International Business Machines Corporation

James C Sexton

John A Gunnels

Fred Gehrung Gustavson

James Allan Kahle

Charles Ray Johns

Amanda Peters

Shakti Kapoor

Ravi Kothari

Patrick Joseph McCarthy

Yogish Sabharwal

David L Satterfield

Alan Jonathan King

Renato John Recio

Alan G Gara

Philip Heidelberger

Ravi Nair

Martin Ohmacht

Bernard Metzler

Jose R Brunheroto

Michael B Mundy

Abdullah Kayi

Alessandro Morari

Eran Gampel

Changhoan Kim

Gordon G Stewart

Robert E Walkup

Richard Ormond Simpson

Constantinos Evangelinos

Ravinder Reddy Amanaganti

Gal Sagi

Gyan Vatsaraja Bhanot

Thomas A Budnik

Patrick D Siegl

Kirk E Jordan

Matthew R Drahzal

Eoin M Lawless

Todd A Inglet

Embodiments for providing enhanced endpoint multicast emulation in a computing environment. One or more multicast operations may be executed on an overlay network using endpoint multicast emulation by using an overlay layer or a virtual extensible LAN ('VXLAN') layer to maintain control over one or more multicast groups.

Enhanced endpoint multicast emulation

An approach is disclosed that maintains a consistent view of a virtual address by a local node which writes a first value to the virtual address and, after writing the first value, establishes a snapshot consistency state of the virtual address. The virtual address is shared amongst any number of processes and the processes includes a writing process and other processes that read from the virtual address. After writing the first value, the writing process writes a second value to the virtual address. Even after writing the second value, the first value is still visible to the other processes.

Global virtual address space consistency model

An approach is described that provides a distributed directory structure within a storage of an information handling system (a local node). A request is received with the request corresponding to a shared virtual address. The shared virtual address that is shared amongst a number of nodes that includes the local node and some remote nodes. A Global Address Space Directory (GASD) is retrieved that corresponds to a global virtual address space. The GASD is stored in a Coordination Namespace that is stored in a memory that is distributed amongst the nodes. A mapping that is included in the GASD is used to determine the node where the shared virtual address currently resides. The shared virtual address is then accessed from the node where it currently resides.

Distributed directory of named data elements in coordination namespace

A method for system level acceleration includes managing, by a system level acceleration server, an accelerator program running on an accelerator machine on behalf of a client, receiving, by the system level acceleration server, a communication from the client, interacting, by the system level acceleration server, with the accelerator program on behalf of the client according to the communication from the client, and generating, by the accelerator machine, a deliverable for provision to the client based upon an output of the accelerator program.

System level acceleration server

A method and structure for an in-place transformation of matrix data. For a matrix A stored in one of a standard full format or a packed format and a transformation T having a compact representation, blocking parameters MB and NB are chosen, based on a cache size. A sub-matrix Aof A, Ahaving size M=m*MB by N=n*NB, is worked on, and any of a residual remainder of A is saved in a buffer B. Sub-matrix Ais worked on by contiguously moving and contiguously transforming Ain-place into a New Data Structure (NDS), applying the transformation T in units of MB*NB contiguous double words to the NDS format of A, thereby replacing Awith the contents of T(A), and moving and transforming NDS T(A) to standard data format T(A) with holes for the remainder of A in buffer B. The contents of buffer B is contiguously copied into the holes of A, thereby providing in-place transformed matrix T(A).

Method and structure for fast in-place transformation of standard full and packed matrix data formats

A plurality of processing cores, are central storage unit having at least memory connected in a daisy chain manner, forming a daisy chain ring layout on an integrated chip. At least one of the plurality of processing cores places trace data on the daisy chain connection for transmitting the trace data to the central storage unit, and the central storage unit detects the trace data and stores the trace data in the memory co-located in with the central storage unit.

Distributed trace using central performance counter memory

A method (and structure) of executing a matrix operation, includes, for a matrix A, separating the matrix A into blocks, each block having a size p-by-q. The blocks of size p-by-q are then stored in a cache or memory in at least one of the two following ways. The elements in at least one of the blocks is stored in a format in which elements of the block occupy a location different from an original location in the block, and/or the blocks of size p-by-q are stored in a format in which at least one block occupies a position different relative to its original position in the matrix A.

Method and structure for producing high performance linear algebra routines using register block data format routines

A general computer-implement method and apparatus to optimize problem layout on a massively parallel supercomputer is described. The method takes as input the communication matrix of an arbitrary problem in the form of an array whose entries C(i, j) are the amount to data communicated from domain i to domain j. Given C(i, j), first implement a heuristic map is implemented which attempts sequentially to map a domain and its communications neighbors either to the same supercomputer node or to near-neighbor nodes on the supercomputer torus while keeping the number of domains mapped to a supercomputer node constant (as much as possible). Next a Markov Chain of maps is generated from the initial map using Monte Carlo simulation with Free Energy (cost function) F=&#x3a3;C(i,j)H(i,j)&#x2212; where H(i,j) is the smallest number of hops on the supercomputer torus between domain i and domain j. On the cases tested, found was that the method produces good mappings and has the potential to be used as a general layout optimization tool for parallel codes. At the moment, the serial code implemented to test the method is un-optimized so that computation time to find the optimum map can be several hours on a typical PC. For production implementation, good parallel code for our algorithm would be required which could itself be implemented on supercomputer.

Optimizing layout of an application on a massively parallel supercomputer

Apparatus for evaluating the performance of DMA-based algorithmic tasks on a target multi-core processing system includes a memory and at least one processor coupled to the memory. The processor is operative: to input a template for a specified task, the template including DMA-related parameters specifying DMA operations and computational operations to be performed; to evaluate performance for the specified task by running a benchmark on the target multi-core processing system, the benchmark being operative to generate data access patterns using DMA operations and invoking prescribed computation routines as specified by the input template; and to provide results of the benchmark indicative of a measure of performance of the specified task corresponding to the target multi-core processing system.

Performance evaluation of algorithmic tasks and dynamic parameterization on multi-core processing systems

A computerized method provides for an in-place transformation of matrix A data including a New Data Structure (NDS) format and a transformation T having a compact representation. The NDS represents data of the matrix A in a format other than a row major format or a column major format, such that the data for the matrix A is stored as contiguous sub matrices of size MB by NB in an order predetermined to provide the data for a matrix processing. The transformation T is applied to the MB by NB blocks, using an in-place transformation processing, thereby replacing data of the block Awith the contents of T(A).

Executing Multiple Instructions Multiple Data ('MIMD') programs on a Single Instruction Multiple Data ('SIMD') machine, the SIMD machine including a plurality of compute nodes, each compute node capable of executing only a single thread of execution, the compute nodes initially configured exclusively for SIMD operations, the SIMD machine further comprising a data communications network, the network comprising synchronous data communications links among the compute nodes, including establishing a SIMD partition comprising a plurality of the compute nodes; booting the SIMD partition in MIMD mode; executing by launcher programs a plurality of MIMD programs on compute nodes in the SIMD partition; and re-executing a launcher program by an operating system on a compute node in the SIMD partition upon termination of the MIMD program executed by the launcher program.

Executing Multiple Instructions Multiple Data ('MIMD') programs on a Single Instruction Multiple Data ('SIMD') machine

Executing MIMD programs on a SIMD machine, including establishing on the SIMD machine a plurality of SIMD partitions; booting a first SIMD partition in MIMD mode; executing, on a compute node of the first SIMD partition booted in MIMD mode, a MIMD accelerator program; executing a SIMD program in a second SIMD partition, one instance of the SIMD program executing on each compute node of the second SIMD partition, each instance of the SIMD program carrying out a portion of the data processing effected by the SIMD program; and accelerating, by an instance of the SIMD program through the MIMD accelerator program, a portion of the data processing of the instance of the SIMD program.

Executing multiple instructions multiple date ('MIMD') programs on a single instruction multiple data ('SIMD') machine

A method for evaluating performance of DMA-based algorithmic tasks on a target multi-core processing system includes the steps of: inputting a template for a specified task, the template including DMA-related parameters specifying DMA operations and computational operations to be performed; evaluating performance for the specified task by running a benchmark on the target multi-core processing system, the benchmark being operative to generate data access patterns using DMA operations and invoking prescribed computation routines as specified by the input template; and providing results of the benchmark indicative of a measure of performance of the specified task corresponding to the target multi-core processing system.

Growing community of inventors

Yorktown Heights, NY, United States of America

James C Sexton