Memory arrays are divided into banks which can be operated independent from each other. Read data storing registers and write data storing registers operating independent from each other are provided for the banks. The memory array is divided into a plurality of small array blocks, local IO lines are arranged corresponding to each array block, and the local IO lines are connected to global IO lines. The global IO lines are connected to preamplifier groups and to write buffer groups. By control signal generating circuits and by a register control circuit, inhibition of writing of a desired bit only during successive writing operation can be done, data can be collectively written to the selected memory cells when the final data is input if the data writing should be stopped before reaching the wrap length in successive writing, and the timing for activating the memory array when the write cycle should be repeatedly carried out can be delayed. A synchronous semiconductor memory device having small chip area, high speed of operation, low power consumption and multiple functions is provided.

Synchronous semiconductor memory device

Methods and apparatus are provided for detecting and correcting various data errors that may arise in a mass data storage apparatus comprising a set of physical mass storage devices operating as one or more larger logical mass storage devices. A method and apparatus is provided for detecting and reconstructing incorrectly routed data. A method and apparatus is also provided for detecting when one or more physical devices fails to write a block of data, and for reconstructing lost data.

Data corrections applicable to redundant arrays of independent disks

A method and apparatus are provided for detecting and correcting various data errors that may arise in a mass data storage apparatus comprising a set of physical mass storage devices operating as one or more larger logical mass storage devices. More particularly, there is provided a method and apparatus for determining, on restoration of power to a device set, whether or not a write operation was interrupted when power was removed, and for reconstructing any data that may be inconsistent because of the removal of power.

Non-volatile memory storage of write operation identifier in data storage device

A processing system allows for out of order instruction execution and includes at least one load/store unit for loading instructions to a register for processing by a fixed point unit, floating point unit, or the like, and store the results to memory. A load queue maintains the addresses and program numbers of the load instructions. During execution the address of the store instruction is compared to the address in the load queue of previously executed load instructions. A program counter compares the program number of the store instruction with the program number of the load instruction in the load queue. If the addresses are different, then no impermissible out of order situation exists between the load and store instructions being compared, because the data is not at the same address. If the address is the same, and the store program number is greater than the load program number, then the instructions have been executed in order (the load correctly preceded the store) and no problem exists. However, if the addresses are the same and the load instruction has been incorrectly reordered to precede the store instruction, then a reordering conflict exists and the load instructions must be re-executed.

Out of order instruction load and store comparison

A multiprocessor system includes a number of sub-processor systems, each substantially identically constructed, and each comprising a central processing unit (CPU), and at least one I/O device, interconnected by routing apparatus that also interconnects the sub-processor systems. A CPU of any one of the sub-processor systems may communicate, through the routing elements, with any I/O device of the system, or with any CPU of the system. Communications between I/O devices and CPUs is by packetized messages. Interrupts from I/O devices are communicated from the I/O devices to the CPUs (or from one CPU to another CPU) as message packets. CPUs and I/O devices may write to, or read from, memory of a CPU of the system. Memory protection is provided by an access validation method maintained by each CPU in which CPUs and/or I/O devices are provided with a validation to read/write memory of that CPU, without which memory access is denied.

Self-checked, lock step processor pairs

A high performance pipelined virtual first-in first-out stack structure has a data stack portion and a split control stack portion. The stack structure is intended for use in a pipelined high performance storage unit that can pipeline up to R input requests without having received an acknowledge that a request has been honored. The data stack incorporates R+1 data stack registers to provide over-write protection to ensure that at least R data stack registers are protected from over-write. The split control stack utilizes even address and odd address stack registers. Memory bank request signals are stored sequentially and alternately between the even address and odd address stack registers. An even address read pointer and an odd address read pointer under control of a read pointer control circuit alternates the selection for read out sequentially between the even address and odd address stack registers that decoding of the memory bank request signals for the next reference can be interleaved with completion of the decoding and prioritization of the current stack register. Advancement of stack register addresses at which writing will take place is under control of a request signal. Control of the read pointers for the data stack and split control stack are responsive to bank acknowledge signals received by the read pointer control circuits.

Pipelined split stack with high performance interleaved decode

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.

Memory access system for pipelined data paths to and from storage

Apparatus for detecting and isolating the occurrence of faults in a digital electronic system so as to reduce the mean-time-to-repair. Associated with the logic circuitry to be monitored is a fault indicator which produces a fault signal when a malfunction occurs. Fault capture circuitry is arranged in a hierarchical manner and provides a group fault output signal when one of the fault indicators generates a fault signal. A programmable controller is provided which receives the group fault signal as an interrupt and which then responds by transferring registered fault event signals to a dynamic string register, rearming the error detection used to trap a fault signal and logging the fault location in a memory for later readout by a maintenance processor or the like. The dynamic string allows communications to take place using a scan/set protocol.

Fault capture/fault injection system

A virtual stack structure utilizing Write Pointers and Read Pointers for providing pipelined data words on a first-in first-out basis to a memory structure is described. The virtual stack structure incorporates a plurality of Stack Registers each having an unique Write Tag and Read Tag associated therewith. The Write Tags are utilized to through-check the decoding of the Write Pointer and to issue Write Tag Error signals when it is determined that the Write Pointer has been improperly decoded. Circuitry is provided for checking the appropriate loading of the Write Pointer in the stack, and issuing an error signal when improper loading is sensed. Each of the Stack Registers has an unique Read Tag associated therewith that is utilized to through-check the decoding of the Read Pointer and to issue Read Tag Error signals when improper decoding is detected. Comparison circuitry is utilized to determine that the Read Pointer has been properly loaded in the stack structure, and to issue an error signal when improper loading is sensed.

Pipelined data stack with access through-checking

An error correcting check of a memory system is provided when a memory in which the Dynamic Random Access Memory (DRAM) is of the type which has input lines that are directly coupled to its output lines. Utilizing this type of DRAM, the memory system employs controls, input, output and read circuitry to read bits out of the memory via the output circuitry and write circuitry to write bits into the memory via the input circuitry. An error checking and correction circuit is coupled to the output means which includes a check bit generator and a syndrome generator, and a control means energizes the error checking and correcting means during the write cycle, as well as the read cycle, so that the errors are detected during the write cycle as well as the read cycle. In this manner, errors which occur in circuitry other than the memory, which includes the memory driving and reading logic and also the check bit generator logic translators and syndrome generators, may be separately detected from memory errors.

James H. Scheuneman

Papers

Pipelined split stack with high performance interleaved decode

Memory access system for pipelined data paths to and from storage

Fault capture/fault injection system

Pipelined data stack with access through-checking

Error correction check during write cycles