Intel Corporation

Muralidhar Rajappa

Devadatta V Bodas

Justin J Song

Andy Hoffman

James W Alexander

Ramkumar Nagappan

Manasi Deval

Cong Li

Anjaneya Reddy Chagam Reddy

Eric K Mann

Brian J Griffith

Gopal R Mundada

Jim Daubert

David Wescott

Raed Kanjo

Charles R Yount

James Steven Burns

Joseph A Schaefer

Mariusz Oriol

Ramakrishna Huggahalli

Michael K Patterson

Ilya Sharapov

Scott B Huck

Meenakshi Arunachalam

Sunil K Mahawar

Lingdan (Linda) Zeng

A system with improved power performance for tasks executed in parallel. A plurality of processing cores each to execute tasks. An inter-core messaging unit to conveys messages between the cores. A power management agent transitions a first core into a lower power state responsive to the first core waiting for a second core to complete a second task. In some embodiments long messages are subdivided to allow a receiving core to resume useful work sooner.

Method and apparatus to improve energy efficiency of parallel tasks

A zombie server can be detected. Detecting a zombie server can include labeling a plurality of processes as utility software, calculating a utilization of utility software on the plurality of processes executed in one or more processing resources during an interval of time, and calculating a server utilization of the one or more processing resources during the interval of time. Detecting the zombie server can also include determining whether a difference between the utilization of utility software and the server utilization is greater than a threshold, and identifying a server that hosts the processing resource as a zombie server based on a determination that the difference is smaller than the threshold.

Systems and methods for detecting a zombie server

A method and apparatus for node power regulation among nodes that share a power supply are described. In one embodiment, the apparatus comprises a power supply unit to provide input power and a plurality of nodes coupled to receive the input power, where each node of the plurality of nodes is operable to run power management logic, and wherein two or more nodes of the plurality of nodes alternate between performing power management and providing power regulation control information to other nodes of the plurality of nodes to regulate power consumption by the plurality of nodes, with, at any one time, only one node of plurality of nodes generating the power regulation control to regulate power for the plurality of nodes.

Method and apparatus to maintain node power budget for systems that share a power supply

A non-transitory computer readable storage medium having stored thereon instructions, the instructions being executable by one or more processors to perform operations including: receiving, by a calibration module executed by the one or more processors, a calibration request including (i) a workload type, (ii) a list of compute nodes belonging to a distributed computer system, and (iii) one or more frequencies; responsive to identifying the workload type as a clustered workload type, instructing a plurality of compute nodes on the list of compute nodes to begin processing a workload of the workload type; and responsive to identifying the workload type as a clustered workload type, instructing a compute node on the list of compute nodes to begin processing the workload of the workload type is shown.

Method and apparatus to generate and use power, thermal and performance characteristics of nodes to improve energy efficiency and reducing wait time for jobs in the queue

A zombie server can be detected. Detecting a zombie server can include receiving, at a server, network traffic and calculating a percentage of the network traffic as being productivity software layer 7 protocols every first time interval. Detecting a zombie server can also include marking the server as a zombie server based on the percentage every second time interval and processing the network traffic at the server to perform a number of actions by the productivity software.

A method and apparatus for coordinating and authenticating requests for data. In one embodiment, the apparatus comprises: a baseboard management controller (BMC); and a request coordinator coupled to the BMC to intercept BMC requests and to provide intercepted requests to the BMC, where the coordination interface comprises a request parser to parse parameters for each of the BMC requests, one or more queues to store the requests while the BMC is servicing another BMC request, and a command submitter to send individual BMC requests to the BMC, wherein the BMC is operable to generate the responses to the BMC requests received from the coordination interface and to send the responses to the coordination interface.

Method and apparatus to coordinate and authenticate requests for data

Examples are disclosed for determining or using server transaction latency information. In some examples, a network input/output device coupled to a server may be capable of time stamping information related to ingress request and egress response packets for a transaction. For these examples, elements of the server may be capable of determining transaction latency values based on the time stamped information. The determined transaction latency values may be used to monitor or manage operating characteristics of the server to include an amount of power provided to the server or an ability of the server to support one or more virtual servers. Other examples are described and claimed.

Techniques associated with server transaction latency information

A non-transitory computer readable storage medium storing instructions executable by one or more processors of a distributed computer system to perform operations including determining whether a power consumed by the distributed computer system is greater than a power allocated to the distributed computer system, responsive to determining the power consumed by the distributed computer system is greater than the power allocated to the distributed computer system, determining whether all jobs being processed by the distributed computer system are processing at a lowest power state for each job, wherein a job includes one or more calculations performed by the one or more processors of the distributed computer system and responsive to determining all jobs being processed by the distributed computer system are processing at a lowest power state for each job, suspending a job having a lowest priority among all jobs being processed by the distributed computer system is shown.

Methods and apparatus to manage jobs that can and cannot be suspended when there is a change in power allocation to a distributed computer system

A method of assessing energy efficiency of a High-performance computing (HPC) system, including: selecting a plurality of HPC workloads to run on a system under test (SUT) with one or more power constraints, wherein the SUT includes a plurality of HPC nodes in the HPC system, executing the plurality of HPC workloads on the SUT, and generating a benchmark metric for the SUT based on a baseline configuration for each selected HPC workload and a plurality of measured performance per power values for each executed workload at each selected power constraint is shown.

Method to assess energy efficiency of HPC system operated with and without power constraints

A system and method for computing at a facility having systems of multiple compute nodes to execute jobs of computing. Power consumption of the facility is managed to within a power band. The power consumption may be adjusted by implementing (e.g., by a power balloon) activities having little or no computational output.

Control of power consumption

A system and method for distributed computing, including executing a job of distributed computing on compute nodes. The speed of parallel tasks of the job executing on the compute nodes are adjusted to increase performance of the job or to lower power consumption of the job, or both, wherein the adjusting is based on imbalances of respective speeds of the parallel tasks.

Adjustment of execution of tasks

A non-transitory computer readable storage medium having stored thereon instructions executable by one or more processors to perform operations including: receiving a plurality of input parameters including (i) a workload type, (ii) a list of selected nodes belonging to a distributed computer system, and (iii) a list of frequencies; responsive to receiving the plurality of workload parameters, retrieving calibration data from a calibration database; generating a power estimate based on the plurality of workload parameters and the calibration data; and providing the power estimate to a resource manager is shown. Alternatively, the input parameters may include (i) a workload type, (ii) a list of selected nodes belonging to a distributed computer system, and (iii) an amount of available power, wherein the estimator may provide an estimation of the frequency at which the nodes should operate to utilize as much of the available power without exceeding the available power.

Methods and apparatus to estimate power performance of a job that runs on multiple nodes of a distributed computer system

Since the maximum power consumption is largely a concern at the power supply domain, a limited number of nodes may be allowed to consume their maximum power consumption by preventing other nodes from consuming their maximum power consumption. This approach may be used either instead of or in cooperation with existing maximum power consumption regulators.

Growing community of inventors

Chandler, AZ, United States of America

Muralidhar Rajappa