Honeywell International Inc.

Honeywell G.m.b.h.

Aaron Raymond Larson

Larry James Miller

Ryan Lawrence Roffelsen

Arthur Kirk McCready

Srivatsan Varadarajan

Pamela A Binns

Richard L Frost

Murali Rangarajan

Brendan Hall

Nicholas J Wilt

Byron F Birkedahl

Stephen C Vestal

Joseph A Smith

Art McCready

Gary Corbett Kindorf

Systems and methods for scheduling tasks using sliding time windows are provided. In certain embodiments, a system for scheduling the execution of tasks includes at least one processing unit configured to execute multiple tasks, wherein each task in the multiple tasks is scheduled to execute within a scheduler instance in multiple scheduler instances, each scheduler instance in the multiple scheduler instances being associated with a set of time windows in multiple time windows and with a set of processing units in the at least one processing unit in each time window, time windows in the plurality of time windows having a start time and an allotted duration and the scheduler instance associated with the time windows begins executing associated tasks no earlier than the start time and executes for no longer than the allotted duration, and wherein the start time is slidable to earlier moments in time.

Systems and methods for scheduling tasks using sliding time windows

Embodiments of the present invention provide improved systems and methods for processing multiple tasks. In one embodiment a method comprises: selecting a processing unit as a master processing unit from a processing cluster comprising multiple processing units, the master processing unit selected to execute master instruction entities; reading a master instruction entity from memory; scheduling the master instruction entity to execute on the master processing unit; identifying an execution group containing the master instruction entity, the execution group defining a set of related entities; when the execution group contains at least one slave instruction entity, scheduling the at least one slave instruction entity to execute on a processing unit other than the master processing unit during the execution of the master instruction entity; and terminating execution of instruction entities related by the execution group when a master instruction entity is executed that is not a member of the execution group.

Systems and methods for bounding processing times on multiple processing units

A method for scheduling schedulable entities onto an execution timeline for a processing entity in a constrained environment includes determining available capacity on the execution timeline for the processing entity based on constraints on the execution timeline over a plurality of time periods, wherein schedulable entities can only be scheduled onto the execution timeline during schedulable windows of time that are not precluded by constraints. The method further includes determining whether enough available capacity exists to schedule a schedulable entity with a budget at a rate. The method further includes when enough available capacity exists to schedule the schedulable entity with the budget at the rate, scheduling the schedulable entity onto the execution timeline for the processing entity during a schedulable window of time. The method further includes when the schedulable entity is scheduled onto the execution timeline, updating available capacity to reflect the capacity utilized by the schedulable entity.

Constrained rate monotonic analysis and scheduling

One example is directed to a method of generating a set of schedules for use by a partitioning kernel to execute a plurality of partitions on a plurality of processor cores included in a multi-core processor unit. The method includes determining a duration to execute each of the plurality of partitions without interference and generating a candidate set of schedules using the respective duration for each of the plurality of partitions. The method further includes estimating how much interference occurs for each partition when the partitions are executed on the multi-core processor unit using the candidate set of schedules and generating a final set of schedules by, for at least one of the partitions, scaling the respective duration in order to account for the interference for that partition. The method further includes configuring the multi-core processor unit to use the final set of schedules to control the execution of the partitions using at least two of the cores.

Safe partition scheduling on multi-core processors

In a computing system a method and apparatus for cache pooling is introduced. Threads are assigned priorities based on the criticality of their tasks. The most critical threads are assigned to main memory locations such that they are subject to limited or no cache contention. Less critical threads are assigned to main memory locations such that their cache contention with critical threads is minimized or eliminated. Thus, overall system performance is improved, as critical threads execute in a substantially predictable manner.

Cache pooling for computing systems

Methods and systems for a scalable self-checking processing platform are described herein. According to one embodiment, during an execution frame, a first processing element executes both a high-criticality application and a first low-criticality application. During that same execution frame, a second processing element executes both the high-criticality application and a second low-criticality application. The high-criticality application output from the first processing element is compared with that from the second processing element before the next execution frame, and a fault occurs when the output does not match. The low-criticality application is not duplicated or compared. This and other embodiments allow high-criticality applications to be appropriated checked while avoiding the over-dedication of resources to low-criticality applications that do not warrant self-checking.

Scalable self-checking processing platform including processors executing both coupled and uncoupled applications within a frame

In a multitasking system executing real-time harmonic and dynamic tasks having various priority levels, slack is stolen from timeline and reclaimed slack to enable execution of high priority non-essential tasks on a best efforts basis. Counts of the amount of slack consumed, slack reclaimed, and periodic compute time consumed are maintained by individual priority level and dynamically updated at certain times. Idle time is calculated by priority level. Available slack is calculated, allocated and consumed by rate, with the highest rate first and the lowest last. Slack is made available to tasks in more than one time partition. All slack belongs to a common system-wide pool of slack obtained from any one or more time partitions. Common slack can also be time-shared by static, non-harmonic tasks residing in different time partitions. Also described are a computer system and methods that perform slack scheduling in a time-partitioned system.

Methods and apparatus for sharing slack in a time-partitioned system

A time-partitioned system for accounting for processor time consumed by operating system services provided on behalf of an application runs in a real-time environment. The time utilized by the operating system is treated as application processing time, rather than viewing the resultant processor time consumed as an operating system overhead. Each application consists of one or more threads. A time budget is assigned to each thread and to each interrupt. The processor time consumed by the operating system when executing on behalf of a thread or interrupt is charged back to the application or interrupt as part of the application's time budget. Operating system overheads and processor interrupts are thus accounted for accurately enough to determine a schedule for the applications that is enforceable by the operating system at run time.

System and method for robust time partitioning of tasks in a real-time computing environment

A method is provided for optimizing CPU budgets associated with at least first and second threads executable in a time-partitioned real-time operating system. The first thread having a first CPU budget is executed, and at least a portion of the first thread's unused CPU budget is transferred to the second thread.

Budget transfer mechanism for time-partitioned real-time operating systems

A time-partitioned system for accounting for processor time consumed by operating system services provided on behalf of an application running in a real-time environment. The time utilized by the operating system is treated as being application processing time, rather than viewing the resultant processor time consumed as an operating system overhead. Each application consists of one or more threads. A time budget is assigned to each thread and to each interrupt. The processor time consumed by the operating system when executing on behalf of a thread or interrupt is charged back to the application or interrupt as part of the application's time budget. Operating system overheads and processor interrupts are thus accounted for accurately enough to determine a schedule for the applications which is enforceable by the operating system.

Method for time partitioned application scheduling in a computer operating system

A time-efficient inter-process communication method ( ) and protocol allowing producers and consumers to execute at any combination of periods, where producers ( ) and consumers ( ) are isolated from execution ordering variations. The method and protocol is for periodic broadcast-type inter-process communication in a harmonic rate system. The system has a plurality of threads sharing a memory space ( ) and including consuming threads and producing threads having potentially different rates. The producing threads have a period and are adapted to produce a message consumable by said consuming threads. The memory space ( ) includes a plurality of buffers ( ) for messages produced. The method includes the step of writing ( ) a copy of said produced message to a buffer in said memory space during said producing thread's period, wherein only one copy of said produced message is written, regardless of the difference in rates between producing and consuming threads.

Growing community of inventors

Shoreview, MN, United States of America

Aaron Raymond Larson