行ブロックの演算強度データ及び前記各実行ブロックのデッドライン時間を取得する
請求項3の情報処理システム。 - 前記ルーフラインモデルは、前記プロセッサの選択可能な動作周波数及びコア数の組み合わせの各々及び前記主記憶装置の選択可能な動作周波数の各々について演算強度に対する性能の上限値を規定する
請求項1から4までのいずれかの情報処理システム。 - 前記性能電力制御部は、
前記ルーフラインモデル及び前記各実行ブロックの演算強度データから前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数を決定する性能電力判断部と、
前記各実行ブロックの実行時間を計測する実行時間計測部と、
前記性能電力判断部により決定された前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数に従って前記制御を行う性能電力指令部と、
前記制御を前記性能電力指令部に行わせた場合にかかるオーバヘッド時間から前記制御を前記性能電力指令部に行わせるか否かを判定する電力制御レイテンシデータ部と、
を備える
請求項1から5までのいずれかの情報処理システム。 - 前記性能電力判断部は、
前記ルーフラインモデルと前記各実行ブロックの演算強度データとを照合し、
前記主記憶装置のメモリ性能及び前記プロセッサの演算性能のいずれが前記演算アプリケーションの性能面における律速要因であるのかを判定し、
前記メモリ性能が前記律速要因であると判定した場合は、前記ルーフラインモデルデータ記憶部に保持されている前記主記憶装置の選択可能な動作周波数から前記主記憶装置の現在の動作周波数より大きい動作周波数を選択し、
前記演算性能が前記律速要因であると判定した場合は、前記ルーフラインモデルデータ記憶部に保持されている前記プロセッサの選択可能な動作周波数及び/又はコア数から現在の動作周波数及び/又はコア数より大きい動作周波数及び/又はコア数を選択する
請求項6の情報処理システム。 - 前記電力制御レイテンシデータ部は、前記実行時間計測部により計測された前記各実行ブロックの実行時間と、予め定義された各制御を行うのにかかるオーバヘッド時間と、から前記制御を行うか否かを判断する
請求項6又は7の情報処理システム。 - 前記性能電力指令部は、前記電力制御レイテンシデータ部により前記制御を行うと判断された場合に、前記性能電力判断部により決定された前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数を、前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数に設定する
請求項6から8までのいずれかの情報処理システム。 - a) 省電力機構を備えるプロセッサ及び主記憶装置を備える計算機システムの動作環境で動作する演算アプリケーションを構成する各実行ブロックの演算強度データを保持する工程と、
b) 前記各実行ブロックの演算強度データを取得する工程と、
c) 前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数に対応するルーフラインモデルを記憶する工程と、
d) 前記ルーフラインモデル及び前記各実行ブロックの演算強度データから前記プロセッサの動作周波数及びコア数並びに前記主記憶装置の動作周波数の制御を行う工程と、
を備える情報処理システムの制御方法。
Priority Applications (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2022530363A JP7357790B2 (ja) | 2020-06-08 | 2020-06-08 | 情報処理システム及び情報処理システムの制御方法 |
| CN202080101272.4A CN115698950B (zh) | 2020-06-08 | 2020-06-08 | 信息处理系统以及信息处理系统的控制方法 |
| DE112020007310.6T DE112020007310T5 (de) | 2020-06-08 | 2020-06-08 | Informations-verarbeitungssystem und steuerungsverfahren für eininformations-verarbeitungssystem |
| PCT/JP2020/022517 WO2021250737A1 (ja) | 2020-06-08 | 2020-06-08 | 情報処理システム及び情報処理システムの制御方法 |
| US17/918,583 US12222788B2 (en) | 2020-06-08 | 2020-06-08 | Information processing system and information processing system control method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2020/022517 WO2021250737A1 (ja) | 2020-06-08 | 2020-06-08 | 情報処理システム及び情報処理システムの制御方法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2021250737A1 true WO2021250737A1 (ja) | 2021-12-16 |
Family
ID=78845430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2020/022517 Ceased WO2021250737A1 (ja) | 2020-06-08 | 2020-06-08 | 情報処理システム及び情報処理システムの制御方法 |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US12222788B2 (ja) |
| JP (1) | JP7357790B2 (ja) |
| CN (1) | CN115698950B (ja) |
| DE (1) | DE112020007310T5 (ja) |
| WO (1) | WO2021250737A1 (ja) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023238276A1 (ja) * | 2022-06-08 | 2023-12-14 | 三菱電機株式会社 | 情報処理装置および情報処理方法 |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005085164A (ja) * | 2003-09-10 | 2005-03-31 | Sharp Corp | マルチプロセッサシステムの制御方法およびマルチプロセッサシステム |
| JP2020077242A (ja) * | 2018-11-08 | 2020-05-21 | シャープ株式会社 | 電子機器、制御装置、電子機器の制御方法、および制御プログラム |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006006174A2 (en) * | 2004-07-15 | 2006-01-19 | N-Trig Ltd. | A tracking window for a digitizer system |
| JP4231516B2 (ja) | 2006-08-04 | 2009-03-04 | 株式会社日立製作所 | 実行コードの生成方法及びプログラム |
| WO2008120274A1 (ja) | 2007-03-29 | 2008-10-09 | Fujitsu Limited | 演算能力を制御する処理装置 |
| EP2793133B1 (en) * | 2011-12-13 | 2019-09-25 | Toyota Jidosha Kabushiki Kaisha | Multi-core processor |
| US20140089699A1 (en) * | 2012-09-27 | 2014-03-27 | Advanced Micro Devices | Power management system and method for a processor |
| US9600392B2 (en) * | 2014-08-11 | 2017-03-21 | International Business Machines Corporation | Tracking pipelined activity during off-core memory accesses to evaluate the impact of processor core frequency changes |
| US10048740B2 (en) * | 2015-07-31 | 2018-08-14 | International Business Machines Corporation | Controlling power consumption |
| US10509677B2 (en) * | 2015-09-30 | 2019-12-17 | Lenova (Singapore) Pte. Ltd. | Granular quality of service for computing resources |
| KR102940878B1 (ko) * | 2021-03-03 | 2026-03-18 | 삼성전자주식회사 | 뉴럴 프로세싱 장치 및 뉴럴 프로세싱 장치의 동작 방법 |
-
2020
- 2020-06-08 JP JP2022530363A patent/JP7357790B2/ja active Active
- 2020-06-08 DE DE112020007310.6T patent/DE112020007310T5/de active Pending
- 2020-06-08 WO PCT/JP2020/022517 patent/WO2021250737A1/ja not_active Ceased
- 2020-06-08 US US17/918,583 patent/US12222788B2/en active Active
- 2020-06-08 CN CN202080101272.4A patent/CN115698950B/zh active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005085164A (ja) * | 2003-09-10 | 2005-03-31 | Sharp Corp | マルチプロセッサシステムの制御方法およびマルチプロセッサシステム |
| JP2020077242A (ja) * | 2018-11-08 | 2020-05-21 | シャープ株式会社 | 電子機器、制御装置、電子機器の制御方法、および制御プログラム |
Non-Patent Citations (1)
| Title |
|---|
| YASUDA, IPPEI: "Evaluation of Performance Models taking Massively Parallel System Inter-node Communications into Consideration", 2012 IPSJ SIG TECHNICAL REPORT, vol. 24, no. 5, 15 February 2013 (2013-02-15), pages 1 - 6, ISSN: 1884-0930 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023238276A1 (ja) * | 2022-06-08 | 2023-12-14 | 三菱電機株式会社 | 情報処理装置および情報処理方法 |
| JPWO2023238276A1 (ja) * | 2022-06-08 | 2023-12-14 | ||
| JP7837409B2 (ja) | 2022-06-08 | 2026-03-30 | 三菱電機モビリティ株式会社 | 情報処理装置および情報処理方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115698950A (zh) | 2023-02-03 |
| US12222788B2 (en) | 2025-02-11 |
| US20230141385A1 (en) | 2023-05-11 |
| JPWO2021250737A1 (ja) | 2021-12-16 |
| JP7357790B2 (ja) | 2023-10-06 |
| DE112020007310T5 (de) | 2023-06-15 |
| CN115698950B (zh) | 2025-10-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5283762B2 (ja) | 演算処理装置、情報処理装置及びその制御方法 | |
| US8276015B2 (en) | Managing the power-performance range of an application | |
| JP5782565B2 (ja) | プロセッサのターボモード動作での電力効率を向上させる方法 | |
| CN107743608B (zh) | 至硬件加速器的动态功率路由 | |
| US8959317B2 (en) | Processor and method for saving designated registers in interrupt processing based on an interrupt factor | |
| JP5316718B1 (ja) | マルチコア・プロセッサ | |
| US10120712B2 (en) | Instruction pre-fetching | |
| US10120713B2 (en) | Hardware controlled instruction pre-fetching | |
| CN118868077B (zh) | 一种基于动态电压频率调整的能源效率优化方法 | |
| JP2012048545A (ja) | 特定のプロセスを短時間で処理する方法およびコンピュータ | |
| CN110795238A (zh) | 负载计算方法、装置、存储介质及电子设备 | |
| WO2021250737A1 (ja) | 情報処理システム及び情報処理システムの制御方法 | |
| US12524269B2 (en) | Stack memory allocation control based on monitored activities | |
| JP4965638B2 (ja) | タスクの切り換えを制御するシステムおよび方法 | |
| US10180839B2 (en) | Apparatus for information processing with loop cache and associated methods | |
| JP4972522B2 (ja) | データ処理システム | |
| JP2019113984A (ja) | 計算制御プログラム、計算制御方法及び画像形成装置 | |
| US20220214993A1 (en) | Electronic computing device | |
| JP2006126947A (ja) | 情報処理装置、情報処理方法、およびプログラム | |
| JP5858947B2 (ja) | 情報処理装置及び動作周波数/動作電圧制御方法 | |
| CN120196527B (zh) | 处理器性能硬件优化方法、系统、电子设备及存储介质 | |
| JP7157542B2 (ja) | プリフェッチコントローラ | |
| JP2003271455A (ja) | キャッシュメモリ制御装置およびキャッシュメモリシステム | |
| JP2011008617A (ja) | マルチスレッド実行装置、オブジェクトプログラムの生成方法、プログラム | |
| JPH0391046A (ja) | データ処理装置 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20940262 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 2022530363 Country of ref document: JP Kind code of ref document: A |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 20940262 Country of ref document: EP Kind code of ref document: A1 |
|
| WWG | Wipo information: grant in national office |
Ref document number: 202080101272.4 Country of ref document: CN |