| Patent application number | Description | Published |
|---|
| 20090007140 | REDUCING LAYERING OVERHEAD IN COLLECTIVE COMMUNICATION OPERATIONS - A communication method for collective operations is provided which eliminates the need for a three layered structure. Instead, the method of the present invention employs a control structure which is specifically designed to keep track of the progress of the collective communication operations and the facilities for the handling of asynchronous message passing. The method of the present invention thus eliminates the need for an intermediary, point-to-point communication layer that is required to follow message matching semantics for sending messages, receiving messages and for awaiting message arrivals. Application layer tasks and user defined programs running in the communication layer share access to the control structure as a mechanism for synchronizing and scheduling the progress of the collective communication operation. | 01-01-2009 |
| 20090100196 | GENERIC SHARED MEMORY BARRIER - A method, information processing node, and a computer program storage product are provided for performing synchronization operations between participants of a program. Each participant includes at least one of a set of processes and a set of threads. Each participant in a first subset of participants of a program updates a portion of a first local vector that is local to the respective participant. Each participant in a second subset of participants of the program updates a portion of a second local vector that is local to the respective participant. The participants in the second subset exit the synchronization barrier in response to determining that all of the participants in the first subset have reached the synchronization barrier. | 04-16-2009 |
| 20100122268 | COMMUNICATOR-BASED TOKEN/BUFFER MANAGEMENT FOR EAGER PROTOCOL SUPPORT IN COLLECTIVE COMMUNICATION OPERATIONS - A method, system, method and computer program product for facilitating collective communication in parallel computing. A system for parallel computing includes one or more communicators. Each of the one or more communicators comprises a plurality of processes. A memory pool including one or more early arrival buffers is provided. One or more tokens are assigned to a specified communicator included in the communicators. Each of the processes comprised by the specified communicator may consume any token assigned to the specified communicator. Requesting an early arrival buffer included in the memory pool requires consuming at least one token. A collective communication operation is performed using the specified communicator. The collective communication operation is performed eagerly using early arrival buffers obtained by consuming the tokens assigned to the communicator. | 05-13-2010 |
| 20110060891 | PARALLEL PIPELINED VECTOR REDUCTION IN A DATA PROCESSING SYSTEM - A parallel processing data processing system builds at least one data structure indicating a communication schedule for a plurality of processes each having a respective one of a plurality of equal length vectors formed of multiple equal size chunks. The data processing system, based upon the at least one data structure, communicates chunks of the plurality of vectors among the plurality of processes and performs partial reduction operations on chunks in accordance with the communication schedule. The data processing system then stores a result vector representing reduction of the plurality of vectors. | 03-10-2011 |
| Patent application number | Description | Published |
|---|
| 20080216089 | CHECKPOINT/RESUME/RESTART SAFE METHODS IN A DATA PROCESSING SYSTEM TO ESTABLISH, TO RESTORE AND TO RELEASE SHARED MEMORY REGIONS - A method is provided in which checkpointing operations are carried out in data processing systems running multiple processes which employ shared memory in a manner which preserves data coherence and integrity but which places no timing restrictions or constraints which require coordination of checkpointing operations. Data structures within local process memory and within shared memory provide the checkpoint operation with application level information concerning shared memory resources specific to at least two processes being checkpointed. Methods are provided for establishing, restoring and releasing shared memory regions that are accessed by multiple cooperating processes. | 09-04-2008 |
| 20090037450 | DATA GATHER SCATTER - REDISTRIBUTION MACHINE - A method, system, and computer program product to transfer data between two application data structures by copying a data gather scatter program (DGSP) from an exporting process address space where a first data structure is located, to a location in shared memory visible to an importing process address space; assembling a parameter set identifying the data structure; starting a data gather scatter-redistribution machine (DGS-RM) in an importing process space where a second application data structure is located; passing the first parameter set, the DGSP copy, and a second parameter set identifying a second application data structure and a second DGSP to the DGS-RM; and creating master and worker stack machines. The master stack machine identifies a contiguous chunk of the first data structure. The worker stack machine identifies contiguous chunks of the second data structure representing the same number of bytes as the contiguous chunk of the first data structure and transferring to (from) one or more identified chunks of the second data structure from (to) the single chunk of the first application data structure. | 02-05-2009 |
| 20090210635 | FACILITATING INTRA-NODE DATA TRANSFER IN COLLECTIVE COMMUNICATIONS, AND METHODS THEREFOR - Intra-node data transfer in collective communications is facilitated. A memory object of one task of a collective communication is concurrently attached to the address spaces of a plurality of other tasks of the communication. Those tasks that attach the memory object can access the memory object as if it was their own. Data can be directly written into or read from an application data structure of the memory object by the attaching tasks without copying the data to/from shared memory. | 08-20-2009 |
| 20110228789 | CONTENTION FREE PIPELINED BROADCASTING WITHIN A CONSTANT BISECTION BANDWIDTH NETWORK TOPOLOGY - In an interconnection network, multiple nodes are connected to one of a first layer of switches. The first layer of switches is connected to one another through a second layer of switches, wherein each of the nodes is connected through one of multiple shared links connecting the first layer switches and the second layer of switches. A pipelined broadcast manager schedules a hierarchical pipelined broadcast through at least one switch of the first layer switches comprising a multiple non-root nodes by selecting two nodes of multiple non-root nodes connected to the at least one switch and scheduling each of multiple broadcast steps for the pipelined broadcast with at least one of an inter-switch broadcast phase and an intra-switch broadcast phase. The inter-switch broadcast phase is scheduled with a first node of the two nodes receiving a first data packet from another switch from among the first layer of switches and a second node of the two nodes sending a second data packet previously received from another switch to one other switch from among the first layer switches. The intra-switch broadcast phase is scheduled with the first node acting as a source for sending a previously received data packet to at least one other non-root node and the second node acting as a sink for receiving the previously received data packet, wherein each of the non-root nodes sends and receives the previously received data packet once throughout the broadcast steps and the sink receives the previously received data packet last. In scheduling the hierarchical pipelined broadcast, the first node and the second node alternate roles each broadcast step. | 09-22-2011 |
