# US 11,445,631 B2 · High performance computing based holistic liquid cooled rack cost optimization > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,445,631 B2 - Document ID: US11445631B2 - Status: Granted - Kind: B2 - Application number: 16/823,685 - Filed: 2020-03-19 - Published: 2021-09-23 - Granted: 2022-09-13 - Earliest claimed: 2020-03-19 - Assignees: Baidu USA LLC - Applicants: Baidu USA LLC - Inventors: Shuai Shao; Tianyi Gao - Web page: https://patentgrep.com/patent/US11445631B2 ## CPC - H05K 7/1487: {Blade assemblies, e.g. blade cases or inner arrangements within a blade} - G06F 1/206: {comprising thermal management} - G06F 1/3228: Monitoring task completion, e.g. by use of idle timers, stop commands or wait commands - G06F 1/324: by lowering clock frequency - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 20/00: Machine learning - H05K 7/1488: {Cabinets therefor, e.g. chassis or racks or mechanical interfaces between blades and support structures} - H05K 7/20272: {Accessories for moving fluid, for expanding fluid, for connecting fluid conduits, for distributing fluid, for removing gas or for preventing leakage, e.g. pumps, tanks or manifolds} - H05K 7/20281: {Thermal management, e.g. liquid flow control} - H05K 7/20772: {within server blades for removing heat from heat source} - H05K 7/20781: {within cabinets for removing heat from server blades} - G06F 2200/201: Cooling arrangements using cooling fluid - H05K 7/207: {Thermal management, e.g. cabinet temperature control} - H05K 7/20836: {Thermal management, e.g. server temperature control} - Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management - H05K: PRINTED CIRCUITS; CASINGS OR CONSTRUCTIONAL DETAILS OF ELECTRIC APPARATUS; MANUFACTURE OF ASSEMBLAGES OF ELECTRICAL COMPONENTS - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - 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 ## Abstract An electronic rack includes an array of server blades arranged in a stack. Each server blade contains one or more servers and each server includes one or more processors to provide data processing services. The electronic rack includes a coolant distribution unit (CDU) and a rack management unit (RMU). The CDU supplies cooling liquid to the processors and receives the cooling liquid carrying heat from the processors. The CDU includes a liquid pump to pump the cooling liquid. The RMU is configured to manage the operations of the components within the electronic rack such as CDU, etc. The RMU includes control logic to determine an optimal pump speed and an optimal processor clock rate based on a first relationship between cost of the electronic rack and clock rate of the processors and a second relationship between energy consumption of the electronic rack, clock rate of the processors and pump speed. ## Claims ### Claim 1 An electronic rack of a data center, comprising: a plurality of server blades arranged in a stack, each server blades containing one or more servers and each server including one or more processors to provide data processing services; a coolant distribution unit (CDU) to supply cooling liquid to the processors and to receive the cooling liquid carrying heat exchanged from the processors, wherein the CDU includes a liquid pump to pump the cooling liquid and a pump controller to control a pump speed of the liquid pump; and a rack management unit (RMU) coupled to the server blades and the CDU, wherein the RMU includes control logic configured to determine a first relationship between a cost of each server and a clock rate of the one or more processors of the server based on a performance requirement of the electronic rack, determine a second relationship between energy consumption of the server to perform a defined task, the clock rate of processors of the server, and a pump speed of the liquid pump, determine an optimal clock rate and an optimal pump speed based on the first relationship and the second relationship, and control the pump speed of the liquid pump via the pump controller based on the optimal pump speed and control the optimal clock rate of the one or more processors based on the optimal clock rate. ### Claim 2 (depends on 1) The electronic rack of claim 1, wherein determining the optimized clock rate and optimized pump speed comprises optimizing an objective function such that the objective function reaches a minimum, while a set of predetermined constraints is satisfied. ### Claim 3 (depends on 2) The electronic rack of claim 2, wherein the set of predetermined constraints comprises: a first condition that a clock speed is within a predetermined clock speed range; and a second condition that the pump speed of the liquid pump is within a predetermined pump speed range. ### Claim 4 (depends on 2) The electronic rack of claim 2, wherein the objective function indicates a total cost of the electronic rack to train a machine learning model based on at least one of server acquisition cost, pump system acquisition cost, server power consumption cost, or pump system power consumption cost. ### Claim 5 (depends on 2) The electronic rack of claim 2, wherein the objective function indicates a total cost of the electronic rack over an estimated lifetime of the electronic rack. ### Claim 6 (depends on 1) The electronic rack of claim 1, wherein the optimized clock rate and optimized pump speed are used during training of an artificial intelligence (AI) model to minimize cost of training the AI model. ### Claim 7 (depends on 1) The electronic rack of claim 1, wherein determining the second relationship is further based on an operating temperature of the one or more processors. ### Claim 8 A method for determining optimal operating parameters for an electronic rack of a data center, the method comprising: determining, for an electronic rack having a plurality of server blades, each server blade including one or more servers and each server including one or more processors to provide data processing services, a first relationship between a cost of the electronic rack and a clock rate of the one or more processors of the electronic rack based on a performance requirement of the electronic rack; determining a second relationship between energy consumption of the electronic rack to perform a defined task, a clock rate of one or more processors, and a pump speed of a liquid pump of a coolant distribution unit (CDU) of the electronic rack; determining an optimized clock rate and an optimized pump speed based on the first relationship and the second relationship; and controlling the pump speed of the liquid pump via a pump controller based on the optimal pump speed and controlling the clock rate of the one or more processors based on the optimal clock rate. ### Claim 9 (depends on 8) The method of claim 8, wherein determining the optimized clock rate and optimized pump speed comprises optimizing an objective function such that the objective function reaches a minimum, while a set of predetermined constraints is satisfied. ### Claim 10 (depends on 9) The method of claim 9, wherein the set of predetermined constraints comprises: a first condition that a clock speed is within a predetermined clock speed range; and a second condition that the pump speed of the liquid pump is within a predetermined pump speed range. ### Claim 11 (depends on 9) The method of claim 9, wherein the objective function indicates a total cost of the electronic rack to train a machine learning model based on at least one of server acquisition cost, pump system acquisition cost, server power consumption cost, or pump system power consumption cost. ### Claim 12 (depends on 9) The method of claim 9, wherein the objective function indicates a total cost of the electronic rack over an estimated lifetime of the electronic rack. ### Claim 13 (depends on 8) The method of claim 8, wherein the optimized clock rate and optimized pump speed are used during training of an artificial intelligence (AI) model to minimize cost of training the AI model. ### Claim 14 (depends on 8) The method of claim 8, wherein determining the second relationship further is based on an operating temperature of the one or more processors. ### Claim 15 A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations of determining optimal operating parameters for an electronic rack of a data center, the operations comprising: determining, for an electronic rack having a plurality of server blades, each server blade including one or more servers and each server including one or more processors to provide data processing services, a first relationship between a cost of the electronic rack and a clock rate of the one or more processors of the electronic rack based on a performance requirement of the electronic rack; determining a second relationship between energy consumption of the electronic rack to perform a defined task, the clock rate of one or more processors, and a pump speed of a liquid pump of a coolant distribution unit (CDU) of the electronic rack; determining an optimized clock rate and an optimized pump speed based on the first relationship and the second relationship; and controlling the pump speed of the liquid pump via the pump controller based on the optimal pump speed and controlling the clock rate of the one or more processors based on the optimal clock rate. ### Claim 16 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein determining the optimized clock rate and optimized pump speed comprises optimizing an objective function such that the objective function reaches a minimum, while a set of predetermined constraints is satisfied. ### Claim 17 (depends on 16) The non-transitory machine-readable medium of claim 16, wherein the set of predetermined constraints comprises: a first condition that a clock speed is within a predetermined clock speed range; and a second condition that the pump speed of the liquid pump is within a predetermined pump speed range. ### Claim 18 (depends on 16) The non-transitory machine-readable medium of claim 16, wherein the objective function indicates a total cost of the electronic rack to train a machine learning model based on at least one of server acquisition cost, pump system acquisition cost, server power consumption cost, or pump system power consumption cost. ### Claim 19 (depends on 16) The non-transitory machine-readable medium of claim 16, wherein the objective function indicates a total cost of the electronic rack over an estimated lifetime of the electronic rack. ### Claim 20 (depends on 15) The non-transitory machine-readable medium of claim 15, wherein the optimized clock rate and optimized pump speed are used during training of an artificial intelligence (AI) model to minimize cost of training the AI model.