Electrical computers and digital processing systems: multicomput – Computer-to-computer data routing – Least weight routing
Reexamination Certificate
1997-09-29
2002-11-26
Maung, Zarni (Department: 2151)
Electrical computers and digital processing systems: multicomput
Computer-to-computer data routing
Least weight routing
Reexamination Certificate
active
06487578
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates generally to the field of microprocessors. More specifically, the invention relates to operating system policy regarding the utilization of microprocessors.
2. Description of the Related Art
PC (Personal Computer) based applications, such as videoconferencing that involve heavily multimedia (audio, video and data) streams which must be simultaneously processed, place heavy demands upon the processor and resources of the platforms where these applications are executed. In addition to their appetite for processing power, these applications must execute in real-time, i.e., the multimedia processing must occur in tightly constrained time intervals. More recently, desktop personal computers have used true demand-paged multi-tasking operating systems (such as Windows NT™) wherein both multimedia applications and third party applications (other than multimedia, such as a spreadsheet) vie and contend for key system resources such as processor “bandwidth” (referring to the amount of computational resource utilized per unit time) and memory. Where a third party application (an application other than multimedia) and the multimedia application so contend, two situations arise: (1) the third party application will take resources away from the multimedia application causing the multimedia application to be unable to meet its real-time deadlines for a desired quality of service (QOS) and/or (2) the third party application will execute sluggishly and nonresponsively and may be “starved” by the multimedia application for the computational resource.
Operating systems are ordinarily designed to distribute system resources evenly by preventing the monopolization of resources by one application to the exclusion of others. Most commercial operating systems for desktop PCs are not designed to accommodate real-time applications specifically, but rather, are designed to allocate resources without regard to any real-time constraints by a particular application. Real-time multimedia applications that are sensitive to latency are becoming more and more prevalent and will only increase with the advent of innovations such as the Intel Pentium Processor with MMX Technology™ (having special instructions geared toward the processing of multimedia). The emergence of videoconferencing transmission standards, broadband data delivery and high-bandwidth information pathways will additionally serve to increase the use of multimedia applications. The increased load on microprocessors and system resources due to the widening use of applications will not be appropriately accounted for by operating system policy. Commercial operating systems are by necessity targeted to a broad base of applications of different types, and favors fairness and resource use equalization between applications.
Because an operating system is designed to execute a broad range of application with widely differing usage characteristics, it is very difficult to tailor a commercial shrink-wrapped desktop operating system to the needs of any particular application. Rather, it is more feasible for individual applications to be able to control the individual resources being consumed in response to contention for system resources by other applications. This will allow such an application to consume heavily when there is no contention and voluntarily consume less when there is contention.
Currently, there is no practice for allowing applications to modify their own operation depending upon their utilization of system resources. Further, though an operating system may compute resource utilization by application, it does not compute, load or utilization for the individual and subsystems of running applications. Operating systems do track utilization of threads. However, they have no knowledge of the internal structure of applications that are multi-threaded. If an application is partitioned into a number of subsystems with each subsystem consisting of several distinct threads, current operating systems provide no mechanism to back the processor utilization of individual subsystems. Further, there is no mechanism for meaningfully presenting and packaging statistics related to load to an application such that the application can intelligently respond by modifying its operating parameters. Traditional operating system costing does not allow for distinguishing between short-term transient aberrations and long-term utilization trends which meaningful statistics would provide. The most effective and efficient way of utilizing available resources is to allow application to control the operating parameters of their own threads and subsystems based on dynamic-statistical feedback of and estimates (costing metrics) of resource utilization.
There is a need for providing costing metrics to applications to allow the application to modify its operation to more effectively respond to contention for system resources by other applications running concurrently.
SUMMARY OF THE INVENTION
What is disclosed is an adaptive resource utilization apparatus for an application. The apparatus includes a costing subsystem engine configured to measure resource utilization of application-defined subsystems within that application as well as a feedback mechanism configured to interface the application to the costing engine. The application is enabled to modify its operating parameters based on resource utilization of its individual subsystems as presented to the application by the feedback mechanism.
REFERENCES:
patent: 5682530 (1997-10-01), Shimamura
patent: 5761091 (1998-06-01), Agrawal et al.
patent: 5826083 (1998-10-01), Prasad
patent: 5838968 (1998-11-01), Culbert
patent: 5878258 (1999-03-01), Pizi et al.
patent: 6003061 (1999-12-01), Jones et al.
patent: 6026362 (2000-02-01), Kim et al.
Blakely , Sokoloff, Taylor & Zafman LLP
Intel Corporation
Maung Zarni
Opie George L.
LandOfFree
Dynamic feedback costing to enable adaptive control of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Dynamic feedback costing to enable adaptive control of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Dynamic feedback costing to enable adaptive control of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2930614