Abstract: An apparatus for efficiently detecting errors in a system having a plurality of memory devices. The present invention uses a single parity bit configuration to detect common data errors caused by faulty memory devices including multiple data errors within one memory device. This is accomplished by effectively turning a multiple bit error detection situation into a single bit error detection situation. Thus, instead of allocating a contiguous block of bits to the same memory unit, the present invention allocates bits across all memory units in a round-robin fashion. The parity domains are defined such that multiple errors within one SRAM can be detected despite only using a single bit parity configuration.

Abstract: A system and method for updating partial blocks of file data stored in a non-volatile storage within a file cache system connected to a host computer system. A first buffer and a last buffer receive from the non-volatile storage the existing portions of the blocks that are to be retained. A write buffer receives new data of a size not equal to an integral multiple of a block from a host computer system. The new data is merged under hardware control with the existing portions contained in the first buffer and the last buffer, thereby updating the cached file.

Abstract: A method and apparatus for executing conditional branch instructions by a data processing system containing a programmable microprocessor and multiple VLSI gate arrays connected by a bi-directional bus, whereby a branch condition is obtained from a storage location resident on a VLSI gate array executing asynchronous and external to the microprocessor. The branch condition is fetched and evaluated in parallel with the fetching of the branch target address and the incrementing of the program counter. The microprocessor changes instruction sequence control depending on the results of the branch condition evaluation.

Abstract: A microsequencer bus controller system provides a flexible and efficient mechanism for controlling multiple gate arrays called stations embedded within a larger computer system. A control store memory, loaded at system initialization time, holds fixed-length instructions simultaneously executed by dual reduced instruction set (RISC) microprocessors which interface with the multiple stations over a bi-directional bus. The master microprocessor compares the result of the processing of each instruction with the slave microprocessor's result to detect any differences, thereby minimizing error latency. Master and slave microprocessors each control half of the stations on the bus. Data widths of 32-bit and 36-bit words are supported by the microprocessors, bus, and stations.

Abstract: A bus architecture includes address lines, data lines, and control signals to allow a processor to communicate with a VLSI gate array. The address lines are interpreted by the VLSI gate array to select either multi-bit registers or single bit designators resident on the VLSI gate array depending on which control signal is received from the processor. Dual address decode logic on the VLSI gate array senses control signals indicating a request to read from a register, write to a register, and set, clear, or test a designator, and decodes the address received to select the appropriate storage location for the requested function.

Abstract: A bus architecture and associated circuitry for providing communication between processors and multiple gate arrays whereby the size of the data being transferred may be either full words of 32-bits or 36-bits per word, or half words of 16-bits or 18-bits per word. Parity generation logic operates on the data to be sent over the bus to generate a parity value from the correct data bits depending on the selected data word size. Parity checking logic operates on the data received from the bus to check the parity of the correct data bits depending on the selected data word size.

Abstract: A method and apparatus for detecting stuck faults in a signal line used to communicate a branch condition for executing conditional branch instructions by a data processing system containing a programmable microprocessor and multiple VLSI gate arrays connected by a bi-directional bus, whereby the branch condition is obtained from a storage location resident on a VLSI gate array executing asynchronous and external to the microprocessor. The branch condition is fetched and evaluated in parallel with the fetching of the branch target address and the incrementing of the program counter. The microprocessor changes instruction sequence control depending on the results of the branch condition evaluation.

Abstract: A system and method for detecting errors during the storage and retrieval of file information between a file cache system and a host computer system utilizes a block check sequence key as redundant data included in each block of file data transferred. The block check sequence key is generated by key generation logic and accompanies each block of file data stored in the file cache system by the host computer system. The block check sequence key is a compressed representation of the data within the selected block, as well as unique file and block identification information supplied by the requester of the write operation. When the block is retrieved from the file cache system, the system generates a new block check sequence key based on the data within the retrieved block and the unique file and block identification information supplied by the requester of the read operation. Validation logic ensures that if the retrieved key and the newly generated key does not match, an error signal is activated.

Abstract: A method and system for executing shift, mask, and merge operations on two operands specified by one instruction contains two registers holding operand data and separate shift, mask, and merge logic. A programmer-defined set of mask and merge indicators controls the mask and merge operations. Each mask and merge indicator is represented as a single bit but controls a pair of bits in an operand. If the first operand is selected by the programmer, it is shifted and then masked. The result of the shift and mask operations is merged with the second operand. If the second operand is selected, it is shifted and masked, and the result is merged with the first operand. Final results are stored for processing by subsequent instructions.

Abstract: A system and method for executing conditional branch instructions by a processor using dynamic branch address tables containing branch target addresses. The branch target addresses are selected by the result of a computation of fields in the branch instruction and an index generated during the execution of previous instructions. The relevant fields include the base address of a branch address table and a mask value. The contents, size, and location of the branch address tables in a random-access-memory local store may be changed during run-time by program control.

Abstract: An address verification system for providing address error detection whether the error originates at the address generation circuitry, the address transmission path, or the address receiving circuitry. Multiple address generation circuits which simultaneously generate equivalent addresses each have associated parity generation circuits to provide parity bits for its associated address. Monitoring for unequal parity bits generated by the multiple parity generation circuits allows detection of address generation errors. Predetermined address parity bits for each potential address to be sent to the address-receiving circuitry are stored at the address-receiving circuitry to be compared to the parity bits issued by the multiple parity generation circuits.

Abstract: A counter system having associated counter error detection circuitry that utilizes the current parity, the previous parity, and a predicted parity for evaluating counter operation is described. In successive count cycles, a predicted parity is utilized, during the next subsequent count cycle is stored in flip-flop as the current parity, and in the next subsequent count cycle is stored a second flip-flop as a previous parity. Circuit are described for performing parity check and parity prediction functions. The previous parity, current parity and predicted parity will not be alike for any binary counter that operates properly. Circuity is described that holds and compares the parity of the Count, the current parity, and the previous parity, during each counter advance cycle and to provide an error signal when the counter is detected to be stuck.

Abstract: A bus control and error detection system is provided for a bus system in which data and address signals are transferred between a microsequencer and a number of operational stations which are coupled to the bus. Tri-state drivers are employed in the microsequencer and in the stations which are constructed such that two of the three states of these tri-state drivers are utilized to provide the two states of binary logic operation, and the third state is a high impedance state that protects the components that are coupled to the bus during predefined abort condition which are detected in the system. An abort detection circuit is included in each of the operational stations which is coupled to receive control signals from the microsequencer and which is constructed to emit an ABORT signal output to the microsequencer when the control signals indicate that an abort condition has occurred for the associated operational station.

Abstract: A system for apportioning serially supplied data among eight contending memory banks tends to equalize usage among the banks despite their arrangement in a predetermined, sequential priority. Each bank has a data hold register, an OR logic gate to generate a request signal whenever its register contains data, and a negative AND gate for enabling the bank for clearing data from its register. All except the lowest priority bank further include a blocking latch and an enabling NOR gate. Each blocking latch is set when its associated bank is enabled, and then inhibits its associated AND gate and each higher priority AND gate, while enabling each lower priority NOR gate. Each enabled NOR gate provides an enabling signal to all lower priority AND gates. When the lowest priority AND gate is enabled by the NOR gates and its request signal, all blocking latches are cleared. The banks thus are utilized in a sequence that is maintained even if one or more banks are bypassed on occasion.

Abstract: A novel memory access system is provided for simultaneously processing request for access to a plurality of memory banks. A plurality of input-output ports are coupled to a read bus and to a write bus which are in turn coupled to the memory banks to be accessed by read and write commands initiated by processors coupled to the I/O ports. Pipeline control means receive the request for access functions from the processors and are operable to resolve conflict between plural request. The pipeline control means sequentially raise either write or read request on control and address buses and generate time slot windows during which subsequent write or read data transfer operations will occur so that data being pipelined on the write and read buses is being simultaneously accessed.

Abstract: A system for apportioning serially supplied data among eight contending memory banks tends to equalize usage among the banks despite their arrangement in a predetermined, sequential priority. Each bank has a data hold register, an OR logic gate to generate a request signal whenever its register contains data, and a negative AND gate for enabling the bank for clearing data from its register. All except the lowest priority bank further include a blocking latch and an enabling NOR gate. Each blocking latch is set when its associated bank is enabled, and then inhibits its associated AND gate and each higher priority AND gate, while enabling each lower priority NOR gate. Each enabled NOR gate provides an enabling signal to all lower priority AND gates. When the lowest priority AND gate is enabled by the NOR gates and its request signal, all blocking latches are cleared. The banks thus are utilized in a sequence that is maintained even if one or more banks are bypassed on occasion.

Abstract: Logic checking circuits are provided for verifying whether or not the data bus enable logic circuits are operating properly in response to operational commands to transmit or to NOT transmit data. The transmit latches in the bus interface logic circuits are continuously monitored to determine if they are set or NOT set in a position to enable transmission of data or NOT to enable transmission of data to a bus. Transmit gating circuit means are couple to the output of said transmit latches for determining if all of the transmit latches are in the same state and are in the state ordered by the central controller, and for determining whether the state ordered by the central controller occurs in the exact time period during which the command to transmit should be executed.

Abstract: The logic cards for a main storage unit or computer logic which receive request operations for access to portions of the memory or logic are divided into banks or elements. When a request operation attempts to access one of the elements a return busy signal is raised from that element. The present invention structure generates a predicted busy signal which occurs during the same time the return busy signal should be activated or operable. The return busy signal and predict busy signal are compared in novel circuitry to verify that the element performing the operaton is in fact performing an operation during the predetermined time slot allowed for performance of the requested operation. Fault signals for bank invalidation are stored in internal check trap circuitry for future reference when the requestor raises a subsequent request operation.

Abstract: A dynamic random access memory (DRAM) memory refreshing scheme utilizes at least two separate refresh channels. Each of the channels consists of a pair of identical counters which are coupled through two different types of timing chains. One of the timing chains is associated with one of the counters and generates a refresh request signal, while the other timing channel generates a refresh error signal. As long as the refresh error signal matches the refresh request signal, no error is present, and a validated refresh request signal will be generated from that timing channel and supplied to an OR gate to refresh all of the memory banks for the memory.

Abstract: A dual priority hold register enables the transfer of data to memory ports having serial priority in accordance with two stages of priority. First, all latches of a high priority sector of the register are cleared. Then, the highest priority latch of the low priority sector of the register is cleared, while the latches of the higher priority register are loaded with further data. Following clearance of the low priority latch, all latches of the higher priority register are cleared once again, followed by clearance of the next highest priority latch of the lower priority register sector while the higher priority register is loaded once again. The sequence is repeated until both the higher and lower priority sectors of the register are clear.