Sun Microsystems, Inc.

Thinking Machines Corporation

Oracle International Corporation

Oracle America, Inc.

Other

Continuum Software, Inc.

Steven J Sistare

Joshua E Simons

Karen C Jourdenais

James L Frankel

Richard Title

Donald C Allen

Richard Bowker

Terry D Dontje

Michael J McCarthy

Eric Christopher Saxe

Jonathan Chew

Pallab Bhattacharya

Nicholas O'Donnell

Anthony L Kimball

Michael L Best

Rafael Vanoni Polanczyk

John A Sullivan

Theodore F Tabloski, Jr

David Plauger

Nicholas J Nevin

A method for specifying the layout of computing system resources includes receiving a request for a virtual machine, the request comprising a processor requirement and a memory requirement, gathering resource groupings, each resource grouping comprising a latency penalty between a processor and at least a memory. The method further includes calculating a proportionality for each resource grouping, wherein the proportionality comprises a minimum of: an amount of the processor requirement the resource grouping can satisfy and an amount of the memory requirement the resource grouping can satisfy, sorting the resource groupings based on at least the proportionality for each resource grouping to create an ordered list and binding, based on the ordered list, at least one resource group to the virtual machine, wherein the at least one resource group satisfies the processor requirement and the memory requirement.

Method and system for specifying the layout of computer system resources

A resource credit tree for resource management includes leaf nodes and non-leaf nodes. The non-leaf nodes include a root node and internal nodes. Resource management includes initializing a operation corresponding to a resource pool, selecting, using a hash function, a leaf node of a resource credit tree, and identifying a number of available credits of the leaf node. Resource management may further include traversing, using a backward traversal path, from the leaf node to a non-leaf node based on determining that the number of available credits is less than a required number of credits or determining that capacity of the leaf node is less than the summation of the number of credits to free to the resource credit tree and the number of available credits. Resource management may further allocating and freeing credits from and to the resource credit tree.

Method for resource management allocating and freeing credits from and to a resource credit tree

A method for scheduling a new thread involves identifying a criticality level of the new thread, selecting a processor group according to the criticality level of the new thread and an existing assigned utilization level of the processor group to obtain a selected processor group, increasing an assigned utilization level of the selected processor group based on the new thread, and executing the new thread by the selected processor group.

Method and system for scheduling threads

A debugger for aiding in the debugging of multi-threaded program, in response to an event such as, for example, a breakpoint in a thread which has caused an operating system to stop execution of all threads, identifies the thread which contained the breakpoint. After identifying the thread which contained the breakpoint, the debugger enables the operating system to resume execution of the other threads, that is, the threads which did not contain the breakpoint. By allowing the other threads, that is, the threads which did not contain the breakpoint, to continue execution, the debugger's impact on program execution is substantially reduced, particularly for programs which contain a large number of threads.

Parallel and asynchronous debugger and debugging method for multi-threaded programs

A system is described for controlling co-scheduling of processes in a computer comprising at least one process and a spin daemon. The process, when it is waiting for a flag to change condition, transmits a flag monitor request to the spin daemon and enables itself to be de-scheduled. The spin daemon, after receiving a flag monitor request, monitors the flag and, after the flag changes condition, enables the process to be re-scheduled for execution by the computer. Since the spin daemon can monitor flags for a number of processes, the ones of the processes that are waiting will not need to be executed, allowing other processes that are not just waiting to be processed more often.

System and method for controlling co-scheduling of processes of parallel program

A communications arrangement is described for facilitating transfer of messages among a plurality of processes in a computer system. The communications arrangement comprises a channel data structure, a status daemon and an exit handler. The channel data structure includes a channel status flag normally having one of a plurality of conditions, and a plurality of storage locations each configured to receive message information. The status daemon is configured to determine the operational status of the processes. The exit handler is configured to, in response to the status daemon determining a predetermined condition in connection with at least one of the processes, condition the channel status flag to another of the conditions, thereby to indicate to the other processes a failure condition in connection with the communications arrangement.

Robust communications connection system and method for shared memory

A data visualization arrangement for facilitating the display of data items comprising a selected region of an object in a selected one of a plurality of display modes. The arrangement comprises a data object store, an interface, an object region retrieval component and a display. The data object store stores the data object, the data object comprising a plurality of data items in a predetermined organization. The interface receives a region identification for identifying a particular region of the object and a display mode identification. The object region retrieval component retrieves data items from a region of the data object as identified by the region identification received by the interface. Finally, the display receives the data items as retrieved by the object region retrieval mechanism and displays them in the display mode as identified by the display mode identification.

Visualization system including data navigator for use with digital computer system

A communication arrangement facilitates transfer of messages among a plurality of processes in with a computer, the computer having a memory shared by the processes. The communication arrangement comprises, allocated to each process, a plurality of buffers, and a plurality of postboxes each associated with one of the other processes. Each process includes a message size determination module and a message transfer module. The message size determination module is configured to determine whether a message to be transferred to another process can be accommodated by a postbox. The message transfer module is configured to (i) in response to a positive determination by the message size determination module, store the message in the postbox associated with the process as allocated to the other process, and (iii) in response to a negative determination by the message size determination module, store the message in one of the buffers allocated thereto, and providing a pointer to the one of the buffers in the postbox associated with the process as allocated to the other process.

System and method for allocating buffers for message passing in a shared-memory computer system

A system for generating a 'where graph' for a parallel processing system comprising a parallel processing array controlled by a control processor. The processing array comprises a plurality of processing elements each processing a program comprising a main routine and a plurality of subroutines, the main routine calling at least one subroutine and at least some of said subroutines calling others of said subroutines. The control processor determines a consolidated program routine sequence identifying sequences of routines extending from said main routine to a current routine being processed by said at least some of said processing elements and generates and displays, on, for example, a video display, the program routine sequence in graph form. The graph comprises a plurality of graph nodes, each representing a routine, with the graph nodes being interconnected to depict the consolidated program routine sequence.

System for generating and graphically displaying call stack information

A parallel program development and processing system includes a parallel program development section and a parallel program execution section. The parallel program development section allows a program developer to develop programs for execution by a parallel computer system using a predetermined set of components which can be selected by the program developer, using a graphical user interface, and linked in a dataflow graph that represents the order of operations to be performed by the program on the data to be processed. After the program developer has developed the graph, the parallel program development section generates executable program code from the instances of the components and their interconnections, for execution by a parallel computer. In the executable program code, instances of the components selected by the program developer form executable objects which are executed by the parallel computer under control of a run-time system that includes an execution control object. The execution control object controls processing of the executable objects using a dataflow processing model, in which each executable object processes data which it receives, but blocks if it has no data to process or if a downstream element is unable to receive data that it has processed. When an executable object blocks, the execution control object is called to correct the blocking condition. When the blocking condition is corrected, the execution control object enables the blocked executable object to resume operations.

System and method for developing computer programs for execution on

A data visualization arrangement facilitates the display of a selected region of a multi-dimensional data object in a selected one of a plurality of display modes. The arrangement comprises a data object store, an interface, an object region retrieval component and a display. The data object store stores the data object, the data object comprising a plurality of data items in a predetermined organization. The interface receives a region identification for identifying a particular region of the object and a display mode identification. The object region retrieval component retrieves data items from a region of the data object as identified by the region identification received by the interface. Finally, the display receives the data items as retrieved by the object region retrieval mechanism and displays them in the display mode as identified by the display mode identification.

System and method for displaying a selected region of a

A debugger for use in connection with a parallel computer including a plurality of processing nodes. The debugger enables the operator to establish a processing node set in response to certain criteria, such as the respective identifications of the processing node and their prior processing under the debugger. The debugger, in response to a processing node set establishment command from the operator, enables each processing node to establish and condition a set membership flag in response to the operator-supplied set membership criteria. The debugger is then responsive to a debugging command from the operator to enable the processing nodes to use their respective set membership flags to condition their execution of the debugging command.

Parallel processing system including arrangement for establishing and

An optimizer for optimizing an intermediate representation (IR) tree having multiple nodes. The IR tree represents a partial compilation of a source code. The source code is written using a high level language supporting data parallel processing. According to the present invention, the optimizer optimizes the IR tree by minimizing the number and size of temporary parallel variables in the IR tree. Such minimization optimizes the IR tree because temporary parallel variables require an enormous amount of space in memory.

System and method for parallel variable optimization

A compiler for compiling a computer program wherein the computer program is adapted for use with a data parallel computer. The compiler comprises an optimizer which optimizes the compiled code. In optimizing the compiled code, the optimizer performs live variable analysis. With regard to performing live variable analysis, the optimizer of the present invention is adapted for use with data parallel languages. Additionally, the optimizer is computationally efficient at compile time. Further, the optimizer operates in an incremental manner.

Growing community of inventors

Somerville, MA, United States of America

Steven J Sistare