# US 2019/0377548 A1 · Arithmetic processing apparatus, control method, and recording medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2019/0377548 A1 - Document ID: US20190377548A1 - Status: Published application - Kind: A1 - Application number: 16/401,128 - Filed: 2019-05-02 - Published: 2019-12-12 - Earliest claimed: 2018-06-06 - Assignees: Fujitsu Limited - Applicants: Fujitsu Limited - Inventors: Manabu Matsuyama - Web page: https://patentgrep.com/patent/US20190377548A1 ## CPC - G06F 7/485: Adding; Subtracting {(G06F7/4833, G06F7/4836 take precedence)} - G06F 7/487: Multiplying; Dividing {(G06F7/4833, G06F7/4836 take precedence)} - G06F 7/4876: {Multiplying} - G06F 2207/3824: Accepting both fixed-point and floating-point numbers - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract An arithmetic processing apparatus includes a memory; and a processor coupled to the memory and configured to: acquire input data and output data in each of a plurality of operations using predetermined data for an information processing including the plurality of operations of a floating-point format to, extract a specific operation represented by a complicated function including at least a transcendental function among the plurality of operations, obtain an alternative function having a smaller computation amount than the complicated function in the extracted specific operation based on the input data, and substitute the specific operation in the information processing with the alternative function. ## Claims ### Claim 1 An arithmetic processing apparatus comprising: a memory; and a processor coupled to the memory and configured to: acquire input data and output data in each of a plurality of operations using predetermined data for an information processing including the plurality of operations of a floating-point format to, extract a specific operation represented by a complicated function including at least a transcendental function among the plurality of operations, obtain an alternative function having a smaller computation amount than the complicated function in the extracted specific operation based on the input data, and substitute the specific operation in the information processing with the alternative function. ### Claim 2 (depends on 1) The arithmetic processing apparatus according to claim 1, wherein the processor is configured to: obtain an approximation expression of the complicated function based on the maximum value, the minimum value, and a median of the input data, and substitute the approximation expression with the alternative function when an error between the approximation expression and the complicated function falls within an allowable range. ### Claim 3 (depends on 2) The arithmetic processing apparatus according to claim 2, wherein the processor is configured to: obtain a linear approximation expression using least squares method based on the maximum value, the minimum value, and the median, and substitute the linear approximation expression with the alternative function when an error between the linear approximation expression and the complicated function falls within an allowable range. ### Claim 4 (depends on 1) The arithmetic processing apparatus according to claim 1, wherein the processor is configured to: hold in advance a correspondence table between information on the input data and an approximation expression, obtain the approximation expression corresponding to the complicated function from the correspondence table based on the input data, and substitute the obtained approximation expression with the alternative function. ### Claim 5 (depends on 1) The arithmetic processing apparatus according to claim 1, wherein the processor is configured to: specify a transformable operation in which the input data and the output data fall within a predetermined range, among the plurality of operations, transform the transformable operation into an operation of a fixed-point format, adjust input/output of an operation other than the transformable operation, among the plurality of operations, adjust the final output of the information processing, extract an operation that is the transformable operation and the complicated function, as the specific operation. ### Claim 6 (depends on 5) The arithmetic processing apparatus according to claim 5, wherein the processor is configured to specify an operation in which the maximum value, the minimum value, and the distribution frequency of the input data and the output data fall within the predetermined range, as the transformable operation. ### Claim 7 A control method executed by a processor included in an arithmetic processing apparatus, the method comprising: acquiring input data and output data in each of a plurality of operations using predetermined data for an information processing including the plurality of operations of a floating-point format to; extracting a specific operation represented by a complicated function including at least a transcendental function among the plurality of operations; obtaining an alternative function having a smaller computation amount than the complicated function in the extracted specific operation based on the input data; and substituting the specific operation in the information processing with the alternative function. ### Claim 8 (depends on 7) The control method according to claim 7, the method further comprising: obtaining an approximation expression of the complicated function based on the maximum value, the minimum value, and a median of the input data; and substituting the approximation expression with the alternative function when an error between the approximation expression and the complicated function falls within an allowable range. ### Claim 9 A non-transitory computer-readable recording medium storing a program that causes a processor included in an arithmetic processing apparatus to execute a process, the process comprising: acquiring input data and output data in each of a plurality of operations using predetermined data for an information processing including the plurality of operations of a floating-point format to; extracting a specific operation represented by a complicated function including at least a transcendental function among the plurality of operations; obtaining an alternative function having a smaller computation amount than the complicated function in the extracted specific operation based on the input data; and substituting the specific operation in the information processing with the alternative function.