Qualcomm Incorporated

Ati Technologies Ulc

Qualcomm Innovation Center, Inc.

Wilson Kwan

James Lyall Esliger

Hee Jun Park

Subrato Kumar De

Mark Bapst

Aris Balatsos

Dineel Diwakar Sule

Kevin O'Neil

Stewart Chao

Neural network workload re-allocation in a system-on-chip having multiple heterogenous processors executing one or more neural network units may be based on measurements associated with the processors' conditions and on metadata associated with the neural network units. Metadata may be contained in an input file along with neural network information. Measurements characterizing operation of the processors may be obtained and compared with one or more thresholds. A neural network unit executing on a processor may be identified as a candidate for re-allocation based on metadata associated with the neural network unit and results of the comparisons. A target processor may be identified based on the metadata and results of the comparisons, and the candidate neural network neural network unit may be re-allocated to the target processor.

Run-time neural network re-allocation across heterogeneous processors

Various embodiments of methods and systems for proactive resource allocation and configuration are disclosed. An exemplary method first compiles and links a profile instrumented application with a compiler comprising a profile guided optimization feature that inserts calls to a profiler runtime. The profile instrumented application is executed on a target device using one or more workload datasets representative of probable workloads. During execution, based on recognition of the inserted calls, an instrumentation-based profile dataset is generated in association with each of the one or more workload datasets. Next, the profile instrumented application is recompiled and relinked based on the instrumentation-based profile datasets to create a set of profile guided optimizations to the source code, thereby resulting in an optimized application. The optimized application may be executed and monitored to generate a revised profile dataset useful for providing instructions to the target device for optimal workload allocation and resource configuration.

Application profile driven scheduling and configuration in a system on a chip

Systems and methods for profiling application code are disclosed. The method is hybrid in nature as it may include inserting instrumentation within application code and also periodic sample gathering, by employing a runtime app profile generator that provides the hybrid profiling infrastructure and is linked to the application code. An executable user application is then generated from the application code, and the executable user application is executed. The runtime app profile generator is then launched in response to the execution of the application code, and hybrid profiling results are generated by obtaining samples from the different threads of the executed application code and accumulating instrumented execution information. In some implementations, the hybrid profiling results capture even cold regions of the code and can also be used for a next round of profiling through automated targeted instrumentation.

Profiling system for computing devices

Systems, methods, and apparatus for improving a user experience during viewport scrolling is herein disclosed. Rendering and drawing of a viewport is performed for a destination viewport based on a viewport scroll vector calculated from a user scroll input at a display device. The destination viewport is rendered before viewports between a starting viewport and the destination viewport so that there is no lag in rendering when the scrolling action reaches the destination viewport. The compromise is that intermediate viewports may be rendered based on a thumbnail image in low resolution, but since these low-resolution portions of the webpage are viewed briefly as the webpage scrolls to the destination viewport, a user is unlikely to notice the reduced resolution.

Predictive web page rendering using a scroll vector

A method and apparatus for reducing net power consumption in a computer system includes identifying a plurality of processing states operable to execute a task. A processing state and current drain pattern is selected that is most power efficient. A selected processing state may include one or more processing elements of the computer system such as one or more processors or accelerators and indicates the manner in which one or more portions of the received task may be distributed among the processing elements of the computer system. The current drain pattern selected may be a constant current drain pattern or a pulsed current drain pattern and may be selected to optimize power consumption when executing the task among the one or more processing elements.

Method and apparatus for optimizing power consumption in a multiprocessor environment

A technique for adjusting or modifying content delivery to devices is provided, thereby offering a degree of power consumption control in such devices. The device can determine a need to modify power consumption, either in response to an input or via detection of an unfavorable power supply condition. In response, the device sends a request to an infrastructure for at least one reduced rate data stream. Because the at least one reduced rate data stream requires less processing power when undergoing decoding operations, the device subsequently consumes less power. In one embodiment of the present invention, the determination of the unfavorable power supply condition may be based on a configurable power condition profile. Furthermore, subsequent to receiving the at least one reduced rate data stream, the device may provide a request to the infrastructure for at least one increased rate data stream.

Growing community of inventors

Toronto, Canada

Wilson Kwan