Honeywell Information Systems Inc.

Bull Hn Information Systems, Inc.

Honeywell Bull, Inc.

Bull Nh Information Systems, Inc.

Thomas F. Joyce

Richard P. Kelly

Ming T. Miu

Robert V. Ledoux

Forrest M. Phillips

Richard P Kelly

Thomas F Joyce

Jian-Kuo Shen

Robert V Ledoux

Thomas O Holtey

James C Raymond

Steven S Noyes

Richard A Lemay

George J Barlow

James W Keeley

Ming T Miu

David E Cushing

Gary J Goss

Robert C Miller

Chester M Nibby, Jr

Thomas L Murray, Jr

Michel M Raguin

Forrest M Phillips

Deborah K Staplin

Daniel G Peters

Robert V Ledoux, Deceased

James L Brown

Richard P Wilder

The decoding of certain instructions cause an instruction unit of a production line data processing system to stall. Instructions still in the production line are executed, but no new instructions are sent into the production line until the instruction that caused the stall condition is executed. The execution of the instruction that caused the stall is completed by an execution unit taking over control of an address unit.

Control store addressing from multiple sources

The pipelined central processing system (CSS) units of a multiprocessor system are tightly coupled to connect in common to a system bus for sharing main memory and input/output controllers/devices. The CSS includes several circuit boards for the different VLSI circuit chip pipelined stages and associated control circuits in addition to the bus interface unit (BIU) circuits. Each board includes one or more unusual event (UEV) detector circuits for signaling when the behavior of a stage is abnormal. The UEV fault signals from each board are collected by the BIU board. When a UEV fault is detected, the BIU board circuits prevent any further communications with the system bus and broadcasts the UEV fault signal to the other boards causing the different pipelined stages to emulate the completion of the instructions within the pipeline thereby flushing it. It is thereafter placed in a nonpipelined mode. Control circuits execute a UEV handler microcode routine which reads the contents of syndrome registers included on each board containing the UEV indicator states in addition to error signals into register file working locations. The UEV signal is then cleared enabling the BIU to resume communications with the sytem bus.

Recovery method and apparatus for a pipelined processing unit of a

A data processing system includes cache memories for storing instructions and operands. An execution unit stores instructions in an instruction FIFO, operands in a data FIFO and offsets in an offset FIFO. Offsets indicate the location of operands relative to a memory word boundary. Instructions read from the instruction FIFO are applied to a control store subsystem which reads out a firmware word. Specified firmware bits condition multiplexers in the data path to align the operands on the fly during the execution of the instruction.

Apparatus for generating first and second selection signals for aligning

In a data processing system using a virtual memory adressing scheme, certain software instructions call for the virtual address to be stored in a base register. The virtual address stored in the base register is incremented or decremented during the read out cycle of the previous operand to address the next operand. If the operand is not in physical memory, then the contents of the base register is restored to its original value.

Method for reexecuting instruction by altering high bits of instruction

In a data processing system in which each of the data processing units is implemented using pipeline techniques and has a cache memory unit employing a store through strategy, the time required to prepare a write instruction operand address can be substantially shorter than the time required by the execution unit to prepare the associated write instruction operand. In order to utilize the time difference, apparatus is included in the execution cache unit for storing the write instruction operand address during the preparation of the associated write instruction operand. After storing the write instruction operand address, a next address is entered in an input register of the execution cache unit. When the newly entered address is associated with a read instruction, does not conflict with the write instruction operand address, and produces a 'hit' signal when applied to the execution cache unit tag directory, the read instruction is processed by the execution unit. When a second write instruction operand address is entered in the input register, the read instruction operand address conflicts with the stored write instruction operand address or the read instruction operand address results in a 'miss' when applied to the execution cache tag directory unit, the address is stored in the input register until the write instruction operand has been determined and the associated write instruction has been procossed by the execution cache unit.

Apparatus and method for simultaneous execution of a write instruction

During the execution of an instruction by an execution unit, the instruction is stored in an instruction register, the operand including its ring number is stored in a data register and the ring number developed by the Virtual Memory Management Unit is stored in a ring effective register. The instruction addresses a control store which stores a firmware word in a control store register. A firmware field is decoded to generate a plurality of ring control signals. The ring numbers from the data and ring effective registers are compared and an effective ring number is generated. Depending on the states of the secure process signal, the ring control signals and the relative value of the ring numbers, the effective ring number is binary 00 or the larger ring number.

Ring reduction logic using parallel determination of ring numbers in a

A separate register file memory is included in at least two units of a pipelined processor which are located on separate integrated circuit chips. The register file memories of the units are interconnected so as to share certain input data register stages to enable updating to take place within a minimum of time. Each unit has a microprogrammed control unit which automatically provides update commands during the unit's cycles of operation. The signals from each microprogrammed control unit are applied to both register file memories enabling both memories to be updated during successive cycles of operation and thereby function collectively as one unit. This ensures that both units have access to the same most recently updated user visible information enabling both units to complete the execution of different instructions entering pipeline.

Mechanism for automatically updating multiple unit register file

A data processing system includes the functionality of a commercial instruction processor, a scientific instruction processor and a basic instruction processor integrated into a single semiconductor logic element.

Multiprocessors on a single semiconductor chip

A firmware controlled microprocessor plugged into a printed circuit board received firmware signals from a control store mounted on the printed circuit board. The number of pins required for transferring firmware signals is reduced by time sharing pins with firmware signal required for the full cycle of operation and firmware signal required only during the second half of the cycle of operation.

Means for transferring firmware signals between a control store and a

In a data processing system using a virtual memory addressing scheme, certain software instructions call for the virtual address to be stored in a base register. The virtual address stored in the base register is incremented or decremented during the read out cycle of the previous operand to address the next operand. If the operand is not in physical memory, then the contents of the base register is restored to its original value. This invention minimizes the amount of logic required to back out of a software instruction after execution has begun and is faster than checking if all resources are present before any state change is made during the execution of a software instruction.

Method and apparatus for backing out of a software instruction after

Apparatus that provides interrupt operation in a central processor based system wherein internal subsystems are operated via addresses generated by a next address generator in the processor and sent to control stores associated with each subsystem to thereby read out firmware instructions which are used by a controller in each subsystem to control the operations of same. When a special condition is detected in ones of the subsystems a trap signal is sent to the next address generator which responds by generating a microinstruction address to the subsystem that generated the trap signal. The subsystem responds to the microinstruction to read out a register, the contents of which indicate the status of processing in the subsystem including the special condition. The register contents are forwarded to the processor which tests same to determine the nature of the special condition and calls a microprogram the microinstructions of which are applied to the control store of the subsystem that generated the trap signal. The subsystem responds to the microinstructions to clear the special condition. Certain subsystems may alternately send a special condition indicating signal directly to the next address generator, rather than a trap signal, and responsive thereto the next address generator calls the required microprogram to be applied to the control store of the subsystem that generated the indicating signal.

Distributed control store word architecture

A method and apparatus for a microinstruction controlled unit to recover from a read error in reading microinstructions from a control store. The method provides for the overlapping of the execution of a current microinstruction while the next microinstruction is being addressed and read from the control store. Execution of the current microinstruction is begun before it is known whether or not it was read without error. The apparatus provides for aborting the execution of the current microinstruction with the read error and the next microinstruction. During the aborted execution of the next microinstruction, the current microinstruction is reread from the control store and then executed while the next microinstruction is being reread. The execution of microinstructions is aborted in a manner that does not alter the state of the microinstruction controlled unit beyond the point that would inhibit the re-execution of the aborted microinstructions.

Control store memory read error resiliency method and apparatus

In a data processing system having a plurality of units coupled for the transfer of information therebetween over a common electrical bus or for transferring information via a communication channel to other data processing systems during asynchronously generated information bus transfer cycles, an apparatus exists for reformatting data for transfer over the common electrical bus or via the communication channel. The apparatus is comprised of an eight-way multiplexer responsive to control bits for selecting one of eight different formats during write or read operations to or from a memory. Additionally, during read operations, the apparatus transfers a return address from a unit requesting information to the address bus so that data read from memory may be transferred to the requesting device. The formatting control bit is similarly reformatted from the data bus to the address bus bit.

Automatic data steering and data formatting mechanism

A communications controller of a data processing system uses a microprocessor to control communication operations. Apparatus in the controller stretches the microprocessor clock cycle signals for selected operations to allow the microprocessor speed to match the speed of the logic performing the selected operation. The apparatus includes a counter which is freerunning for the stretched cycle and reset on a predetermined cycle for the 'no stretch' cycle. A decoder coupled to the counter conditions logic gates to generate the microprocessor clock cycle signals.

Microprocessor controlled communications controller having a stretched

A data processing system operating in a bit oriented protocol (BOP) mode of operation senses a transmit underrun; that is, the subsystem is not receiving data from a microprocessor fast enough to maintain the synchronous transmission over the communication line. Apparatus senses the transmit underrun state and generates an abort sequence of bits containing from 8 to 13 successive binary ONE bits.

Growing community of inventors

Nashua, NH, United States of America

Richard P Kelly