View source for 17948159. OPTIMIZING OPERATION OF HIGH-PERFORMANCE COMPUTING SYSTEMS simplified abstract (HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP)

=OPTIMIZING OPERATION OF HIGH-PERFORMANCE COMPUTING SYSTEMS=


==Organization Name==

[[:Category:HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP|HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP]]


[[Category:HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP]]

==Inventor(s)==

[[:Category:Jan Maximilian M�der of Fort Collins CO (US)|Jan Maximilian M�der of Fort Collins CO (US)]][[Category:Jan Maximilian M�der of Fort Collins CO (US)]]

[[:Category:Christian Simmendinger of Baden-Württemberg (DE)|Christian Simmendinger of Baden-Württemberg (DE)]][[Category:Christian Simmendinger of Baden-Württemberg (DE)]]

[[:Category:Tobias Walter Wolfgang Schiffmann of Baden-Württemberg (DE)|Tobias Walter Wolfgang Schiffmann of Baden-Württemberg (DE)]][[Category:Tobias Walter Wolfgang Schiffmann of Baden-Württemberg (DE)]]

[[:Category:Torsten Wilde of Berlin (DE)|Torsten Wilde of Berlin (DE)]][[Category:Torsten Wilde of Berlin (DE)]]

==OPTIMIZING OPERATION OF HIGH-PERFORMANCE COMPUTING SYSTEMS - A simplified explanation of the abstract==


This abstract first appeared for US patent application 17948159 titled 'OPTIMIZING OPERATION OF HIGH-PERFORMANCE COMPUTING SYSTEMS

==Simplified Explanation==

The abstract describes a method for optimizing operations of high-performance computing (HPC) systems using machine learning techniques to classify workloads and determine optimal settings for hardware execution parameters.

* The method involves collecting data on workload performance profiling counters during runtime.
* A machine-learning technique is used to classify workloads and determine workload-specific fingerprints.
* An optimization metric is identified to optimize during workload execution.
* Optimal settings for tunable hardware execution parameters are determined by varying parameters and measuring against the optimization metric.
* Workload-specific fingerprints, optimization metrics, and optimal settings are stored in an architecture-specific knowledge database.

== Potential Applications ==
This technology can be applied in various fields such as scientific research, data analysis, financial modeling, and artificial intelligence where high-performance computing systems are used.

== Problems Solved ==
This technology helps in optimizing the performance of HPC systems by identifying workload-specific characteristics and determining optimal hardware execution parameters, leading to improved efficiency and productivity.

== Benefits ==
The benefits of this technology include enhanced performance, increased efficiency, reduced energy consumption, and improved overall productivity of high-performance computing systems.

== Potential Commercial Applications ==
Potential commercial applications of this technology include cloud computing services, data centers, research institutions, financial organizations, and technology companies looking to optimize their HPC systems for better performance.

== Possible Prior Art ==
One possible prior art in this field is the use of performance profiling tools and techniques to analyze and optimize the performance of HPC systems. Researchers have been exploring various methods to improve the efficiency and effectiveness of high-performance computing operations.

=== What are the specific machine-learning techniques used in classifying workloads and determining optimal settings for hardware execution parameters? ===

The specific machine-learning techniques used in this method are not explicitly mentioned in the abstract. Further details on the algorithms and models employed for workload classification and parameter optimization would provide a clearer understanding of the technology's implementation.

=== How does the method ensure the security and privacy of the collected data and stored information in the architecture-specific knowledge database? ===

The abstract does not address the security and privacy measures implemented to protect the collected data and stored information in the knowledge database. An explanation of encryption methods, access controls, and data protection mechanisms would be essential to address concerns regarding data security and privacy in the context of this technology.


==Original Abstract Submitted==

A method for optimizing operations of high-performance computing (HPC) systems includes collecting data associated with a plurality of workload performance profiling counters associated with a workload during runtime of the workload in an HPC system. Based on the collected data, the method includes using a machine-learning technique to classify the workload by determining a workload-specific fingerprint for the workload. The method includes identifying an optimization metric to optimize during running of the workload in the HPC system. The method includes determining an optimal setting for a plurality of tunable hardware execution parameters as measured against the optimization metric by varying at least a portion of the plurality of tunable hardware execution parameters. The method includes storing the workload-specific fingerprint, the optimization metric, and the optimal setting for the plurality of tunable hardware execution parameters as measured against the optimization metric in an architecture-specific knowledge database.



[[Category:G06F9/50]]
[[Category:G06F11/34]]