International Business Machines Corporation

Arquimedes Martinez Canedo

Hideaki Komatsu

Takeo Yoshizawa

A computer-implemented method, system, and article of manufacture for parallelizing a code configured by coupling a functional block having an internal state and a functional block without any internal state. The method includes: creating and storing a graphical representation where functional blocks are chosen as nodes and connections between functional blocks are chosen as links; visiting the nodes on the graphical representation sequentially, detecting inputs from functional blocks without any internal state to functional blocks having an internal state and storing these functional blocks as a set of use blocks, and detecting inputs from functional blocks having an internal state to functional blocks without any internal state and storing these functional blocks as a set of definition blocks; and forming strands of functional blocks based on information on the set of use blocks and information on the set of definition blocks stored in association with the functional blocks.

Parallelization method, system and program

Paralleling processing system and method. When clusters are formed based on strongly connected components, a single cluster (fat cluster) having at least a predetermined number of blocks, or an expected processing time exceeding a predetermined threshold, is formed. The fat cluster is subjected to an unrolling process to make multiple copies of the processing of the fat cluster and to assign the copies to individual processors. Processing of the fat cluster is executed by the multiple processor devices in a pipelined manner. If a fat cluster to be iteratively executed cannot be executed in the pipelined manner because a processing result of an niteration of the fat cluster depends on a processing result of a preceding iteration of the fat cluster an input value needed for execution of the fat cluster is generated based on a certain prediction, and the fat cluster is speculatively executed.

Paralleling processing method, system and program

A system, method and program to improve the processing speed of a simulation system. A processing system finds an entry point so that functional blocks cover a broad range. The processing system places code of a look-ahead dispatcher for assigning processing. The look-ahead dispatcher monitors an input state at the entry point to determine whether the input state is a stable state. If the input state is stable, the look-ahead dispatcher calls an adaptive execution module at some frequency or otherwise calls an idle execution module. The adaptive execution module performs processing on multiple timestamps at once. When a discrete system receives an input event, the look-ahead dispatcher calls a recovery execution module. Based on the input event on that occasion, the timestamp, and a value stored in a state vector, the recovery execution module calculates a state for which recovery is performed.

Growing community of inventors

Kanagawa, Japan

Arquimedes Martinez Canedo