Marvell Asia Pte., Ltd.

Cavium, Inc.

Cavium International

Jason Daniel Zebchuk

Wilson Parkhurst Snyder, Ii

Tejas Maheshbhai Bhatt

Hong Jik Kim

Ahmed Shahid

Gregg Alan Bouchard

Richard Eugene Kessler

David K Carlson

Shubhendu Sekhar Mukherjee

David Asher

Isam Wadih Akkawi

Eric Marenger

Mark Jon Kwong

Kalyana Sundaram Venkataraman

Timothy Toshio Nakada

Steven W Aiken

Michael S Bertone

Shahe Hagop Krakirian

A system and corresponding method isolate work within a virtualized scheduler using tag-spaces. The system comprises a tag-space resource configured to store at least one respective assignment of at least one scheduling group to a given tag-space. The given tag-space defines a given ordering-atomicity domain that isolates, within the virtualized scheduler, (i) work belonging to the at least one scheduling group from (ii) work belonging to at least one other scheduling group, assigned, in the tag-space resource, to a respective tag-space different from the given tag-space. The system further comprises a work scheduler that schedules, for processing, work belonging to the at least one scheduling group and work belonging to the at least one other scheduling group. Such scheduling may have independent ordering and atomicity effectuated therebetween by the given ordering-atomicity domain. Such independency of ordering and atomicity improves quality-of-service of the virtualized scheduler.

System and method for isolating work within a virtualized scheduler using tag-spaces

A method and system for flexibly assigning hardware resources to physical and virtual functions in a processor system supporting hardware virtualization is disclosed. The processor system includes a resource virtualization unit which is used to flexibly assign hardware resources to physical functions and also flexibly assign local functions to virtual functions associated with one or more of the physical functions. Thereby, standard PCI software is compatible with the physical functions and any associated virtualized hardware resources that have been flexibly assigned to the virtual and local functions.

Flexible resource assignment to physical and virtual functions in a virtualized processing system

A memory request, including an address, is accessed. The memory request also specifies a type of an operation (e.g., a read or write) associated with an instance (e.g., a block) of data. A group of caches is selected using a bit or bits in the address. A first hash of the address is performed to select a cache in the group. A second hash of the address is performed to select a set of cache lines in the cache. Unless the operation results in a cache miss, the memory request is processed at the selected cache. When there is a cache miss, a third hash of the address is performed to select a memory controller, and a fourth hash of the address is performed to select a bank group and a bank in memory.

Methods and systems for distributing memory requests

A system and corresponding method queue work within a virtualized scheduler based on in-unit accounting (IUA) of in-unit entries (IUEs). The system comprises an IUA resource and arbiter. The IUA resource stores, in association with an IUA identifier, an IUA count and threshold. The IUA count represents a global count of work-queue entries (WQEs) that are associated with the IUA identifier and occupy respective IUEs of an IUE resource. The IUA threshold limits the global count. The arbiter retrieves the IUA count and threshold from the IUA resource based on the IUA identifier and controls, as a function of the IUA count and threshold, whether a given IUE from a given scheduling group, assigned to the IUA identifier, is moved into the IUE resource to be queued for scheduling. The IUA count and threshold prevent group(s) assigned to the IUA identifier from using more than an allocated amount of IUEs.

System and method for queuing work within a virtualized scheduler based on in-unit accounting of in-unit entries

An apparatus, and corresponding method, for input/output (I/O) value determination, generates an I/O instruction for an I/O device, the I/O device including a state machine with state transition logic. The apparatus comprises a controller that includes a simplified state machine with a reduced version of the state transition logic of the state machine of the I/O device. The controller is configured to improve instruction execution performance of a processor core by employing the simplified state machine to predict at least one state value of at least one I/O device true state value to be affected by the I/O instruction at the I/O device.

Complex I/O value prediction for multiple values with physical or virtual addresses

A system and corresponding method queue work within a virtualized scheduler based on in-unit accounting (IUA) of in-unit entries (IUEs). The system comprises an IUA resource and arbiter. The IUA resource stores, in association with an IUA identifier, an IUA count and threshold. The IUA count represents a global count of work-queue entries (WQEs) that are associated with the IUA identifier and occupy respective IUEs of an IUE resource. The IUA threshold limits the global count. The arbiter retrieves the IUA count and threshold from the IUA resource based on the IUA identifier and controls, as a function of the IUA count and threshold, whether a given WQE from a given scheduling group, assigned to the IUA identifier, is moved into the IUE resource to be queued for scheduling. The IUA count and threshold prevent group(s) assigned to the IUA identifier from using more than an allocated amount of IUEs.

Method and system embodying the method for scheduling arbitration among a plurality of service requestors encompassing: designating among the plurality of service requestors all the service requestors that have an active request; determining whether at least one of the designated service requestors has an un-served status indicator which is set; and when the determining is positive then: selecting one of the at least one designated service requestors in accordance with a pre-determined policy; and clearing the un-served status indicator for the selected service requestor, is disclosed.

Method and apparatus for scheduling arbitration among a plurality of service requestors

Method and system embodying the method for programmable scheduling encompassing: enqueueing at least one command into one of a plurality of queues having a plurality of entries; determining a category of the command at the head entry of each of the plurality of queues; processing each determined non-job category command by a non-job command arbitrator; and processing each determined job category command by a job arbitrator and assignor, is disclosed.

Programmable hardware scheduler for digital processing systems

Method and system embodying the method for a direct memory access between a data storage and a data processing device via one or more direct memory access units, comprising transferring data between the data storage and a first direct memory access engine of a respective one or more direct memory access units and providing the data for a second direct memory access engine of the respective one or more direct memory access units; and transferring the data provided by the first direct memory access engine by a second direct memory access engine to the data processing device via the second direct memory access engine is disclosed.

Method and apparatus for efficient and flexible direct memory access

A method utilizing a system encompassing a free pool buffer; a deadlock avoidance buffer; and a controller communicatively coupled to the free pool buffer and the deadlock avoidance buffer to reorder out-of-order responses to fetch requests into correct order by: receiving a fetch request on behalf of a consumer; allocating space first in the free pool buffer and when such space is not available then allocating space in a division associated with the consumer in the deadlock avoidance buffer. Issuing segment(s) of the fetch request including associated tag(s) to one of one or more memories; writing response data for each of the segment(s) to the allocated space in the free buffer or the deadlock avoidance buffer according to each of the associated tag(s); and transferring the response data to the consumer according to an entry in an ordering first-in, first-out buffer and an entry in a pending request array.

Re-ordering buffer for a digital multi-processor system with configurable, scalable, distributed job manager

Method and system embodying the method for load balancing of scheduled jobs among a plurality of engines encompassing determining a number of cluster credits for each of a plurality of clusters that comprise at least one engine capable of processing a scheduled job; determining a number of engine credits for each of the plurality of engines comprising each of the at least one engine in accordance with a number of jobs assigned to each of the plurality of engines; evaluating the determined number of cluster credits and the determined number of engine credits in accordance with a credit evaluation policy; and assigning the job to one of the plurality of engines in accordance with the evaluation, is disclosed.

Method and apparatus for load balancing of jobs scheduled for processing

A method and a system embodying the method for job pre-scheduling in a processing system comprising distributed job management, encompassing: determining a maximum amount of pre-schedulable jobs for each of a plurality of engines; setting for each of the plurality of engines a threshold less than or equal to the maximum amount; pre-scheduling by a scheduler an amount of jobs less than or equal to the threshold to at least one of a plurality of job managers; determining at the at least one of the plurality of job managers managing one of the plurality of engines one of a plurality of data processing devices in order for each pre-scheduled job; and assigning the job to the determined data processing device.

Growing community of inventors

Watertown, MA, United States of America

Jason Daniel Zebchuk