Abstract: A surface computer includes an address generator for generating an address for adjusting surface region data concerning at least a storage region and a concurrent computer, provided at a subsequent stage of the address generator, having a plurality of unit computers.

Abstract: Embodiments of the invention relate to a processing cell for use in computing systems. Generally, a processing cell generates remote instructions to be received and processed by at least one other processing cell. A processing cell may include a program counter, an instruction memory, and appropriate elements such as a branch lookup, a branch unit, etc. Alternatively, the processing cell may include a state machine that replaces the program counter and the instruction memory. Embodiments of the invention are able to support the VLIW mode, the MIMD) mode, a mixture of both modes of execution, etc.

Abstract: A hybrid switching module includes a hybrid switching module processor data channel; a hybrid switching module main data channel; an input/output link data channel; a switch coupled to the hybrid switching module processor data channel; and a bridge coupled to the hybrid switching module main data channel; wherein the switch selectively coupled to the bridge and selectively coupled to the input/output link data channel, wherein the hybrid switching module processor data channel is thereby selectively coupled to the bridge and selectively coupled to the input/output link data channel.

Abstract: A data processor can assign a greater number of operations to instruction codes with shorter length, thereby implementing high performance, high code efficiency and low cost data processor. The data processor is a VLIW (Very Long Instruction Word) system that can execute a plurality of operations in parallel, and specify the execution sequence of the operations. It can assign a plurality of operations to the same operation code, and the operations that are executed in a second or subsequent sequence are limited to only predetermined operations among the plurality of operations.

Abstract: A system for designing and implementing digital integrated circuits utilizing a set of synchronized sequencers that permit quick and efficient parallel processing of system level designs. The system and method converts digital schematics and hardware description language (HDL) based designs into a set of logic equations and single bit arithmetic-logic operations executed by a set of parallel operating sequencers. The system includes software for converting netlists and HDL designs into Boolean logic equations, and a compiler for distributing these logic equations between multiple sequencers. Each sequencer is comprised of a logic processor and the associated program memory for storing the executable code of the assigned Boolean logic equations and data memory for storing the results of processing of logic equations. To synchronize execution of logic equations by multiple sequencers, all program memories are addressed by one common address register.

Abstract: There is disclosed a technique for accessing a register file which comprises defining a first register address as a plurality of bits and using said first register address to access said register file generating a second register address by using a sequence of said plurality of bits with at least one of said plurality of bits supplied via a unitary operator, the unitary operator being effective to selectively alter the logical value of said bit depending on its logical value in the first register address, and using said second register address to access said register file. A computer system for carrying out such a technique is also enclosed.

Abstract: An embedded processor complex contains multiple protocol processor units (PPUs). Each unit includes at least one, and preferably two independently functioning core language processors (CLPs). Each CLP supports dual threads thread which interact through logical coprocessor execution or data interfaces with a plurality of special purpose coprocessors that serve each PPU. Operating instructions enable the PPU to identify long and short latency events and to control and shift priority for thread execution based on this identification. The instructions also enable the conditional execution of specific coprocessor operations upon the occurrence or non occurrence of certain specified events.

Abstract: Techniques for achieving the effects of significantly reducing the amount of computer memory needed to simulate the behavior of a multi-stage pipelined processor, as well as, significantly increasing the performance of the simulation process by eliminating the storing and copying of redundant information are described. These beneficial effects are achieved by reordering the chronological sequence of execution of software models of the various pipeline stages with respect to the actual instruction-flow sequence implemented by the processor hardware. This approach takes advantage of the independence of the stages within a cycle to make the results computed by a previous stage directly available to its subsequent stage without the use of transient data space or data copying. In particular, it is shown how to apply this technique to the simulation of a multi-parallel-stage VLIW array processor, such as the manifold array (ManArray) processor.

Abstract: A multiprocessor system including a master processor, a plurality of processor elements, each of which is provided with a local memory, the processor elements being controlled in accordance with commands from the foregoing master processor, and a global memory shared by the plurality of processor elements is disclosed. The processor elements are provided with a command pooling buffer capable of accumulating a plurality of commands, respectively. DMA controllers are also provided with a command pooling buffer capable of accumulating a plurality of commands, respectively. The master processor persistently issues a plurality of commands to the DMA controller and each processor element. A counter array manages the number of the issued commands which have received no response. When the responses are returned with respect to all issued commands, the counter array notifies the master processor of this.

Abstract: A circuit is provided to provide instruction streams to a processing device: embodiments of the circuit are appropriate for use with RISC CPUs, whereas other embodiments are useable with other processing devices, such as small processing devices used in a field programmable array. The circuit receives an external instruction stream which provides a first set of instruction values, and has a memory which contains a second set of instruction values. Two or more outputs provide instruction streams to the processing device. The circuit has a control input in the form of a mask which causes a selection means to allocate bits from the first and second sets of instruction values to different instruction streams to the processing device.

Abstract: A processing cell for use in computing systems is disclosed. Generally, a processing cell generates branch commands to be received and processed by at least one other processing cell. A processing cell may be instruction-based that includes a program counter, an instruction memory, and appropriate elements such as a branch lookup, a branch unit, an ALU, etc., for computations. Alternatively, the processing cell is state-machine based, which is comparable to an instruction-based cell, but includes a state machine that replaces the program counter and the instruction memory. Embodiments of the invention are able to support at least the VLIW mode, the MIMD mode, and a mixture of both modes of execution.

Abstract: Processing element to processing element switch connection control is described using a receive model that precludes communication hazards from occurring in a synchronous MIMD mode of operation. Such control allows different communication topologies and various processing effects such as an array transpose, hypercomplement or the like to be efficiently achieved utilizing architectures, such as the manifold array processing architecture. An encoded instruction method reduces the amount of state information and setup burden on the programmer taking advantage of the recognition that the majority of algorithms will use only a small fraction of all possible mux settings available. Thus, by means of transforming the PE identification based upon a communication path specified by a PE communication instruction an efficient switch control mechanism can be used.

Abstract: Instruction and data registers of processors of a multiprocessing computing system are joined and forked to allow processing in multiple modes of operation. When joined, the registers of the processors each contain a same piece of information, hence generating single instruction and data streams. In contrast, when forked, the registers of the processors contain different pieces of information, thereby generating multiple instruction and data streams. Additionally, information may be stored into partitions of memory and fetched and broadcast by processors local to the particular memory sections thereby resulting in a faster cycle time.

Abstract: An array of processor elements has multiple instruction streams and multiple data streams broadcast to all of the processor elements. The processor elements are each connected to multiple neighbouring processor elements within a cruciate neighbourhood. The architecture is suitable for use in fine-grained applications. The array may have a processor element for each pixel of an image. The array is preferably provided on a single integrated circuit having 10,000 or more processor elements.

Abstract: Methods, signals, devices, and systems are provided for matching tasks with processing units. A region within a multi-faceted task space is allocated to a processing unit. A point in the multi-faceted task space is assigned to a task. The task is then associated with the processing unit if the region allocated to the processing unit is close to the point assigned to the task. The region allocated to a processing unit may be changed. If no assigned point for a task is sufficiently close to any allocated processing unit region, the task is suspended. Overlapping regions may be assigned to different processing units. In some implementations, the union of the allocated regions covers the task space, while in others it does not. Regions may also be allocated to wait conditions and one or more dimensions of a region may be allocated to conventional processor allocators.

Abstract: Synergetic computing system contains a unidirectional each-to-each switchboard (2) with N inputs and 2*N outputs, with N functional units (1.1, . . . , 1.N) attached, each unit executing its own program (a sequence of binary and unary operations). Results of operations are sent to the switchboard and used as operands by other functional units. The final result of computation is formed as a result of programmed coordinated interaction (synergy) of the functional units (1.1, . . . , 1.N). Two operating modes are suggested, synchronous and asynchronous. The synchronous mode uses a two-stage pipeline and duration of individual operations has to be taken into account when writing the code. An instruction using a result of another instruction should begin execution in the cycle immediately following the generation of this result. In the asynchronous mode, programming does not need to account for instruction duration and operations are performed upon operand availability.

Abstract: A system for designing and implementing digital integrated circuits utilizing a set of synchronized sequencers that permit quick and efficient parallel processing of system level designs. The system and method converts digital schematics and hardware description language (HDL) based designs into a set of logic equations and single bit arithmetic-logic operations executed by a set of parallel operating sequencers. The system includes software for converting netlists and HDL designs into Boolean logic equations, and a compiler for distributing these logic equations between multiple sequencers. Each sequencer is comprised of a logic processor and the associated program memory for storing the executable code of the assigned Boolean logic equations and data memory for storing the results of processing of logic equations. To synchronize execution of logic equations by multiple sequencers, all program memories are addressed by one common address register.

Abstract: There is disclosed a technique for accessing a register file which comprises defining a first register address as a plurality of bits and using said first register address to access said register file generating a second register address by using a sequence of said plurality of bits with at least one of said plurality of bits supplied via a unitary operator, the unitary operator being effective to selectively alter the logical value of said bit depending on its logical value in the first register address, and using said second register address to access said register file. A computer system for carrying out such a technique is also enclosed.

Abstract: A system for designing and implementing digital integrated circuits utilizing a set of synchronized sequencers that permit quick and efficient parallel processing of system level designs. The system and method converts digital schematics and hardware description language (HDL) based designs into a set of logic equations and single bit arithmetic-logic operations executed by a set of parallel operating sequencers. The system includes software for converting netlists and HDL designs into Boolean logic equations, and a compiler for distributing these logic equations between multiple sequencers. Each sequencer is comprised of a logic processor and the associated program memory for storing the executable code of the assigned Boolean logic equations and data memory for storing the results of processing of logic equations. To synchronize execution of logic equations by multiple sequencers, all program memories are addressed by one common address register.

Abstract: A programmable processing engine processes transient data within an intermediate network station of a computer network. The engine comprises an array of processing elements symmetrically arrayed as rows and columns, and embedded between input and output buffer units with a plurality of interfaces from the array to an external memory. The external memory stores non-transient data organized within data structures, such as forwarding and routing tables, for use in processing the transient data. Each processing element contains an instruction memory that allows programming of the array to process the transient data as processing element stages of baseline or extended pipelines operating in parallel.

Abstract: In a computer having a single execution pipeline, the invention provides a method for executing paired MMX-type instructions. The method includes executing two MMX-type instructions as paired MMX instructions. If execution of the paired MMX instructions causes an exception, pairing of instructions is disabled, and the two MMX-type instructions are re-executed in sequential fashion. Paired execution is re-enabled following re-execution of the two MMX-type instructions.

Abstract: A SIMD machine employing a plurality of parallel processor (PEs) in which communications hazards are eliminated in an efficient manner. An indirect Very Long Instruction Word instruction memory (VIM) is employed along with execute and delimiter instructions. A masking mechanism may be employed to control which PEs have their VIMs loaded. Further, a receive model of operation is preferably employed. In one aspect, each PE operates to control a switch that selects from which PE it receives. The present invention addresses a better machine organization for execution of parallel algorithms that reduces hardware cost and complexity while maintaining the best characteristics of both SIMD and MIMD machines and minimizing communication latency. This invention brings a level of MIMD computational autonomy to SIMD indirect Very Long Instruction Word (iVLIW) processing elements while maintaining the single thread of control used in the SIMD machine organization.

Abstract: A high speed remote storage controller system for a computer system has cluster nodes of symmetric multiprocessors. A plurality of clusters of symmetric multiprocessors each of has a plurality of processors, a shared cache memory, a plurality of I/O adapters and a main memory accessible from the cluster. Each cluster has an interface for passing data between cluster nodes of the symmetric multiprocessor system. Each cluster has a local interface and interface controller. The system provides one or more remote storage controllers each having a local interface controller and a local-to-remote data bus. A remote resource manager manages the interface between clusters of symmetric multiprocessors. The remote store controller is responsible for processing data accesses across a plurality of clusters and processes data storage operations involving shared memory.

Abstract: According to the SIMD control parallel processing method for performing common operation in parallel in a plurality of elements, comprising first retaining means for retaining operation data specified by n-bit for each of said plurality of elements; second retaining means for previously retaining operated result with all possible combinations comprising said data according to a predetermined operation; and selecting means for selecting said operated data retained in said first retaining means from among said operated results retained by said second retaining means, from among retained data obtained through operation, data corresponding to that resultant from the operation is selected for each element, thereby enabling a configuration to be simplified, smaller and less costly.

Abstract: Processing element to processing element switch connection control is described using a receive model that precludes communication hazards from occurring in a synchronous MIMD mode of operation. Such control allows different communication topologies and various processing effects such as an array transpose, hypercomplement or the like to be efficiently achieved utilizing architectures, such as the manifold array processing architecture. An encoded instruction method reduces the amount of state information and setup burden on the programmer taking advantage of the recognition that the majority of algorithms will use only a small fraction of all possible mux settings available. Thus, by means of transforming the PE identification based upon a communication path specified by a PE communication instruction an efficient switch control mechanism can be used.

Abstract: Three parallel instruction processing pipelines of a microprocessor share two data memory ports for obtaining operands and writing back results. Since a significant proportion of the instructions of a typical computer program do not require reading operands from the memory, the probability is high that at least one of any three program instructions to be executed at the same time need not fetch an operand from memory. The two memory ports are thus connected at any given time with the two of the three pipelines which are processing instructions that require memory access, the pipeline without access to the memory processing an instruction that does not need it. To do so, the added third pipeline need not have all the same resources as the other two pipelines, so its stages are made to have a reduced capability in order to save space and reduce power consumption.

Abstract: A hierarchical instruction set architecture (ISA) provides pluggable instruction set capability and support of array processors. The term pluggable is from the programmer's viewpoint and relates to groups of instructions that can easily be added to a processor architecture for code density and performance enhancements. One specific aspect addressed herein is the unique compacted instruction set which allows the programmer the ability to dynamically create a set of compacted instructions on a task by task basis for the primary purpose of improving control and parallel code density. These compacted instructions are parallelizable in that they are not specifically restricted to control code application but can be executed in the processing elements (PEs) in an array processor. The ManArray family of processors is designed for this dynamic compacted instruction set capability and also supports a scalable array of from one to N PEs.

Abstract: A highly parallel data processing system includes an array of n processing elements (PEs) and a controller sequence processor (SP) wherein at least one PE is combined with the controller SP to create a Dynamic Merged Processor (DP) which supports two modes of operation. In its first mode of operation, the DP acts as one of the PEs in the array and participates in the execution of single-instruction-multiple-data (SIMD) instructions. In the second mode of operation, the DP acts as the controlling element for the array of PEs and executes non-array instructions. To support these two modes of operation, the DP includes a plurality of execution units and two general-purpose register files. The execution units are “shared” in that they can execute instructions in either mode of operation. With very long instruction word (VLIW) capability, both modes of operation can be in effect on a cycle by cycle basis for every VLIW executed.

Abstract: A processor particularly useful in multimedia applications such as image processing is based on a stream programming model and has a tiered storage architecture to minimize global bandwidth requirements. The processor has a stream register file through which the processor's functional units transfer streams to execute processor operations. Load and store instructions transfer streams between the stream register file and a stream memory; send and receive instructions transfer streams between stream register files of different processors; and operate instructions pass streams between the stream register file and computational kernels. Each of the computational kernels is capable of performing compound vector operations. A compound vector operation performs a sequence of arithmetic operations on data read from the stream register file, i.e., a global storage resource, and generates a result that is written back to the stream register file.

Abstract: A pipelined data processing unit includes an instruction sequencer and n functional units capable of executing n operations in parallel. The instruction sequencer includes a random access memory for storing very-long-instruction-words (VLIWs) used in operations involving the execution of two or more functional units in parallel. Each VLIW comprises a plurality of short-instruction-words (SIWs) where each SIW corresponds to a unique type of instruction associated with a unique functional unit. VLIWs are composed in the VLIW memory by loading and concatenating SIWs in each address, or entry. VLIWs are executed via the execute-VLIW (XV) instruction. The iVLIWs can be compressed at a VLIW memory address by use of a mask field contained within the XV1 instruction which specifics which functional units are enabled, or disabled, during the execution of the VLIW. The mask can be changed each time the XV1 instruction is executed, effectively modifying the VLIW every time it is executed.

Abstract: An alternate route to improved multimedia performance without replacing the central processor unit (CPU) is presented, through the utilization of general-purpose components available in a computer. The method relies on the use of integrated circuit memory boards having a data port for directly inputting encoded image signals from an I/O device into the memory. An on-board decoder provided on the IC memory is used to decode the variable-length encoded input signals. This approach enalbes to reduce the computational load on the CPU so that the usual bottleneck which is the slow process of data exchange between the CPU and the memory boards is eliminated. The CPU directly accesses the processed image data in the memory and displays the final image on the monitor. This route to increasing the image processing speed of a computer has considerable merits because it is low cost and is readily applicable to mass-produced IC memories with only a few additional fabrication steps.

Abstract: A multiprocessor computer system is provided with a BIOS that allows parallel execution of system initialization tasks by at least two processors to reduce system boot-up time. At power-on, one of the processors is designated as a bootstrap processor and the remaining processors are designates as application processors. The processors are coupled to a shared memory module by a shared processor bus. The bootstrap processor is configured to instruct the application processor to test and initialize memory locations in the shared memory module while the bootstrap processor proceeds with other system initialization tasks which may include determining the system configuration, initializing peripheral devices, testing the keyboard, and setting up the BIOS data area with configuration information. After completing its tasks, the bootstrap processor determines whether the application processor has completed the memory test, and if so, the bootstrap processor proceeds to locate and execute an operating system.

Abstract: A multi-processor array organization is dynamically configured by the inclusion of a configuration topology field in instructions broadcast to the processors in the array. Each of the processors in the array is capable of performing a customized data selection and storage, instruction execution, and result destination selection, by uniquely interpreting a broadcast instruction by using the identity of the processor executing the instruction. In this manner, processing elements in a large multi-processing array can be dynamically reconfigured and have their operations customized for each processor using broadcast instructions.

Abstract: A method and apparatus for syncing multiple bus masters (110, 120, 130, 140) utilize a sync bus (160) to communicate synchronization information between a plurality of bus masters. For each bus master, a check is made after instruction fetch (350) for a "sync indicator" (360). When a sync indicator is detected (360), the bus master transmits sync signals on the sync bus (160) indicating which masters it needs to sync with (365). The bus master then stalls instruction dispatch (390) until the corresponding sync signals are received (370) on the sync bus (160) from the other bus masters. A further programmed delay may be introduced after the bus masters have matched sync signals (375), enabling through testing for timing windows.

Abstract: A method is disclosed for reproducible sampling of data items of a dataset which is shared across a plurality of nodes of a parallel data processing system.In data mining of large databases, segmentation of the database is often necessary either to obtain a summary of the database or prior to an operation such as link analysis. A sample of data records are taken to create an initial segmentation model. The records of this sample and the initial model created from them can be critical to the results of the data mining process, and the initial model may not be reproducible unless the same sampling of data records is repeatable. Reproducible sampling is enabled without polling of all nodes to locate particular records. Parametric control information with a small number of control parameters is generated which describes the particular partitioning of the dataset. The parametric control information enables computing of the location of a data record.

Abstract: A circuit arrangement and method utilize a distributed extensible processing architecture to allocate various DSP functions or operations between multiple processing cores disposed on an integrated circuit device. Each processing core includes one or more hardwired datapaths to provide one or more DSP operations. Moreover, each processing core includes a programmable controller that controls the operation of each hardwired datapath via a local computer program executed by the controller. Furthermore, the processing cores are coupled to one another over a communications bus to permit data to be passed between the cores and thereby permit multiple DSP operations to be performed on data supplied to the device.

Abstract: There is disclosed a multi-processor system arranged, in one embodiment, as an image and graphics processor. The processor is structured with several individual processors all having communication links to several memories. The individual processors can, on a cycle by cycle basis, be grouped in any configuration to run in synchronism (but from different instruction streams) with the other processors in that group. More than one such synchronized group can be formed concurrently. A crossbar switch serves to establish the processor memory links and the entire image processor, including the individual processors, the crossbar switch and the memories are contained on a single silicon chip.

Abstract: A remote resource management system for managing resources in a symmetrical multiprocessing comprising a plurality of clusters of symmetric multiprocessors having interfaces between cluster nodes of the symmetric multiprocessor system. Each cluster of the system has a local interface and interface controller. There are one or more remote storage controllers each having its local interface controller, and a local-to-remote data bus. The remote resource manager manages the interface between two clusters of symmetric multiprocessors each of which clusters has a plurality of processors, a shared cache memory, a plurality of I/O adapters and a main memory accessible from the cluster. This remote resource manager manages resources with a remote storage controller to distribute work to a remote controller acting as an agent to perform a desired operation without requiring knowledge of a requester who initiated the work request.