Back to results
US 11,231,767 B2Granted
Dynamic frequency scaling
AssigneeTaiwan Semiconductor Manufacturing Company, Ltd.
InventorsKai-Yuan Ting, Ashok Mehta, Stanley John, Sandeep Kumar Goel
Life of this patent
Estimating the term…
Abstract
A method for dynamic frequency scaling (DFS) on the electronic systems level (ESL). The method can run in a virtual environment and dynamically scale the frequency of a virtual component based on a first transaction time and a second transaction time.
Claims20
A method, comprising:
- determining a total runtime for a plurality of application tasks running on a computer simulator;
- determining a first transaction time based on the time the application tasks spend utilizing at least one hardware component simulator, wherein the first transaction time represents time spent performing input/output operations;
- calculating a second transaction time from the total runtime and the first transaction time, wherein the second transaction time represents time spent utilizing at least one component having a scalable frequency;
- determining from the second transaction time and the total runtime whether a frequency of the processing unit simulator is to be adjusted;
- and
- adjusting the frequency to the processing unit simulator, wherein the frequency is adjusted based on a ratio between the second transaction time and the total transaction time.
3 more elements
A non-transient machine readable storage medium encoded with program code, wherein when the program code is executed by a processor, the processor performs a method comprising the steps of:
- establishing a computer simulator configured to simulate a hardware environment, wherein the computer simulator is comprised of at least one first frequency hardware component simulator and at least one second frequency hardware component simulator, wherein the at least one first frequency hardware component comprises a scalable frequency and the at least one second frequency hardware component comprises an input/output component;
- running at least one computer task on the computer simulator;
- and
- adjusting a frequency of the computer simulator based on a first time spent by the at least one first frequency hardware component to perform the at least one computer task and a second time spent by the at least one second frequency hardware component to perform the at least one computer task, wherein the frequency is increased when the first time is greater than the second time and the frequency is decreased with the first time is less than the second time.
1 more element
A method, comprising:
- establishing a computer simulator configured to simulate a hardware environment, wherein the computer simulator is comprised of at least one processing unit simulator and at least one hardware component simulator, wherein the at least one hardware component simulator comprises an input/output component;
- running a first plurality of tasks on the computer simulator, wherein the first plurality of tasks include at least one low frequency task and at least one high frequency task, wherein each low frequency task corresponds to time spent performing input/output operations and each high frequency task corresponds to time spent utilizing a component having a scalable frequency;
- and
- adjusting a frequency of the processing unit simulator based on a first time spent performing the at least one low frequency task and a second time spent performing the at least one high frequency task, wherein the frequency is increased when the first time is less than the second time and the frequency is decreased with the first time is greater than the second time,
- wherein the simulated hardware environment comprises a bus monitoring unit (BMU) configured to monitor traffic between the at least one processing unit simulator and the at least one hardware component simulator, wherein the BMU is further configured to record bus usage data in a non-transient computer readable storage medium, wherein the bus usage data includes the total transaction time.
2 more elements
Dependencies are inferred from the claim text.