Patents by Inventor Tze-Chiang HUANG
Tze-Chiang HUANG has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 10970439Abstract: A System On Chip (SOC) current profile model for Integrated Voltage Regulator (IVR) co-design may be provided. A first current profile model may be extracted corresponding to an SOC at a first design stage of the SOC. Then it may be determined that an IVR and the SOC pass a first co-simulation based on the extracted first current profile model. Next, a second current profile model may be extracted corresponding to the SOC at a second design stage of the SOC. Then it may be determined that the IVR and the SOC pass a second co-simulation based on the extracted second current profile model. A third current profile model may be extracted corresponding to the SOC at a third design stage of the SOC. Then it may be determined that the IVR and the SOC pass a third co-simulation based on the extracted third current profile model.Type: GrantFiled: October 11, 2019Date of Patent: April 6, 2021Inventors: Haohua Zhou, Tze-Chiang Huang, Mei Hsu Wong
-
Publication number: 20210089696Abstract: A method includes receiving a source code for executing a plurality of operations associated with a machine learning algorithm, classifying each operation into a fast operation group and/or a slow operation group, defining a neuron network for executing operations of the slow operation group, and mapping the neuron network to an initial machine learning hardware configuration. The method also includes executing the slow operation group operation on the machine learning hardware configuration, finalizing the machine learning hardware configuration capable of successfully executing least one test data set.Type: ApplicationFiled: December 8, 2020Publication date: March 25, 2021Inventors: Kai-Yuan TING, Sandeep Kumar GOEL, Tze-Chiang HUANG, Yun-Han LEE
-
Patent number: 10867089Abstract: Electronic system level (ESL) design and verification of the present disclosure is utilized to provide an electronic simulation of various loads on one or more batteries of an electronic device resulting from the electronic device performing one or more functional behaviors. Before this electronic simulation occurs, the electronic device is modeled using the high-level software language or the high-level software format. For example, a battery discharge model, a regulator efficiency model, a power delivery network (PDN) model, or a component power model are used to model behaviors of the one or more batteries, regulator circuitry, power delivery network (PDN) circuitry, and other electronic circuits, respectively, of the electronic device.Type: GrantFiled: July 19, 2017Date of Patent: December 15, 2020Assignee: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Charlie Zhou, Kai-Yuan Ting, Sandeep Kumar Goel, Tze-Chiang Huang, Yun-Han Lee
-
Patent number: 10867098Abstract: A method includes receiving a source code for executing a plurality of operations associated with a machine learning algorithm, classifying each operation into a fast operation group or a slow operation group, defining a neuron network for executing operations of the slow operation group, and mapping the neuron network to an initial machine learning hardware configuration. The method also includes executing operations of the slow operation group on the initial machine learning hardware configuration, modifying the initial machine learning hardware configuration in response to a determination that the slow group operation fails to produce an expected result in response to at least one set of inputs; and executing a fast group operation using a machine learning software code.Type: GrantFiled: September 25, 2019Date of Patent: December 15, 2020Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY, LTD.Inventors: Kai-Yuan Ting, Sandeep Kumar Goel, Tze-Chiang Huang, Yun-Han Lee
-
Publication number: 20200372124Abstract: A method includes receiving a source code for executing a plurality of operations associated with a machine learning algorithm, classifying each operation into a fast operation group or a slow operation group, defining a neuron network for executing operations of the slow operation group, and mapping the neuron network to an initial machine learning hardware configuration. The method also includes executing operations of the slow operation group on the initial machine learning hardware configuration, modifying the initial machine learning hardware configuration in response to a determination that the slow group operation fails to produce an expected result in response to at least one set of inputs; and executing a fast group operation using a machine learning software code.Type: ApplicationFiled: September 25, 2019Publication date: November 26, 2020Inventors: Kai-Yuan TING, Sandeep Kumar GOEL, Tze-Chiang HUANG, Yun-Han LEE
-
Publication number: 20200274685Abstract: An integrated circuit includes a first through fourth devices positioned over a substrate, the first device including first through third transceivers, the second device including a fourth transceiver, the third device including a fifth transceiver, and the fourth device including a sixth transceiver. A first radio frequency interconnect (RFI) includes the first transceiver coupled to the fourth transceiver through a first guided transmission medium, a second RFI includes the second transceiver coupled to the fifth transceiver through a second guided transmission medium, and a third RFI includes the third transceiver coupled to the sixth transceiver by the second guided transmission medium.Type: ApplicationFiled: May 13, 2020Publication date: August 27, 2020Inventors: Huan-Neng CHEN, William Wu SHEN, Chewn-Pu JOU, Feng Wei KUO, Lan-Chou CHO, Tze-Chiang HUANG, Jack LIU, Yun-Han LEE
-
Patent number: 10719648Abstract: A method is disclosed that includes providing an IP bank, an application bank, and a technology bank; generating a hierarchical table based on the IP bank and the application bank; performing an estimation of at least one of a performance value, a power value, an area value and a cost value, which corresponds to the hierarchical table, by using the technology bank, to output an result data as a basis of fabrication of a system.Type: GrantFiled: September 8, 2016Date of Patent: July 21, 2020Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING CO., LTD.Inventors: Tze-Chiang Huang, Kai-Yuan Ting, Sandeep Kumar Goel, Yun-Han Lee, Shereef Shehata, Mei Wong
-
Publication number: 20200175129Abstract: A System On Chip (SOC) current profile model for Integrated Voltage Regulator (IVR) co-design may be provided. A first current profile model may be extracted corresponding to an SOC at a first design stage of the SOC. Then it may be determined that an IVR and the SOC pass a first co-simulation based on the extracted first current profile model. Next, a second current profile model may be extracted corresponding to the SOC at a second design stage of the SOC. Then it may be determined that the IVR and the SOC pass a second co-simulation based on the extracted second current profile model. A third current profile model may be extracted corresponding to the SOC at a third design stage of the SOC. Then it may be determined that the IVR and the SOC pass a third co-simulation based on the extracted third current profile model.Type: ApplicationFiled: October 11, 2019Publication date: June 4, 2020Inventors: Haohua Zhou, Tze-Chiang Huang, Mei Hsu Wong
-
Patent number: 10673603Abstract: An integrated circuit includes a first radio frequency interconnect (RFI) transceiver, a second RFI transceiver, a third RFI transceiver, a fourth RFI transceiver and a guided transmission medium. The first RFI transceiver is configured to transmit or receive a first data signal. The second RFI transceiver is configured to transmit or receive a second data signal. The third RFI transceiver is configured to transmit or receive the first data signal. The fourth RFI transceiver is configured to transmit or receive the second data signal. The guided transmission medium is configured to carry the first data signal and the second data signal. The first RFI transceiver and the second RFI transceiver are connected to the third RFI transceiver and the fourth RFI transceiver by the guided transmission medium.Type: GrantFiled: October 23, 2015Date of Patent: June 2, 2020Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY, LTD.Inventors: Huan-Neng Chen, William Wu Shen, Chewn-Pu Jou, Feng Wei Kuo, Lan-Chou Cho, Tze-Chiang Huang, Jack Liu, Yun-Han Lee
-
Publication number: 20200134249Abstract: A true random metastable flip-flop (TRMFF) compiler generates an electrical architecture for a TRMFF chain. The compiler selects components for the TRMFF chain from a library of standard cells and logically connects these components in accordance with a primitive polynomial to generate the electrical architecture. The TRMFF chain provides a sequence of random numbers from one or more physical processes in accordance with the primitive polynomial. During operation, one or more microscopic phenomena inside and/or outside of the TRMFF chain can cause one or more low-level, statistically random entropy noise signals to be present within the TRMFF chain. The TRMFF chain advantageously utilizes the one or more low-level, statistically random entropy noise signals to provide the sequence of random numbers.Type: ApplicationFiled: December 26, 2019Publication date: April 30, 2020Applicant: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Charlie ZHOU, Tze-Chiang HUANG, Jack LIU
-
Patent number: 10599796Abstract: A true random metastable flip-flop (TRMFF) compiler generates an electrical architecture for a TRMFF chain. The compiler selects components for the TRMFF chain from a library of standard cells and logically connects these components in accordance with a primitive polynomial to generate the electrical architecture. The TRMFF chain provides a sequence of random numbers from one or more physical processes in accordance with the primitive polynomial. During operation, one or more microscopic phenomena inside and/or outside of the TRMFF chain can cause one or more low-level, statistically random entropy noise signals to be present within the TRMFF chain. The TRMFF chain advantageously utilizes the one or more low-level, statistically random entropy noise signals to provide the sequence of random numbers.Type: GrantFiled: October 4, 2017Date of Patent: March 24, 2020Assignee: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Charlie Zhou, Tze-Chiang Huang, Jack Liu
-
Patent number: 10540462Abstract: A method includes providing a register transfer level (RTL) description of a circuit design, providing a plurality of RTL-to-gate-level mapping details by translating the RTL description into a gate-level netlist, providing one or more input/output (I/O) variables as stimulus to simulate the RTL description of the circuit design, capturing a plurality of internal operation values from the simulated RTL description at a beginning time of a specified period of time wherein the specified period of time is less than a time period required to compete a full-scale simulation, mapping the captured internal operation values to corresponding gate-level nodes of the gate-level netlist, capturing a plurality of I/O values from the I/O variables at the beginning time of the specified period of time, and simulating the circuit design in a gate-level for the specified period of time based on the mapped internal operation values and the captured I/O values.Type: GrantFiled: July 10, 2017Date of Patent: January 21, 2020Assignee: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Chih-Yuan Stephen Yu, Wenyuan Lee, Boh-Yi Huang, Brent Lui, Tze-Chiang Huang
-
Publication number: 20190332161Abstract: A device for power estimation is disclosed. The device includes a transformer circuit coupled with a processing circuit and a transaction interface. The transformer circuit is configured to count performance activities executed in the processing circuit and to compare count values of the performance activities with a pre-determined value to determine a power state of the processing circuit. The transaction interface is configured to receive a request from the processing circuit and record a first timestamp, and further configured to receive a response from a memory model and record a second timestamp, the transaction interface being further configured to record a time difference between the first timestamp and the second timestamp as a time difference. The transformer circuit is further configured to determine the power state of the processing circuit based on both of the count values and the time difference.Type: ApplicationFiled: July 8, 2019Publication date: October 31, 2019Applicant: TAIWAN SEMICONDUCTOR MANUFACTURING CO., LTD.Inventors: Kai-Yuan TING, Shereef SHEHATA, Tze-Chiang HUANG, Sandeep Kumar GOEL, Mei WONG, Yun-Han LEE
-
Patent number: 10345883Abstract: A power state transformer, a system and a method thereof are disclosed. The power state transformer is coupled with a processing unit model. The power state transformer is configured for counting performance activities executed in the processing unit model, and further for determining a power state of the processing unit model according to count values of the performance activities.Type: GrantFiled: May 31, 2016Date of Patent: July 9, 2019Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING CO., LTD.Inventors: Kai-Yuan Ting, Shereef Shehata, Tze-Chiang Huang, Sandeep Kumar Goel, Mei Wong, Yun-Han Lee
-
Publication number: 20180314781Abstract: A true random metastable flip-flop (TRMFF) complier generates an electrical architecture for a TRMFF chain. The complier selects components for the TRMFF chain from a library of standard cells and logically connects these components in accordance with a primitive polynomial to generate the electrical architecture. The TRMFF chain provides a sequence of random numbers from one or more physical processes in accordance with the primitive polynomial. During operation, one or more microscopic phenomena inside and/or outside of the TRMFF chain can cause one or more low-level, statistically random entropy noise signals to be present within the TRMFF chain. The TRMFF chain advantageously utilizes the one or more low-level, statistically random entropy noise signals to provide the sequence of random numbers.Type: ApplicationFiled: October 4, 2017Publication date: November 1, 2018Applicant: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Charlie ZHOU, Tze-Chiang Huang, Jack Liu
-
Publication number: 20180314772Abstract: Electronic system level (ESL) design and verification of the present disclosure is utilized to provide an electronic simulation of a various loads on one or more batteries of an electronic device resulting from the electronic device performing one or more functional behaviors. Before this electronic simulation occurs, the electronic device is modeled using the high-level software language or the high-level software format. For example, a battery discharge model, a regulator efficiency model, a power delivery network (PDN) model, or a component power model are used to model behaviors of the one or more batteries, regulator circuitry, power delivery network (PDN) circuitry, and other electronic circuits, respectively, of the electronic device.Type: ApplicationFiled: July 19, 2017Publication date: November 1, 2018Applicant: Taiwan Semiconductor Manufacturing Co., Ltd.Inventors: Charlie ZHOU, Kai-Yuan TING, Sandeep Kumar GOEL, Tze-Chiang HUANG, Yun-Han LEE
-
Patent number: 10108764Abstract: A method of estimating power consumption for a system on chip (SOC) includes simulating operation of a first sub-block to obtain power consumption information for the first sub-block including first activation information for a first IP block. The method further includes simulating operation of a second sub-block to obtain power consumption information for the second sub-block including second activation information for the first IP block and activation information for a plurality of second IP blocks. The method further includes determining a weighting factor for the first activation information for the first IP block, the second activation information for the first IP block and the activation information for each second IP block. The method further includes estimating power consumption for the SOC based on the first and second activation information for the first IP block, the activation information for at least one second IP block, and corresponding weighting factors.Type: GrantFiled: April 15, 2016Date of Patent: October 23, 2018Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY, LTD.Inventors: Shereef Shehata, Sandeep Kumar Goel, Tze-Chiang Huang, Yun-Han Lee, Mei Wong
-
Patent number: 10101796Abstract: A method of estimating power consumption of a processor includes accessing an electronic system level (ESL) model of the processor, the ESL model including a plurality of functional blocks, identifying a plurality of processor events by tracing activity of the plurality of functional blocks for a plurality of machine code instructions, and calculating a first power consumption value based on the plurality of processor events. The method also includes identifying a plurality of cycles by analyzing a plurality of micro-code operation codes corresponding to the plurality of machine code instructions, calculating a second power consumption value based on the plurality of cycles, and calculating a total power consumption value from the first power consumption value summed with the second power consumption value.Type: GrantFiled: May 27, 2016Date of Patent: October 16, 2018Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY, LTD.Inventors: Kai-Yuan Ting, Sandeep Kumar Goel, Tze-Chiang Huang, Yun-Han Lee
-
Publication number: 20180165394Abstract: A method includes providing a register transfer level (RTL) description of a circuit design, providing a plurality of RTL-to-gate-level mapping details by translating the RTL description into a gate-level netlist, providing one or more input/output (I/O) variables as stimulus to simulate the RTL description of the circuit design, capturing a plurality of internal operation values from the simulated RTL description at a beginning time of a specified period of time wherein the specified period of time is less than a time period required to compete a full-scale simulation, mapping the captured internal operation values to corresponding gate-level nodes of the gate-level netlist, capturing a plurality of I/O values from the I/O variables at the beginning time of the specified period of time, and simulating the circuit design in a gate-level for the specified period of time based on the mapped internal operation values and the captured I/O values.Type: ApplicationFiled: July 10, 2017Publication date: June 14, 2018Inventors: Chih-Yuan Stephen YU, Wenyuan LEE, Boh-Yi HUANG, Brent LUI, Tze-Chiang HUANG
-
Patent number: 9887863Abstract: A transceiver group includes a plurality of transceivers; wherein the transceiver group performs transmission and receiving through a wire, and each of the transceivers includes a transmitter and a receiver, and the transmitter includes: a carrier generator arranged to generate a plurality of carriers having different frequencies for a plurality of data streams to be transmitted; a modulator, coupled to the data streams to be transmitted and the carrier generator, to generate a plurality of modulated data streams carried on the plurality of carriers; and a summer arranged to merge the plurality of modulated data streams to an output signal to the wire; and the receiver includes: a carrier generator arranged to generate a plurality of carriers having different frequencies for an input signal received from the wire; and a demodulator, coupled to the input signal and the carrier generator, to generate a plurality of demodulated data streams.Type: GrantFiled: October 16, 2015Date of Patent: February 6, 2018Assignee: TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY LTD.Inventors: Huan-Neng Chen, William Wu Shen, Lan-Chou Cho, Feng Wei Kuo, Chewn-Pu Jou, Tze-Chiang Huang, Jack Liu, Yun-Han Lee