Integrated MPSoC Profiling Environment for Task Scheduling Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Multi-processor System on Chips (MPSoCs) require complex programming techniques to fully utilize their resources, and existing profiling and visualization tools fail to provide a high-level view of processes, leading to difficulties in task partitioning, application mapping, and task scheduling, which can result in applications failing to meet real-time processing requirements.

Innovation Solution

An integrated environment for execution monitoring and profiling of MPSoCs that includes an adaptive task scheduler, graphical user interface for task visualization, and optional debugging capabilities, allowing for real-time or after-execution profiling, and enabling manual optimization through stub tasks and simulated parameters for rapid prototyping and optimization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated task scheduling techniques are used, then productivity is improved, but the ability to provide high-level application code understanding deteriorates

Engineering Contradiction:

Improvetask scheduling efficiencyVSAvoidhigh-level application code understanding

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The patent introduces a profiling tool as an intermediary between the automated scheduling system and the application developer. This tool captures execution data from low-level hardware events and transforms it into high-level application code information, allowing developers to understand and optimize their code without manually dealing with complex hardware scheduling details.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces manual code analysis and optimization processes with an automated profiling system that uses hardware performance counters and event-based monitoring to automatically gather execution statistics, replacing the need for developers to manually trace and analyze low-level hardware behavior.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If low-level hardware element profiling is performed, then measurement precision is improved, but ease of operation deteriorates

Engineering Contradiction:

Improvehardware execution measurement accuracyVSAvoidprofiling tool usability

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The profiling tool acts as an intermediary that automatically collects precise low-level hardware execution data through hardware performance counters and event monitors, then transforms this data into high-level application code metrics, providing both measurement precision and ease of operation through automated data transformation.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system creates a virtual model or copy of the application execution at the hardware level using performance counters and event data, allowing developers to analyze execution behavior without directly interacting with complex hardware mechanisms, thus maintaining precision while improving usability.

Inventive Principle:

Principle #26Copying

3Manufacturing precision

If manual optimization through application profiler is performed, then manufacturing precision is improved, but productivity deteriorates

Engineering Contradiction:

Improveapplication optimization qualityVSAvoidoptimization development time

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The profiling tool performs preliminary automated analysis of application execution behavior before the developer begins manual optimization, providing pre-computed statistics and insights that guide the optimization process, thereby reducing the time and effort required to achieve high-quality optimization results.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms where execution data from hardware performance counters is continuously monitored and fed back to the developer through the profiling interface, enabling iterative optimization with immediate visibility into the impact of code changes on actual hardware execution behavior.

Inventive Principle:

Principle #23Feedback

Data Source

PatentEP2666083B1Integrated environment for execution monitoring and profiling of applications running on multi-processor system-on-chip

Publication Date: 2019.06.19 INTEL CORP

Figure 1
Figure 2
Figure 3

AI summary

There is provided a system and method for providing an integrated environment for execution monitoring and profiling of applications running on multi- processor system-on-chips. There is provided a method comprising obtaining task execution data of an application, the task execution data including a plurality of task executions assigned to a plurality of hardware resources, showing a scheduler view of the plurality of task executions on a display, receiving a modification request for a selected task execution from the plurality of task executions, reassigning the plurality of task executions to the plurality of hardware resources based on implementing the modification request, and updating the scheduler view on the display. As a result, the high level results of specific low level optimizations may be tested and retried to discover which optimization routes provide the greatest benefits.