International Business Machines Corporation

Globalfoundries Inc.

Intel Corporation

Lenovo (singapore) Pte. Ltd.

Charles J Archer

Brian E Smith

Michael A Blocksome

Joseph D Ratterman

James E Carey

Philip James Sanders

Amanda Peters

Kurt Walter Pinnow

Matthew W Markland

Todd Alan Inglett

Gary Ross Ricard

Albert Sidelnik

Jeffrey John Parker

Sameh Sherif Sharkawi

Brent Allen Swartz

Brian Paul Wallenfelt

Nysal Jan K A

Douglas R Miller

Mark Gregory Megerian

Bob Richard Cernohous

Roy Glenn Musselman

Ahmad A Faraj

Philip Heidelberger

Michael E Aho

Tsai-Yang Jea

Gheorghe Almasi

Nysal Jan Ka

Timothy J Schimke

Thomas M Gooding

Sameer Kumar

Daniel A Faraj

Patrick Joseph McCarthy

Amanda Randles

Gabor J Dozsa

Matthew Ryan Weaver

Jeffrey M Ceason

Jeremy E Berg

Emily J Howe

Benjamin E Lynam

Bin Cao

Phillip V Mann

David Lee Darrington

Mark Edwin Giampapa

Hanhong Xue

Chulho Kim

David Pardo Keppel

Dean J Burdick

Gregory Howard Bellows

Timothy John Mullins

Gordon G Stewart

Ruth Janine Poole

Camesha R Hardwick

Hanghong Xue

This disclosure describes, in one embodiment an apparatus. The apparatus includes a processor; a memory, an application, collector circuitry and aggregator circuitry. The memory is to store one or more tasks. The application is associated with the one or more tasks. The collector circuitry is to identify a local free address range in at least one address space. The aggregator circuitry is to provide address range data to a subgroup aggregator. The provided address range data includes at least one local free address range.

Symmetric addressing

Data communications may be carried out in a distributed computing environment that includes a plurality of computers coupled for data communications through communications adapters and an active messaging interface ('AMI'). In distributed computing environment, data communications may include: receiving in the AMI from an application an eager SEND instruction that describes the location and size of send data in an application SEND buffer; copying by the AMI the send data from the application SEND buffer to a temporary AMI buffer; advising the application of completion of the SEND instruction before sending the SEND data to the receiver; and after advising the application of completion of the SEND instruction, sending the SEND data by the sender to the receiver.

Data communications in a distributed computing environment

In a distributed computing environment that includes hosts that execute a VMM, where each VMM supports execution of one or more VMs, administering VMs may include: assigning, by a VMM manager, the VMMs of the distributed computing environment to a logical tree topology, including assigning one of the VMMs as a root VMM of the tree topology; and executing, amongst the VMMs of the tree topology, a broadcast operation, including: pausing, by the root VMM, execution of one or more VMs supported by the root VMM; sending, by the root VMM, to other VMMs in the tree topology, a message indicating a pending transfer of the paused VMs; and transferring the paused VMs from the root VMM to the other VMMs.

Administering virtual machines in a distributed computing environment

Methods, apparatus, and products are disclosed for providing full point-to-point communications among compute nodes of an operational group in a global combining network of a parallel computer, each compute node connected to each adjacent compute node in the global combining network through a link, that include: receiving a network packet in a compute node, the network packet specifying a destination compute node; selecting, in dependence upon the destination compute node, at least one of the links for the compute node along which to forward the network packet toward the destination compute node; and forwarding the network packet along the selected link to the adjacent compute node connected to the compute node through the selected link.

Providing full point-to-point communications among compute nodes of an operational group in a global combining network of a parallel computer

Executing an all-to-allv operation on a parallel computer that includes a plurality of compute nodes, including: packing, by each task in an operational group of tasks, vectored contribution data from vectored storage in an all-to-allv contribution data buffer into an all-to-all contribution data buffer, wherein two or more entries in the all-to-allv contribution data buffer are different in size and each entry in the all-to-all contribution data buffer is identical in size; executing with the contribution data as stored in the all-to-all contribution data buffer an all-to-all collective operation by the operational group of tasks; and unpacking, by each task in the operational group of tasks, received contribution data from the all-to-all contribution data buffer into the vectored storage in an all-to-allv contribution data buffer.

Executing an all-to-allv operation on a parallel computer that includes a plurality of compute nodes

The present invention provides a system and method for extracting elements from distributed arrays on a parallel processing system. The system includes a module that populates a result array with globally largest elements from input arrays, a module that generates a partition element, a module that counts the number of local elements greater than the partition element, and a module that determines the globally largest elements. The method for extracting elements from distributed arrays on a parallel processing system includes populating a result array with globally largest elements from input arrays, generating a partition element, counting the number of local elements greater than the partition and determining the globally largest elements.

Parallel, low-latency method for high-performance speculative globally-large element extraction from distributed, sorted arrays

Methods, apparatuses, and computer program products for collective operation management in a parallel computer are provided. Embodiments include a parallel computer having a plurality of compute nodes coupled for data communications over a data communications network. Embodiments include a first compute node entering a collective operation. Each compute node of the plurality of compute nodes is associated with the collective operation. In response to entering the collective operation, the first compute node decreases power consumption of the first compute node.

Collective operation management in a parallel computer

Methods, apparatuses, and computer program products for implementing updates to source code executing on a plurality of compute nodes are provided. Embodiments include receiving, by a compute node, a broadcast update-notification message indicating there is an update to the source code executing on the plurality of compute nodes; in response to receiving the update-notification message, implementing a distributed barrier; based on the distributed barrier, halting execution of the source code at a particular location within the source code; based on the distributed barrier, updating in-place the source code including retaining workpiece data in memory of the compute node, the workpiece data corresponding to the execution of the source code; and based on completion of the updating of the source code, resuming with the retained workpiece data execution of the source code at the particular location within the source code where execution was halted.

Implementing updates to source code executing on a plurality of compute nodes

In a distributed computing environment that includes hosts that execute a VMM, with each VMM supporting execution of one or more VMs, administering VMs may include: assigning, by a VMM manager, the VMMs of the distributed computing environment to a logical tree topology, including assigning one of the VMMs as a root VMM of the tree topology; and executing, amongst the VMMs of the tree topology, a scatter operation, including: pausing, by the root VMM one or more executing VMs; storing, by the root VMM in a buffer, a plurality of VMs to scatter amongst the other VMMs of the tree topology; and sending, by the root VMM, to each of the other VMMs of the tree topology a different one of the VMs stored in the buffer.

Growing community of inventors

Rochester, MN, United States of America

Charles J Archer