Memory management parameters derived from system modeling

• US 8,966,462 B2
• Filed: 08/10/2012
• Issued: 02/24/2015
• Est. Priority Date: 08/10/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Concurix Corporation
• Inventors: Gounares, Alexander G., Li, Ying, Garrett, Charles D.
• Primary Examiner(s): Wong, Don
• Assistant Examiner(s): Kabir, Mohammad
• Application Number:
  US13/571,599
• Publication Number:
  US 20130081005A1
• Time in Patent Office: 928 Days
• Field of Search: None
• US Class Current: 717/148
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3447 Performance evaluation by modeling
• CPC Class Codes: G06F 2201/865 Monitoring of software
• CPC Class Codes: G06F 2209/508 Monitor
• CPC Class Codes: G06F 9/4552 Involving translation to a different instruction set architecture, e.g. just-in-time translation in a JVM
• CPC Class Codes: G06F 9/5016 the resource being the memory
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

Optimized memory management settings may be derived from a mathematical model of an execution environment. The settings may be optimized for each application or workload, and the settings may be implemented per application, per process, or with other granularity. The settings may be determined after an initial run of a workload, which may observe and characterize the execution. The workload may be executed a second time using the optimized settings. The settings may be stored as tags for the executable code, which may be in the form of a metadata file or as tags embedded in the source code, intermediate code, or executable code. The settings may change the performance of memory management operations in both interpreted and compiled environments. The memory management operations may include memory allocation, garbage collection, and other related functions.

Claims

1. A method performed by a computer processor, said method comprising:

receiving a first executable code;

monitoring said first executable code in a first execution environment, said first executable code having a plurality of processes;

receiving a set of process classifications, each of said process classifications having a set of garbage collection settings;

classifying each of said plurality of processes and assigning a process classification, each of said plurality of processes having a process classification;

creating a tagged version of said first executable code comprising tags for each of said processes identifying said process classification; and

executing said tagged version of said first executable code such that each process is executed using a set of garbage collection settings associated with said process classification.

2. The method of claim 1, said tagged version being executed in a second execution environment.

3. The method of claim 2, said first execution environment being an instrumented execution environment.

4. The method of claim 3, said second execution environment having less instrumentation than said first execution environment.

5. The method of claim 1, said executable code being compiled prior to executing.

6. The method of claim 5 further comprising:

compiling said tagged version of said first executable code to produce tagged compiled code.

7. The method of claim 6, said tagged compiled code being executed in a second execution environment, said second execution environment:

recognizing a first tag associated with a first process;

determining that said first tag is associated with a first set of garbage collection settings; and

executing said first process using said first set of garbage collection settings.

8. The method of claim 1, said tagged version of said first executable code comprising said first executable code and a metadata file, said metadata file comprising tags for each of said plurality of processes.

9. The method of claim 1, said set of garbage collection settings comprising defining a garbage collection method.

10. The method of claim 9, a first process being executed with a first garbage collection method and a second process being executed with a second garbage collection method.

11. The method of claim 10, said first process and said second process having identical source code.

12. The method of claim 1, said set of garbage collection settings comprising a condition to launch garbage collection.

13. The method of claim 1, said first execution environment being an interpreted execution environment.

14. A system comprising:

a processor;

an optimization system executing on said processor, said optimization system that;

receives observations from executing a first executable code;

determines a set of optimized memory management settings;

generates metadata associating said optimized memory management settings for each of a plurality of processes executed with said first executable code;

tags each process within said first executable code with a separate set of said optimized memory management settings; and

causes said first executable code to be executed with said optimized memory management settings.

15. The system of claim 14, said separate set of said optimized memory management settings being stored in a metadata file.

16. The system of claim 15, said metadata file being consumed by a compiler to generate a second executable code.

17. The system of claim 15, said metadata file being consumed by an interpreter to execute said first executable code.

18. The system of claim 14, said optimized memory management settings comprising memory allocation settings.

19. The system of claim 14, said optimized memory management settings comprising garbage collection settings.