Abstract: An asynchronous circuit having a pipelined completion mechanism to achieve improved throughput.

Abstract: Unlike prior art synchronizers and asynchronous arbiters that produce glitches in their outputs, the present invention provides a failure-free synchronizer that can sample an arbitrary and unstable inputs while maintaining zero probability of system failure. In particular, the invention addresses the synchronization failure problem and the lack of a metastable state in prior art synchronizers. Prior attempts have shown that the conditions rex and rex (where re is the control input and x is the data input) cannot be arbitrated. To overcome this, embodiments of the present invention introduce explicit signals a0 and a1 to hold the values rex and rex, respectively. One embodiment is a fast synchronizer. It has four main components—an input integrator, an inverting component, a SEL component and an output filter. Another embodiment of the present invention is a safe synchronizer that meets the strictest QDI design requirements. Other embodiments use a standard arbiter and a killable arbiter for arbitration.

Abstract: An asynchronous processor having pipelined instruction fetching and execution to implement concurrent execution of instructions by two or more execution units. A writeback unit is connected to execution units and memory units to control information updates and to handle precise exception. A pipelined completion mechanism can be implemented to improve the throughput.

Abstract: An asynchronous processor having pipelined instruction fetching and execution to implement concurrent execution of instructions by two or more execution units. A writeback unit is connected to execution units and memory units to control information updates and to handle precise exception. A pipelined completion mechanism can be implemented to improve the throughput.

Abstract: Exception handling systems and techniques for handling exceptions and sequencing conflicts in an asynchronous processor. Two designated queues are used to respectively keep program counter values of instructions and the assignments of the execution units for executing the instructions according to the program order. An asynchronous circuit is coupled between the program counter unit and the write-back unit of the processor to provide asynchronous communications.

Abstract: An asynchronous and delay-insensitive data processor comprises a plurality of components communicating with each other and synchronizing their activities by communication actions on channels and buses. Each component consists of a control part and a data part. All control parts are implemented with a lazy-active-passive handshake protocol and a sequencing means called a left/right buffer that provides the minimal sequencing constraints on the signals involved. The data parts comprise novel asynchronous ALU, buses, and registers. The control parts and data parts are connected together in an asynchronous and delay-insensitive manner.

Abstract: An asynchronous circuit having a pipelined completion mechanism to achieve improved throughput.

Abstract: A circuit for performing prefix computation in an asynchronous digital processor by implementing a serial process and a tree process for the same prefix computation in parallel. The first output from either processes is selected and used for the subsequent operation. For a prefix computation with N inputs, an average-case latency of O(loglog N) can be achieved. Buffering can be used for a full-throughout operation.

Abstract: The basic building block of the invention is an inverter gate consisting of two stages: The first stage is an input logic switching stage consisting of a depletion mode pull-up FET whose gate is the input node and whose source-to-drain channel is connected in series through a level-shifting Schottky diode with the source-to-drain channel of an depletion mode pull-down FET between drain and source voltage rails. The source of the pull-up FET is connected to the diode's anode while the drain of the pull-down FET is connected to the diode's cathode and is the output node of the input logic switching stage. The level-shifting diode isolates the output node from the input node, which allows the input voltage to switch rail-to-rail without causing problems.

Abstract: A multiprocessor computer system having a two-dimensional array of modules. Each module comprises two source connections for an activation signal for an algorithm, a program memory, a processor element and two destination connections for an activation signal for a corresponding algorithm. The processor element can split the task of a recursive algorithm into two partial tasks (of the same algorithm) or can execute an elementary partial task of this algorithm. The results of the partial tasks are returned in the direction wherefrom the relevant activation signal originated. For this purpose, each module in the array is connected to four neighboring modules. The peripheral modules are not only connected to at least two neighboring modules, but also to at least one module on an opposite side of the array. In each module there is one central process and for each connection there is one channel process, the central process communicating only with the channel processes of the relevant module.