Gradient-priority-based integrated circuit temperature control and power consumption optimization method and device

Abstract

The application relates to the technical field of temperature control of integrated circuits, and discloses an integrated circuit temperature control and power consumption optimization method and device based on gradient priority, which comprises the following steps: calculating the energy efficiency coefficients of each working frequency point of each cluster in an integrated circuit, sorting each energy efficiency coefficient pair to obtain an energy efficiency coefficient array, determining the sorting index of the current working frequency point of each cluster in the energy efficiency coefficient array, determining temperature regulation requirements according to temperature data of the integrated circuit, screening a target current working frequency point and a corresponding target cluster according to the temperature regulation requirements, and performing a temperature regulation operation on the target cluster. It can be seen that the application can realize gradient prioritization of temperature control scheduling by globally and uniformly sorting the energy efficiency coefficients of all working frequency points of the multi-cluster integrated circuit, accurately locates the cluster to be regulated based on the working frequency point, realizes the maximum power consumption reduction with the smallest performance loss when the temperature control is intervened, and greatly improves the temperature control efficiency and temperature control power consumption benefit of the integrated circuit.

Description

Technical Field

[0001] This invention relates to the field of temperature control technology for integrated circuits, and in particular to a method and apparatus for temperature control and power consumption optimization of integrated circuits based on gradient priority. Background Technology

[0002] With the iteration of integrated circuit technology and the widespread application of multi-core, multi-cluster CPU architecture, the requirements for the coordination of temperature control efficiency and operating performance in electronic systems continue to increase. The native temperature control framework of the Linux kernel divides each cluster of the CPU into isolated independent hot zones, and each hot zone executes its own temperature control strategy independently. It does not take into account the energy efficiency differences of different CPU clusters' DVFS frequency points. When temperature control intervenes, it adjusts the operating frequency of each cluster without priority. This not only results in low power consumption benefits and poor temperature control efficiency, but also causes a step change in CPU performance, leading to system lag and failing to balance the temperature control effect with the user's performance experience.

[0003] Existing temperature control technologies mostly achieve cooling by directly reducing CPU frequency or by pre-setting temperature control parameters based on static simulation. These approaches fail to prioritize reducing computing power output at the least energy-efficient frequency points, resulting in insufficient targeted temperature control, low power consumption gains, and poor temperature control efficiency. Therefore, proposing a technical solution that can improve the temperature control efficiency of integrated circuits and enhance power consumption gains is of paramount importance. Summary of the Invention

[0004] This invention provides a gradient-priority integrated circuit temperature control and power consumption optimization method and apparatus, which can help improve the temperature control efficiency of integrated circuits and increase the benefits of temperature control power consumption.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses a gradient-first integrated circuit temperature control and power consumption optimization method, the method comprising: Obtain all operating frequency points of each cluster in the integrated circuit, and calculate the energy efficiency coefficient of each operating frequency point; Sort each operating frequency point according to its energy efficiency coefficient to obtain a monotonically increasing array of energy efficiency coefficients; Determine the current operating frequency point corresponding to each cluster, and determine the sorting index of each current operating frequency point in the energy efficiency coefficient array; Monitor the temperature data of the integrated circuit, and determine the temperature control requirements of the integrated circuit based on the temperature data; Based on the temperature control requirements, the target current operating frequency corresponding to the temperature control requirements is selected from the energy efficiency coefficient array, and the target cluster corresponding to the target current operating frequency is determined, and the temperature control operation is performed on the target cluster.

[0006] As an optional implementation, in the first aspect of the present invention, determining the temperature control requirements corresponding to the integrated circuit based on the temperature data includes: The temperature change trend is determined based on the preset temperature polling cycle and the temperature data. When the temperature data is greater than the preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend. When the temperature data is less than the preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

[0007] As an optional implementation, in the first aspect of the present invention, the step of filtering the target current operating frequency point corresponding to the temperature control requirement from the energy efficiency coefficient array according to the temperature control requirement, and determining the target cluster corresponding to the target current operating frequency point, includes: When the integrated circuit meets the frequency reduction condition, the first current operating frequency point with the largest sort index is selected from the energy efficiency coefficient array, and the first cluster to which the first current operating frequency point belongs is determined as the target cluster; When the integrated circuit meets the frequency increase condition, the second current operating frequency point with the smallest sorting index is selected from the energy efficiency coefficient array, and the second cluster to which the second current operating frequency point belongs is determined as the target cluster.

[0008] As an optional implementation, in the first aspect of the present invention, performing temperature control operations on the target cluster includes: Determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster; Determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold; When the frequency control target value is greater than the idle injection critical ratio threshold, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, a frequency control operation is performed on the target cluster.

[0009] As an optional implementation, in the first aspect of the present invention, performing the idle injection operation on the target cluster includes: The number of cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration are determined. The idle cycle period is calculated based on the number of cores, the idle injection allocation rule for each core, and the single-cycle idle duration. Based on the idle cycle period, the single-cycle idle duration, and the idle injection allocation rules, an idle injection operation is performed on the target cluster.

[0010] As an optional implementation, in the first aspect of the present invention, determining the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster includes: Determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency; Calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster based on the ratio.

[0011] As an optional implementation, in the first aspect of the present invention, the step of performing an idle injection operation on the target cluster when the frequency modulation target value is greater than the idle injection critical ratio threshold, and performing a frequency modulation operation on the target cluster after the idle injection operation is completed, includes: When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval; When the first idle injection ratio is equal to the idle injection critical ratio threshold, it is determined that the idle injection operation has been completed, and a frequency reduction operation is performed on the target cluster. When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency increase condition, an idle injection operation is performed on the target cluster, and the second idle injection ratio is reduced according to a preset time interval; When the second idle injection ratio is lower than the preset low threshold of the idle injection ratio, it is determined that the idle injection operation has been completed, and the frequency increase operation is performed on the target cluster.

[0012] A second aspect of this invention discloses a gradient-first integrated circuit temperature control and power consumption optimization device, the device comprising: The acquisition module is used to acquire all operating frequency points of each cluster in the integrated circuit and calculate the energy efficiency coefficient of each operating frequency point; The sorting module is used to sort each operating frequency point according to the energy efficiency coefficient of each operating frequency point to obtain a monotonically increasing energy efficiency coefficient array. The determination module is used to determine the current operating frequency point corresponding to each cluster, and to determine the sorting index of each current operating frequency point in the energy efficiency coefficient array; A monitoring module is used to monitor the temperature data of the integrated circuit and determine the temperature control requirements of the integrated circuit based on the temperature data. The filtering module is used to filter the target current operating frequency corresponding to the temperature control requirement in the energy efficiency coefficient array according to the temperature control requirement, determine the target cluster corresponding to the target current operating frequency, and perform temperature control operation on the target cluster.

[0013] As an optional implementation, in a second aspect of the present invention, the monitoring module determines the temperature control requirements corresponding to the integrated circuit based on the temperature data in a specific manner including: The temperature change trend is determined based on the preset temperature polling cycle and the temperature data. When the temperature data is greater than the preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend. When the temperature data is less than the preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

[0014] As an optional implementation, in a second aspect of the present invention, the filtering module filters the target current operating frequency point corresponding to the temperature control requirement from the energy efficiency coefficient array according to the temperature control requirement, and determines the target cluster corresponding to the target current operating frequency point in a specific manner including: When the integrated circuit meets the frequency reduction condition, the first current operating frequency point with the largest sort index is selected from the energy efficiency coefficient array, and the first cluster to which the first current operating frequency point belongs is determined as the target cluster; When the integrated circuit meets the frequency increase condition, the second current operating frequency point with the smallest sorting index is selected from the energy efficiency coefficient array, and the second cluster to which the second current operating frequency point belongs is determined as the target cluster.

[0015] As an optional implementation, in a second aspect of the present invention, the screening module performs temperature control operations on the target cluster in a manner that specifically includes: Determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster; Determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold; When the frequency control target value is greater than the idle injection critical ratio threshold, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, a frequency control operation is performed on the target cluster.

[0016] As an optional implementation, in the second aspect of the present invention, the filtering module performs the idle injection operation on the target cluster in a manner that specifically includes: The number of cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration are determined. The idle cycle period is calculated based on the number of cores, the idle injection allocation rule for each core, and the single-cycle idle duration. Based on the idle cycle period, the single-cycle idle duration, and the idle injection allocation rules, an idle injection operation is performed on the target cluster.

[0017] As an optional implementation, in a second aspect of the present invention, the method by which the filtering module determines the idle injection critical ratio threshold corresponding to the current operating frequency point of the target cluster specifically includes: Determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency; Calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster based on the ratio.

[0018] As an optional implementation, in a second aspect of the present invention, the filtering module performs an idle injection operation on the target cluster when the frequency regulation target value is greater than the idle injection critical ratio threshold, and performs a frequency regulation operation on the target cluster after the idle injection operation is completed. Specifically, this includes: When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval; When the first idle injection ratio is equal to the idle injection critical ratio threshold, it is determined that the idle injection operation has been completed, and a frequency reduction operation is performed on the target cluster. When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency increase condition, an idle injection operation is performed on the target cluster, and the second idle injection ratio is reduced according to a preset time interval; When the second idle injection ratio is lower than the preset low threshold of the idle injection ratio, it is determined that the idle injection operation has been completed, and the frequency increase operation is performed on the target cluster.

[0019] A third aspect of the present invention discloses another integrated circuit temperature control and power consumption optimization device based on gradient priority, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the gradient-first integrated circuit temperature control and power consumption optimization method according to any of the first aspects of the present invention.

[0020] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the gradient-first integrated circuit temperature control and power consumption optimization method described in any of the first aspects of the present invention.

[0021] Compared with the prior art, the present invention has the following beneficial effects: In this embodiment of the invention, all operating frequencies of each cluster in the integrated circuit are obtained, and the energy efficiency coefficient (EEC) of each operating frequency is calculated. Each operating frequency is then sorted according to its EEC to obtain a monotonically increasing EEC array. The current operating frequency corresponding to each cluster is determined, and its sorting index in the EEC array is also determined. Temperature data of the integrated circuit is monitored, and the temperature control requirements corresponding to the integrated circuit are determined based on the temperature data. The target current operating frequency corresponding to the temperature control requirements is then selected from the EEC array, and the target cluster corresponding to the target current operating frequency is determined. Temperature control operations are then performed on the target cluster. Therefore, implementing this invention enables gradient priority for temperature control scheduling by uniformly sorting the global EEC of all operating frequencies of multi-cluster integrated circuits. It accurately locates the cluster to be controlled based on energy efficiency levels, achieving maximum power reduction with minimal performance loss when temperature control is intervened. This significantly improves the temperature control efficiency and power consumption benefits of integrated circuits, adapts to multi-core, multi-cluster heterogeneous integrated circuit architectures, and enhances versatility. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1This is a flowchart illustrating a gradient-first integrated circuit temperature control and power consumption optimization method disclosed in an embodiment of the present invention. Figure 2 This is a flowchart illustrating another gradient-first integrated circuit temperature control and power consumption optimization method disclosed in an embodiment of the present invention. Figure 3 This is a schematic diagram of CPUs 0-3 using the same single-cycle idle time as disclosed in the embodiments of the present invention; Figure 4 This is a schematic diagram of CPUs 0-3 using different single-cycle idle durations as disclosed in the embodiments of the present invention; Figure 5 This is a schematic diagram of the structure of an integrated circuit temperature control and power consumption optimization device based on gradient priority, as disclosed in an embodiment of the present invention. Figure 6 This is a schematic diagram of another integrated circuit temperature control and power consumption optimization device based on gradient priority disclosed in an embodiment of the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0026] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0027] This invention discloses a gradient-priority-based integrated circuit temperature control and power consumption optimization method and apparatus. It achieves gradient-priority temperature control scheduling by uniformly ranking the global energy efficiency coefficients of all operating frequencies of multi-cluster integrated circuits. Based on energy efficiency levels, it accurately locates the clusters to be controlled, achieving maximum power reduction with minimal performance loss when temperature control is implemented. This significantly improves the temperature control efficiency and power consumption benefits of integrated circuits, and is adaptable to multi-core, multi-cluster heterogeneous integrated circuit architectures, enhancing versatility. Detailed descriptions follow.

[0028] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a gradient-first integrated circuit temperature control and power consumption optimization method disclosed in an embodiment of the present invention. Figure 1 The gradient-based integrated circuit temperature control and power consumption optimization method described herein can be applied to integrated circuits, including CPUs, but this invention does not limit the scope of the application. Figure 1 As shown, the gradient-first integrated circuit temperature control and power consumption optimization method may include the following operations: 101. Obtain all operating frequencies of each cluster in the integrated circuit and calculate the energy efficiency coefficient of each operating frequency.

[0029] In this embodiment of the invention, the integrated circuit may optionally include one or more combinations of CPU (Central Processing Unit), GPU (Graphics Processing Unit), DSP (Digital Signal Processor), VPU / JPU (Video Processing Unit / Junctional Processing Unit), ISP (Image Signal Processor), etc. This embodiment uses a CPU as an example for illustration. A cluster in an integrated circuit refers to a group of CPU cores located in the same clock domain and sharing a set of DVFS frequency points. This embodiment uses three clusters of CPUs: a large core cluster (CPUB), a medium core cluster (CPUM), and a small core cluster (CPUS). All operating frequencies of each cluster represent the CPU DVFS frequency points, which are preset fixed frequencies and fixed operating voltages for stable operation within the chip. All CPU cores within the same cluster share the operating frequencies of that cluster. This embodiment uses an example of a large core cluster with m operating frequencies, a medium core cluster with n operating frequencies, and a small core cluster with o operating frequencies. This embodiment is not limited to these specific examples.

[0030] In this embodiment of the invention, optionally, the operating frequency and operating voltage parameters corresponding to each operating frequency point can be obtained, and then the energy efficiency coefficient of that operating frequency point can be calculated based on the operating frequency and operating voltage parameters corresponding to each operating frequency point. Specifically:

[0031] Where E represents the energy efficiency coefficient, and K represents the hardware characteristic constant of the integrated circuit. This indicates the core operating voltage corresponding to the operating frequency point f; The energy efficiency coefficient for each operating frequency point corresponding to the three clusters of CPUs (large, medium, and small) can be expressed as: CPUB: , , ... ; CPUM: , , ... ; CPUS: , , ... ; Within the same CPU cluster, the higher the operating frequency, the greater the required operating voltage, and the energy efficiency coefficient shows a monotonically increasing trend. The lower the operating efficiency of the integrated circuit, the greater the energy consumption ratio. The power consumption cost / benefit brought by increasing / decreasing the unit frequency will be greater. That is, the energy efficiency of high frequency points is much lower than that of low frequency points. Therefore, prioritizing the reduction of the operating frequency of integrated circuits operating at low-efficiency frequency points (high-frequency points) will result in greater temperature control benefits. Prioritizing the increase of the operating frequency of integrated circuits operating at high-efficiency frequency points (low-frequency points) will result in the minimum cost of power consumption / temperature rise.

[0032] 102. Sort each operating frequency point according to its energy efficiency coefficient to obtain a monotonically increasing array of energy efficiency coefficients.

[0033] In this embodiment of the invention, optionally, the energy efficiency coefficients of all operating frequencies in the large, medium, and small kernel clusters are merged, duplicate data are removed, and the values ​​are sorted in ascending order. A unique and continuous sorting index is assigned to each energy efficiency coefficient to form an energy efficiency coefficient array with a total length of m+n+o.

[0034] 103. Determine the current operating frequency point corresponding to each cluster, and determine the sorting index of each current operating frequency point in the energy efficiency coefficient array.

[0035] In this embodiment of the invention, optionally, the current operating frequency of each CPU cluster is determined in real time, and the sorting index of each current operating frequency in the energy efficiency coefficient array is determined. For example, the sorting index of the current operating frequency of the large core cluster is... The sorting index of the current working frequency points of the core cluster is The sorting index of the current operating frequency of the small cluster is .

[0036] 104. Monitor the temperature data of the integrated circuit and determine the corresponding temperature control requirements of the integrated circuit based on the temperature data.

[0037] In this embodiment of the invention, optionally, the temperature data of the integrated circuit can be collected at a fixed period by the built-in temperature sensor or an external temperature sensor, and the temperature control requirements of the integrated circuit can be determined based on the temperature data. The temperature control requirements may include frequency reduction cooling requirements or frequency increase release requirements.

[0038] 105. Based on the temperature control requirements, filter the target current operating frequency corresponding to the temperature control requirements in the energy efficiency coefficient array, determine the target cluster corresponding to the target current operating frequency, and perform temperature control operations on the target cluster.

[0039] In this embodiment of the invention, optionally, according to the frequency reduction or increase requirement, the corresponding target current operating frequency point is selected in the energy efficiency coefficient array, and the target cluster corresponding to the target current operating frequency point is determined. Temperature control operation is then performed on the target cluster. The temperature control operation may include a temperature control operation combining idle injection and frequency adjustment to achieve power consumption optimization and temperature control.

[0040] It is evident that implementation Figure 1 The described gradient-first integrated circuit temperature control and power consumption optimization method can acquire all operating frequencies of each cluster in the integrated circuit, calculate the energy efficiency coefficient of each operating frequency, sort each operating frequency according to its energy efficiency coefficient, obtain a monotonically increasing energy efficiency coefficient array, determine the current operating frequency corresponding to each cluster, and determine the sorting index of each current operating frequency in the energy efficiency coefficient array. It monitors the temperature data of the integrated circuit and determines the corresponding temperature control requirements based on the temperature data. Based on the temperature control requirements, it filters the target current operating frequency corresponding to the temperature control requirements in the energy efficiency coefficient array, determines the target cluster corresponding to the target current operating frequency, and performs temperature control operations on the target cluster. By uniformly sorting the global energy efficiency coefficients of all operating frequencies of multi-cluster integrated circuits, it achieves gradient-first temperature control scheduling, accurately locates the cluster to be controlled based on energy efficiency levels, and achieves maximum power consumption reduction with minimal performance loss when temperature control is intervened. This significantly improves the temperature control efficiency and power consumption benefits of integrated circuits, adapts to multi-core, multi-cluster heterogeneous integrated circuit architectures, and improves versatility.

[0041] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a gradient-first integrated circuit temperature control and power consumption optimization method disclosed in an embodiment of the present invention. Figure 2 The gradient-based integrated circuit temperature control and power consumption optimization method described herein can be applied to integrated circuits, including CPUs, but this invention does not limit the scope of the application. Figure 2As shown, the gradient-first integrated circuit temperature control and power consumption optimization method may include the following operations: 201. Obtain all operating frequencies of each cluster in the integrated circuit and calculate the energy efficiency coefficient of each operating frequency.

[0042] 202. Sort each operating frequency point according to its energy efficiency coefficient to obtain a monotonically increasing array of energy efficiency coefficients.

[0043] 203. Determine the current operating frequency point corresponding to each cluster, and determine the sorting index of each current operating frequency point in the energy efficiency coefficient array.

[0044] 204. Monitor the temperature data of the integrated circuit and determine the temperature change trend based on the preset temperature polling cycle and temperature data.

[0045] In this embodiment of the invention, optionally, at least three sets of real-time temperature data of the integrated circuit can be continuously collected based on a preset temperature polling period, and the temperature values ​​of adjacent periods can be compared: if the later temperature is greater than the previous temperature, it is determined to be a continuous rise; if the later temperature is less than the previous temperature, it is determined to be a continuous fall; if the temperature does not change significantly, it is determined to be stable. The temperature polling period can adopt the standard clock period of the Linux kernel, such as 4ms.

[0046] 205. When the temperature data is greater than the preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

[0047] In this embodiment of the invention, optionally, when the temperature data is greater than a preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit, i.e., the frequency reduction cooling requirement, is determined according to the temperature change trend. The upper temperature threshold represents the highest temperature threshold for safe operation of the integrated circuit; if this threshold is exceeded, a frequency reduction cooling operation must be performed.

[0048] 206. When the temperature data is less than the preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

[0049] In this embodiment of the invention, optionally, when the temperature data is less than a preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit, namely the frequency increase release requirement, is determined according to the temperature change trend. The lower limit threshold represents the lowest temperature threshold at which the performance of the integrated circuit can be restored. If the temperature is lower than the threshold, the frequency increase release operation can be performed.

[0050] 207. Based on the temperature control requirements, filter the target current operating frequency corresponding to the temperature control requirements in the energy efficiency coefficient array, determine the target cluster corresponding to the target current operating frequency, and perform temperature control operations on the target cluster.

[0051] In this embodiment of the invention, it should be noted that for other descriptions of steps 201-203 and step 207, please refer to the detailed description of steps 101-103 and step 105 in Embodiment 1 of the invention. These descriptions will not be repeated in this embodiment of the invention.

[0052] It is evident that implementation Figure 2 The described gradient-first integrated circuit temperature control and power consumption optimization method can acquire all operating frequencies of each cluster in the integrated circuit, calculate the energy efficiency coefficient (EEC) of each operating frequency, sort each operating frequency according to its EEC to obtain a monotonically increasing EEC array, determine the current operating frequency of each cluster, and determine the sort index of each current operating frequency in the EEC array. It monitors the temperature data of the integrated circuit, determines the temperature change trend based on a preset temperature polling period and the temperature data, and determines that the integrated circuit meets the frequency reduction condition when the temperature data is greater than a preset upper temperature threshold and the temperature change trend meets the continuous upward condition. Based on the temperature change trend, it determines the first temperature regulation requirement corresponding to the integrated circuit. Conversely, when the temperature data is less than a preset lower temperature threshold and the temperature change trend meets the continuous downward condition, it determines that the integrated circuit meets the frequency increase condition, and based on the temperature change trend, it determines the second temperature regulation requirement corresponding to the integrated circuit. It can accurately identify temperature control needs through a dual-judgment logic of temperature data and temperature change trends, reducing the probability of false temperature control triggers caused by instantaneous temperature fluctuations, improving the stability and accuracy of temperature control need judgment, and is compatible with the Linux kernel's native temperature control polling mechanism to ensure the real-time performance of temperature control response and the reliability of system operation. It determines the temperature control needs corresponding to the integrated circuit based on temperature data, filters the target current operating frequency corresponding to the temperature control needs from the energy efficiency coefficient array, identifies the target cluster corresponding to the target current operating frequency, and performs temperature control operations on the target cluster. It can achieve gradient priority in temperature control scheduling by uniformly sorting the global energy efficiency coefficients of all operating frequencies of multi-cluster integrated circuits, accurately locates the cluster to be controlled based on energy efficiency levels, and achieves maximum power reduction with minimal performance loss when temperature control intervenes, significantly improving the temperature control efficiency and power consumption benefits of integrated circuits. It is adaptable to multi-core, multi-cluster heterogeneous integrated circuit architectures, improving versatility.

[0053] In an optional embodiment, filtering the target current operating frequency corresponding to the temperature control requirement from the energy efficiency coefficient array based on the temperature control requirement, and determining the target cluster corresponding to the target current operating frequency may include the following operations: When the integrated circuit meets the frequency reduction condition, the first current operating frequency point with the largest sorted index is selected from the energy efficiency coefficient array, and the first cluster to which the first current operating frequency point belongs is determined as the target cluster; When the integrated circuit meets the frequency increase condition, the second current operating frequency point with the smallest sort index is selected from the energy efficiency coefficient array, and the second cluster to which the second current operating frequency point belongs is determined as the target cluster.

[0054] In this optional embodiment, when the integrated circuit meets the frequency reduction condition, the sorting index of the current operating frequency of each CPU cluster is retrieved. , , And calculate the maximum value. The frequency point with the largest sorting index is the first current working frequency point, the CPU cluster mentioned in the first current working frequency point is the first cluster, and the first cluster is determined as the target cluster for frequency reduction.

[0055] In this optional embodiment, when the integrated circuit meets the frequency increase condition, the sorting index of the current operating frequency of each CPU cluster is retrieved. , , And calculate the minimum value. The frequency point with the smallest sorting index is the second current working frequency point, the CPU cluster mentioned in the second current working frequency point is the second cluster, and the second cluster is determined as the target cluster for frequency increase.

[0056] As can be seen, implementing this optional embodiment can, when the integrated circuit meets the frequency reduction condition, select the first current operating frequency point with the largest sorting index in the energy efficiency coefficient array and take its first cluster as the target cluster; when the integrated circuit meets the frequency increase condition, select the second current operating frequency point with the smallest sorting index in the energy efficiency coefficient array and take its second cluster as the target cluster. It can prioritize the regulation of the cluster with the worst energy efficiency and the highest power consumption during frequency reduction to maximize the temperature control power consumption benefit, and prioritize the restoration of the cluster with the best energy efficiency and the smallest temperature rise during frequency increase to restore system performance with the minimum temperature rise cost. It accurately matches the energy efficiency gradient scheduling logic and solves the performance waste problem caused by the indiscriminate frequency reduction of traditional temperature control.

[0057] In another alternative embodiment, performing temperature regulation on the target cluster may include the following operations: Determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster; Determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold. When the target value of frequency regulation is greater than the critical ratio threshold for idle injection, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, a frequency regulation operation is performed on the target cluster.

[0058] In this optional embodiment, the idle injection critical ratio threshold corresponding to the target cluster can be calculated based on the frequency values ​​of the current operating frequency and the adjacent frequency points. This threshold serves as the switching boundary between idle injection and frequency regulation. The idle injection critical ratio threshold represents the maximum performance reduction / release ratio threshold that idle injection can bear. If this value is exceeded, frequency regulation must be switched to achieve a smooth transition of CPU frequency reduction and temperature control / frequency increase and performance release operations.

[0059] In this optional embodiment, the target value for frequency control corresponding to the temperature control requirement is optionally compared with the threshold value for idle injection to determine whether a switch in control mode is needed. When the target value for frequency control corresponding to the temperature control requirement is greater than the threshold value for idle injection, idle injection is performed first to smoothly adjust performance. When idle injection reaches the threshold value, idle injection is stopped and frequency control is performed to avoid sudden performance changes. When the target value for frequency control corresponding to the temperature control requirement is less than or equal to the threshold value for idle injection, only idle injection is performed. Idle injection is a lightweight control that only adjusts the percentage of CPU runtime without changing the operating frequency; frequency control is a heavyweight control that directly switches the frequency point. The combination of the two balances temperature control efficiency and performance experience.

[0060] As can be seen, implementing this optional embodiment can determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster, determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold, and when the frequency control target value is greater than the idle injection critical ratio threshold, first perform an idle injection operation on the target cluster, and then perform a frequency control operation on the target cluster after the operation is completed. This can integrate the dual temperature control methods of idle injection and frequency control, and achieve a smooth switching between the two control methods through the critical ratio threshold, avoiding a sudden change in performance when temperature control is involved, and improving the smoothness of system operation and user experience.

[0061] In yet another alternative embodiment, performing an idle injection operation on the target cluster may include the following operations: Determine the number of cores in the target cluster, the idle injection allocation rules for each core, and the preset single-cycle idle duration; The idle cycle period is calculated based on the number of cores, the idle injection allocation rules for each core, and the idle duration of a single cycle. Based on the idle cycle period, the idle duration of a single cycle, and the idle injection allocation rules, perform idle injection operations on the target cluster.

[0062] In this optional embodiment, the number of CPU cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration can be determined. The single-cycle idle duration can be preset by the system, for example, fixed at 4ms, corresponding to 250Hz.

[0063] In this optional embodiment, the idle cycle period can be calculated based on the number of cores, the idle injection allocation rule for each core, and the single-cycle idle duration. For example, when the target cluster has 4 CPU cores, the single-cycle idle duration is 4ms, the target idle injection ratio is 25%, and the idle injection allocation rule is to inject the same single-cycle idle duration on average, then the idle cycle period = single-cycle idle duration 4ms / target idle injection ratio 25% = 16ms. Please refer to [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of CPUs 0-3 using the same single-cycle idle time as disclosed in an embodiment of the present invention. CPUs 0-3 all use the same single-cycle idle time, i.e., each is idle for 4ms. The calculated idle cycle is 16ms. Therefore, each CPU core is idle for 4ms and then works for 12ms. During this time, the target idle injection ratio of the target cluster is controlled at 25%.

[0064] In this optional embodiment, when the idle injection allocation rule is to inject different single-cycle idle durations into different CPU cores, taking as an example that the single-cycle idle duration injected into CPU0-1 is 50% of the CPU0-1 cycle, and the single-cycle idle duration injected into CPU2-3 is 0, then (50%+50%+0+0) / number of CPU cores 4 = 25%. That is, the target idle injection ratio of this allocation scheme is still controlled at 25%. At this time, the idle cycle period = single-cycle idle duration is 4ms / 50% = 8ms. Please refer to [link / reference]. Figure 4 , Figure 4 This is a schematic diagram illustrating that CPUs 0-3 employ different single-cycle idle durations as disclosed in an embodiment of the present invention. Specifically, CPUs 0-1 are idle for 4ms, then work for 4ms, while CPUs 2-3 do not perform idle injection operations.

[0065] In this optional embodiment, optionally, an idle injection operation can be performed on the target cluster based on the idle cycle period, the idle duration of a single cycle, and the idle injection allocation rules. Specifically: If the idle injection allocation rule is followed to inject the same single-cycle idle duration on average, such as... Figure 3 As shown, CPUs 0-3 all use the same single-cycle idle time, i.e., 4ms of idle time. The calculated idle cycle is 16ms. Therefore, each CPU core is idle for 4ms and then works for 12ms. At this time, the target idle injection ratio of the target cluster is controlled at 25%. If different single-cycle idle durations are injected into different CPU cores according to the idle injection allocation rules, taking as an example that the single-cycle idle duration injected into CPU0-1 is 50% of the CPU0-1 cycle, and the single-cycle idle duration injected into CPU2-3 is 0, then... Figure 4 As shown, CPU0-1 is idle for 4ms, then works for 4ms. CPU2-3 does not perform idle injection operations. During this time, the target idle injection ratio of the target cluster is still controlled at 25%. Furthermore, the target idle injection ratio is less than or equal to the idle injection critical ratio threshold.

[0066] As can be seen, implementing this optional embodiment can determine the number of cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration. Based on the number of cores, the idle injection allocation rule, and the single-cycle idle duration, the idle cycle period is calculated, and the idle injection operation is performed on the target cluster according to the idle cycle period, the single-cycle idle duration, and the idle injection allocation rule. This can flexibly adapt to integrated circuit cluster architectures with different numbers of cores and load characteristics, accurately control the computing power reduction ratio of idle injection, ensure the stability and controllability of idle injection execution, and optimize the task scheduling order in the temperature control process.

[0067] In yet another optional embodiment, determining the idle injection critical ratio threshold corresponding to the target current operating frequency of the target cluster may include the following operations: Determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency; Calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster based on the ratio.

[0068] In this optional embodiment, a first frequency value of the target's current operating frequency point may be determined. And the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency point. Then calculate the ratio between the second frequency value and the first frequency value. Then, based on the ratio, determine the idle injection critical ratio threshold corresponding to the current working frequency of the target cluster.

[0069] In this optional embodiment, optionally, the idle injection critical ratio threshold corresponding to the previous operating frequency point. The calculation formula is:

[0070] For example, if the first frequency value of the target's current operating frequency is 2.0 GHz, and the second frequency value of the adjacent frequency range is 1.5 GHz, then... .

[0071] As can be seen, implementing this optional embodiment can determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency points adjacent to the target's current operating frequency, calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the target cluster's current operating frequency based on the ratio. It can dynamically calculate the critical ratio threshold based on the frequencies of adjacent frequency points without the need for static preset parameters, adapt to the full DVFS operating frequency points of integrated circuits, and improve the accuracy of critical threshold calculation and the versatility of the solution.

[0072] In another optional embodiment, when the frequency modulation target value is greater than the idle injection critical ratio threshold, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, the frequency modulation operation on the target cluster may include the following operations: When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval; When the first idle injection ratio equals the idle injection critical ratio threshold, the idle injection operation is determined to be completed, and a frequency reduction operation is performed on the target cluster. When the target value of frequency regulation is greater than the threshold of the idle injection critical ratio, and when the integrated circuit meets the frequency increase condition, an idle injection operation is performed on the target cluster, and the second idle injection ratio is reduced according to a preset time interval. When the second idle injection ratio is lower than the preset low threshold of the idle injection ratio, it is determined that the idle injection operation has been completed, and the frequency increase operation is performed on the target cluster.

[0073] In this optional embodiment, when the frequency control target value is greater than the idle injection critical ratio threshold and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval. The first idle injection ratio represents the idle injection ratio under the frequency reduction scenario, which is gradually increased with the temperature control requirements. Specifically, the first idle injection ratio is increased by increasing the value of the single-cycle idle duration. When the first idle injection ratio is equal to the idle injection critical ratio threshold, it is determined that the idle injection operation has been completed, the idle injection is stopped, a frequency reduction operation is performed on the target cluster, the operating frequency is switched to a lower level, and the sorting index of the current operating frequency of the target cluster is updated to complete the frequency reduction control.

[0074] In this optional embodiment, when the frequency control target value is greater than the idle injection critical ratio threshold and when the integrated circuit meets the frequency increase conditions, an idle injection operation is performed on the target cluster. The second idle injection ratio is gradually reduced at preset time intervals to smoothly restore the performance of the target cluster. The second idle injection ratio represents the idle injection ratio under the frequency increase scenario, which is gradually reduced with temperature control requirements. Specifically, the first idle injection ratio is reduced by decreasing the value of the single-cycle idle duration (idle). When the first idle injection ratio is lower than a preset low threshold (e.g., 0%), it is determined that the idle injection operation has been completed, the idle injection is stopped, a frequency increase operation is performed on the target cluster, the operating frequency is switched to a higher level, the sorting index of the current operating frequency of the target cluster is updated, and the frequency increase control is completed.

[0075] As can be seen, implementing this optional embodiment can gradually increase the first idle injection ratio to equal the idle injection critical ratio threshold when the frequency reduction condition is met and the frequency control target value is greater than the critical ratio threshold, and then perform the frequency reduction operation. When the frequency increase condition is met and the frequency control target value is greater than the critical ratio threshold, the second idle injection ratio is gradually reduced to below the preset low threshold, and then the frequency increase operation is performed. This can achieve bidirectional smooth control of the entire process of temperature control intervention and withdrawal, making performance changes more linear and controllable, reducing temperature control stuttering problems, and further enhancing the temperature control effect and system stability of energy efficiency gradient scheduling.

[0076] Example 3 Please see Figure 5 , Figure 5 This is a schematic diagram of a gradient-first integrated circuit temperature control and power consumption optimization device disclosed in an embodiment of the present invention. Figure 5 The gradient-based integrated circuit temperature control and power consumption optimization device described herein can be applied to integrated circuits, which may include CPUs; however, this invention does not limit the scope of the application. Figure 5 As shown, the gradient-first integrated circuit temperature control and power consumption optimization device may include: The acquisition module 301 is used to acquire all operating frequency points of each cluster in the integrated circuit and calculate the energy efficiency coefficient of each operating frequency point; The sorting module 302 is used to sort each operating frequency point according to the energy efficiency coefficient of each operating frequency point to obtain a monotonically increasing energy efficiency coefficient array. The determination module 303 is used to determine the current operating frequency point corresponding to each cluster and to determine the sorting index of each current operating frequency point in the energy efficiency coefficient array; The monitoring module 304 is used to monitor the temperature data of the integrated circuit and determine the corresponding temperature control requirements of the integrated circuit based on the temperature data. The filtering module 305 is used to filter the target current operating frequency corresponding to the temperature control requirement in the energy efficiency coefficient array according to the temperature control requirement, determine the target cluster corresponding to the target current operating frequency, and perform temperature control operation on the target cluster.

[0077] It is evident that implementation Figure 5 The described gradient-first integrated circuit temperature control and power consumption optimization device can acquire all operating frequencies of each cluster in the integrated circuit, calculate the energy efficiency coefficient of each operating frequency, sort each operating frequency according to its energy efficiency coefficient to obtain a monotonically increasing energy efficiency coefficient array, determine the current operating frequency corresponding to each cluster, and determine the sorting index of each current operating frequency in the energy efficiency coefficient array. It monitors the temperature data of the integrated circuit and determines the temperature control requirements corresponding to the integrated circuit based on the temperature data. Based on the temperature control requirements, it filters the target current operating frequency corresponding to the temperature control requirements in the energy efficiency coefficient array, determines the target cluster corresponding to the target current operating frequency, and performs temperature control operations on the target cluster. By uniformly sorting the global energy efficiency coefficients of all operating frequencies of multi-cluster integrated circuits, it achieves gradient-first temperature control scheduling, accurately locates the cluster to be controlled based on energy efficiency levels, and achieves maximum power consumption reduction with minimal performance loss when temperature control is intervened. This significantly improves the temperature control efficiency and power consumption benefits of integrated circuits, adapts to multi-core, multi-cluster heterogeneous integrated circuit architectures, and improves versatility.

[0078] In an optional embodiment, such as Figure 5 As shown, the specific methods by which the monitoring module 304 determines the temperature control requirements of the integrated circuit based on the temperature data include: The temperature change trend is determined based on the preset temperature polling cycle and temperature data. When the temperature data is greater than the preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend. When the temperature data is less than the preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

[0079] It is evident that implementation Figure 5The described gradient-first integrated circuit temperature control and power consumption optimization device can acquire all operating frequencies of each cluster in the integrated circuit, calculate the energy efficiency coefficient (EEC) of each operating frequency, sort each operating frequency according to its EEC to obtain a monotonically increasing EEC array, determine the current operating frequency of each cluster, and determine the sort index of each current operating frequency in the EEC array. It monitors the temperature data of the integrated circuit, determines the temperature change trend based on a preset temperature polling period and the temperature data, and determines that the integrated circuit meets the frequency reduction condition when the temperature data is greater than a preset upper temperature threshold and the temperature change trend meets the continuous upward condition. Based on the temperature change trend, it determines the first temperature regulation requirement corresponding to the integrated circuit. When the temperature data is less than a preset lower temperature threshold and the temperature change trend meets the continuous downward condition, it determines that the integrated circuit meets the frequency increase condition, and based on the temperature change trend, it determines the second temperature regulation requirement corresponding to the integrated circuit. It can accurately identify temperature control needs through a dual-judgment logic of temperature data and temperature change trends, reducing the probability of false temperature control triggers caused by instantaneous temperature fluctuations, improving the stability and accuracy of temperature control need judgment, and is compatible with the Linux kernel's native temperature control polling mechanism to ensure the real-time performance of temperature control response and the reliability of system operation. It determines the temperature control needs corresponding to the integrated circuit based on temperature data, filters the target current operating frequency corresponding to the temperature control needs from the energy efficiency coefficient array, identifies the target cluster corresponding to the target current operating frequency, and performs temperature control operations on the target cluster. It can achieve gradient priority in temperature control scheduling by uniformly sorting the global energy efficiency coefficients of all operating frequencies of multi-cluster integrated circuits, accurately locates the cluster to be controlled based on energy efficiency levels, and achieves maximum power reduction with minimal performance loss when temperature control intervenes, significantly improving the temperature control efficiency and power consumption benefits of integrated circuits. It is adaptable to multi-core, multi-cluster heterogeneous integrated circuit architectures, improving versatility.

[0080] In another alternative embodiment, such as Figure 5 As shown, the filtering module 305 filters the target current operating frequency points corresponding to the temperature control requirements from the energy efficiency coefficient array according to the temperature control requirements, and determines the target clusters corresponding to the target current operating frequency points in the following specific ways: When the integrated circuit meets the frequency reduction condition, the first current operating frequency point with the largest sorted index is selected from the energy efficiency coefficient array, and the first cluster to which the first current operating frequency point belongs is determined as the target cluster; When the integrated circuit meets the frequency increase condition, the second current operating frequency point with the smallest sort index is selected from the energy efficiency coefficient array, and the second cluster to which the second current operating frequency point belongs is determined as the target cluster.

[0081] It is evident that implementation Figure 5The described gradient-priority integrated circuit temperature control and power consumption optimization device can, when the integrated circuit meets the frequency reduction condition, select the first current operating frequency point with the largest sorted index in the energy efficiency coefficient array and take its first cluster as the target cluster; when the integrated circuit meets the frequency increase condition, select the second current operating frequency point with the smallest sorted index in the energy efficiency coefficient array and take its second cluster as the target cluster. It can prioritize the regulation of the cluster with the worst energy efficiency and the highest power consumption during frequency reduction to maximize the temperature control power consumption benefit, and prioritize the restoration of the cluster with the best energy efficiency and the smallest temperature rise during frequency increase to restore system performance with the minimum temperature rise cost. It accurately matches the energy efficiency gradient scheduling logic and solves the performance waste problem caused by the indiscriminate frequency reduction of traditional temperature control.

[0082] In yet another alternative embodiment, such as Figure 5 As shown, the specific methods by which the screening module 305 performs temperature control operations on the target cluster include: Determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster; Determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold. When the target value of frequency regulation is greater than the critical ratio threshold for idle injection, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, a frequency regulation operation is performed on the target cluster.

[0083] It is evident that implementation Figure 5 The described gradient-first integrated circuit temperature control and power consumption optimization device can determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster, and determine whether the frequency regulation target value corresponding to the temperature regulation requirement is greater than the idle injection critical ratio threshold. When the frequency regulation target value is greater than the idle injection critical ratio threshold, an idle injection operation is first performed on the target cluster, and then a frequency regulation operation is performed on the target cluster. This device can integrate the dual temperature control methods of idle injection and frequency regulation, and achieve a smooth switching between the two regulation methods through the critical ratio threshold, avoiding a sudden change in performance when temperature control is involved, and improving the smoothness of system operation and user experience.

[0084] In yet another alternative embodiment, such as Figure 5 As shown, the specific methods by which the filtering module 305 performs the idle injection operation on the target cluster include: Determine the number of cores in the target cluster, the idle injection allocation rules for each core, and the preset single-cycle idle duration; The idle cycle period is calculated based on the number of cores, the idle injection allocation rules for each core, and the idle duration of a single cycle. Based on the idle cycle period, the idle duration of a single cycle, and the idle injection allocation rules, perform idle injection operations on the target cluster.

[0085] It is evident that implementation Figure 5 The described gradient-first integrated circuit temperature control and power consumption optimization device can determine the number of cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration. It calculates the idle cycle period based on the number of cores, the idle injection allocation rule, and the single-cycle idle duration, and performs idle injection operations on the target cluster according to the idle cycle period, the single-cycle idle duration, and the idle injection allocation rule. It can flexibly adapt to integrated circuit cluster architectures with different numbers of cores and load characteristics, accurately control the computing power reduction ratio of idle injection, ensure the stability and controllability of idle injection execution, and optimize the task scheduling order in the temperature control process.

[0086] In yet another alternative embodiment, such as Figure 5 As shown, the specific method by which the filtering module 305 determines the idle injection critical ratio threshold corresponding to the current operating frequency point of the target cluster includes: Determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency; Calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster based on the ratio.

[0087] It is evident that implementation Figure 5 The described gradient-first integrated circuit temperature control and power consumption optimization device can determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency points adjacent to the target's current operating frequency, calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the target cluster's current operating frequency based on the ratio. It can dynamically calculate the critical ratio threshold based on the frequencies of adjacent frequency points without the need for static preset parameters, adapt to the full DVFS operating frequency points of integrated circuits, and improve the accuracy of critical threshold calculation and the versatility of the solution.

[0088] In yet another alternative embodiment, such as Figure 5 As shown, when the frequency control target value is greater than the idle injection critical ratio threshold, the screening module 305 performs an idle injection operation on the target cluster, and after completing the idle injection operation, performs a frequency control operation on the target cluster in the following specific ways: When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval; When the first idle injection ratio equals the idle injection critical ratio threshold, the idle injection operation is determined to be completed, and a frequency reduction operation is performed on the target cluster. When the target value of frequency regulation is greater than the threshold of the idle injection critical ratio, and when the integrated circuit meets the frequency increase condition, an idle injection operation is performed on the target cluster, and the second idle injection ratio is reduced according to a preset time interval. When the second idle injection ratio is lower than the preset low threshold of the idle injection ratio, it is determined that the idle injection operation has been completed, and the frequency increase operation is performed on the target cluster.

[0089] It is evident that implementation Figure 5 The described gradient-first integrated circuit temperature control and power consumption optimization device can gradually increase the first idle injection ratio to equal the idle injection critical ratio threshold and then perform a frequency reduction operation when the frequency reduction condition is met and the frequency regulation target value is greater than the critical ratio threshold. When the frequency increase condition is met and the frequency regulation target value is greater than the critical ratio threshold, it gradually decreases the second idle injection ratio to below the preset low threshold and then performs a frequency increase operation. This enables bidirectional smooth regulation of the entire process of temperature control intervention and withdrawal, making performance changes more linear and controllable, reducing temperature control stuttering problems, and further enhancing the temperature control effect and system stability of energy efficiency gradient scheduling.

[0090] Example 4 Please see Figure 6 , Figure 6 This is a schematic diagram of another integrated circuit temperature control and power consumption optimization device based on gradient priority disclosed in an embodiment of the present invention. Figure 6 As shown, the gradient-first integrated circuit temperature control and power consumption optimization device may include: Memory 401 storing executable program code; Processor 402 coupled to memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the gradient-first integrated circuit temperature control and power consumption optimization method described in Embodiment 1 or Embodiment 2 of the present invention.

[0091] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in any of the gradient-first integrated circuit temperature control and power consumption optimization methods disclosed in Embodiment 1 of this invention.

[0092] Example 6 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the gradient-first integrated circuit temperature control and power consumption optimization method described in Embodiment 1 or Embodiment 2.

[0093] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0094] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0095] Finally, it should be noted that the gradient-first integrated circuit temperature control and power consumption optimization method and apparatus disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A gradient-first integrated circuit temperature control and power consumption optimization method, characterized in that, The method includes: Obtain all operating frequency points of each cluster in the integrated circuit, and calculate the energy efficiency coefficient of each operating frequency point; Sort each operating frequency point according to its energy efficiency coefficient to obtain a monotonically increasing array of energy efficiency coefficients; Determine the current operating frequency point corresponding to each cluster, and determine the sorting index of each current operating frequency point in the energy efficiency coefficient array; Monitor the temperature data of the integrated circuit, and determine the temperature control requirements of the integrated circuit based on the temperature data; Based on the temperature control requirements, the target current operating frequency corresponding to the temperature control requirements is selected from the energy efficiency coefficient array, and the target cluster corresponding to the target current operating frequency is determined, and the temperature control operation is performed on the target cluster.

2. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 1, characterized in that, The step of determining the temperature control requirements corresponding to the integrated circuit based on the temperature data includes: The temperature change trend is determined based on the preset temperature polling cycle and the temperature data. When the temperature data is greater than the preset upper temperature threshold and the temperature change trend meets the condition of continuous increase, it is determined that the integrated circuit meets the frequency reduction condition, and the first temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend. When the temperature data is less than the preset lower limit threshold and the temperature change trend meets the condition of continuous decrease, it is determined that the integrated circuit meets the frequency increase condition, and the second temperature control requirement corresponding to the integrated circuit is determined according to the temperature change trend.

3. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 2, characterized in that, The step of filtering the target current operating frequency point corresponding to the temperature control requirement from the energy efficiency coefficient array according to the temperature control requirement, and determining the target cluster corresponding to the target current operating frequency point, includes: When the integrated circuit meets the frequency reduction condition, the first current operating frequency point with the largest sort index is selected from the energy efficiency coefficient array, and the first cluster to which the first current operating frequency point belongs is determined as the target cluster; When the integrated circuit meets the frequency increase condition, the second current operating frequency point with the smallest sorting index is selected from the energy efficiency coefficient array, and the second cluster to which the second current operating frequency point belongs is determined as the target cluster.

4. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 2 or 3, characterized in that, The temperature control operation performed on the target cluster includes: Determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster; Determine whether the frequency control target value corresponding to the temperature control requirement is greater than the idle injection critical ratio threshold; When the frequency control target value is greater than the idle injection critical ratio threshold, an idle injection operation is performed on the target cluster, and after the idle injection operation is completed, a frequency control operation is performed on the target cluster.

5. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 4, characterized in that, The step of performing an idle injection operation on the target cluster includes: The number of cores in the target cluster, the idle injection allocation rule for each core, and the preset single-cycle idle duration are determined. The idle cycle period is calculated based on the number of cores, the idle injection allocation rule for each core, and the single-cycle idle duration. Based on the idle cycle period, the single-cycle idle duration, and the idle injection allocation rules, an idle injection operation is performed on the target cluster.

6. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 4, characterized in that, The step of determining the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster includes: Determine the first frequency value of the target's current operating frequency and the second frequency value of the adjacent gear frequency point adjacent to the target's current operating frequency; Calculate the ratio between the second frequency value and the first frequency value, and determine the idle injection critical ratio threshold corresponding to the current operating frequency of the target cluster based on the ratio.

7. The gradient-first integrated circuit temperature control and power consumption optimization method according to claim 4, characterized in that, The step of performing an idle injection operation on the target cluster when the frequency modulation target value is greater than the idle injection critical ratio threshold, and performing a frequency modulation operation on the target cluster after the idle injection operation is completed, includes: When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency reduction condition, an idle injection operation is performed on the target cluster, and the first idle injection ratio is increased according to a preset time interval; When the first idle injection ratio is equal to the idle injection critical ratio threshold, it is determined that the idle injection operation has been completed, and a frequency reduction operation is performed on the target cluster. When the frequency control target value is greater than the idle injection critical ratio threshold, and when the integrated circuit meets the frequency increase condition, an idle injection operation is performed on the target cluster, and the second idle injection ratio is reduced according to a preset time interval; When the second idle injection ratio is lower than the preset low threshold of the idle injection ratio, it is determined that the idle injection operation has been completed, and the frequency increase operation is performed on the target cluster.

8. A gradient-first integrated circuit temperature control and power consumption optimization device, characterized in that, The device includes: The acquisition module is used to acquire all operating frequency points of each cluster in the integrated circuit and calculate the energy efficiency coefficient of each operating frequency point; The sorting module is used to sort each operating frequency point according to the energy efficiency coefficient of each operating frequency point to obtain a monotonically increasing energy efficiency coefficient array. The determination module is used to determine the current operating frequency point corresponding to each cluster, and to determine the sorting index of each current operating frequency point in the energy efficiency coefficient array; A monitoring module is used to monitor the temperature data of the integrated circuit and determine the temperature control requirements of the integrated circuit based on the temperature data. The filtering module is used to filter the target current operating frequency corresponding to the temperature control requirement in the energy efficiency coefficient array according to the temperature control requirement, determine the target cluster corresponding to the target current operating frequency, and perform temperature control operation on the target cluster.

9. A gradient-first integrated circuit temperature control and power consumption optimization device, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the gradient-first integrated circuit temperature control and power consumption optimization method as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the gradient-first integrated circuit temperature control and power consumption optimization method as described in any one of claims 1-7.

Images