Abstract: Circuits, systems and methods for operating a processor to process a plurality of sequentially arranged instructions. The method includes various steps, such as receiving (54) into a processor pipeline an instruction from the plurality of sequentially arranged instructions. Next, determines (56) whether the received instruction comprises a memory access instruction. A memory access instruction is operable to access memory information of a specifiable size. In response to determining that the received instruction comprises a memory access instruction, the method generates (58) at least one micro-operation code corresponding to the memory access instruction and it also sets (60) a tag to the at least one micro-operation code, where the set tag requests a subsequent evaluation of the specifiable size. After the tag is set, the method later detects (64, 72) the set tag and, in response to the set tag, retrieves (66, 74) a current value of the specifiable size.

Abstract: A method and system for speculatively sourcing cache memory data within a multiprocessor data-processing system is disclosed. In accordance with the method and system of the present invention, the data-processing system has multiple processing units, each of the processing units including at least one cache memory. In response to a request for data by a first processing unit within the data-processing system, an intervention response is issued from a second processing unit within the data-processing system that contains the requested data. The requested data is then read from a cache memory within the second processing unit before a combined response from all the processing units returns to the second processing unit.

Abstract: Apparatus and methods for allocating shared memory utilizing linked lists are provided which are particularly useful in telecommunications applications such as ATM. A management RAM contained within a VLSI circuit is provided for controlling the flow of data into and out of a shared memory (data RAM), and stores information regarding a number of link lists and a free link list in the shared memory, and a block pointer to unused RAM locations. A head pointer, tail pointer, block counter and empty flag are stored for each data link list. The head and tail pointers each include a block pointer and a position counter. The block counter contains the number of blocks used in the particular queue. The empty flag indicates whether the queue is empty. The free link list includes a head pointer, a block counter, and an empty flag. Each memory page of the shared data RAM receiving the incoming data includes locations for storing data.

Abstract: A data processor supports the use of multiple memory models by computer programs. At a device external to a data processor, such as a memory controller, memory transactions requests are received from the data processor. Each memory transaction request has associated therewith a memory model selected from a predefined plurality of memory models. In a preferred embodiment, the predefined memory models supported are SSO (strong sequential order), TSO (total store order), PSO (partial store order) and RMO (relaxed memory order). Data representing pending memory transactions are stored in one or more pending transaction buffers and a pending transaction status array. The pending transaction status data includes memory transaction order data that indicates which of the pending memory transactions can be performed before other ones of the pending memory transactions.

Abstract: A conventional network file server has a file system that permits file attributes and file data to be written in any order. The conventional network file server may also support an asynchronous write protocol, in which file attributes and file data need not be written to disk storage until a client sends a commit request. This asynchronous write protocol has a data security problem if the attributes are written before the data and the server crashes before completing the writing of the data to disk storage. This security problem is solved by adding a file system cache and following a protocol that writes the attributes to storage after writing the data to storage. For example, the attributes and data are stored in the file system cache and are not written down to storage until receipt of a commit request. When the commit request is received, the data is sent first from the file system cache to storage. Then the attributes are sent from the file system cache to storage.

Abstract: A method for connecting a DRAM module to memory portions of a main processor and memory management PCB board assembly (MPMA PBA) included in a main processor hardware (MPH) block and adapted to perform a higher-order control in a full electronic exchange. In accordance with this method, the DRAM, which is of the zigzag-in-line package type, is replaced by that of the module type so that its parity DRAM area is integrated with memory areas included in the DRAM, thereby enabling waste memory portions of the parity DRAM area to be efficiently managed. A common RAS control signal is divided into a plurality of signals respectively adapted to be used as control signals for defining respective memory areas of the DRAM along with other control signals, namely, CAS and WE control signals. A basic address for the DRAM is determined to obtain an easy memory expansion.

Abstract: A method for caching virtual memory paging requests and disk input/output requests utilizes a portion of the video memory as a location for paged memory as well as an alternative location for a disk cache system. The portion of video memory employed is off screen memory (OSM), access to which is controlled to make OSM available for paging or caching requirements. System operators may be monitored on a continuing basis to provide for a dynamic allocation of OSM.

Abstract: A computer having backup capabilities for backing up data on a host storage disk of the computer to a remote archive repository, the computer including a backup application that determines when data to be backed up should be copied to the remote archive repository and makes read requests to make a copy of data on the host storage disk and makes write requests to store the copy at the remote archive repository, the computer also including a backup system that receives the write requests and the copy of data and has a remote procedure call interface for transmitting the data to the remote archive repository.

Abstract: A DRAM is provided that can carry out data reads or writes in a constant and short access time regardless of the timing with which the reads or writes, or refreshing are executed. When requests for reads from or writes to burst data are continuously input, row decoding (RD) and column decoding (CD) by a row decoder 42 and a column decoder 52, an array access (AR) and precharging (PR) by a data line driver 24, a bit switch 26, and a sense amplifier 28, and data transfer (TR) by a write buffer 52 and a read buffer 54 are executed in parallel in a pipelined manner. When the time has come to refresh a DRAM array 22, a refresh address held in a refresh controller 40 is output while the burst data is being transferred, and a series of refreshing operations comprising (RD), (AR), and (PR) is performed.

Abstract: A method and system in which X-bit packets of bits (where X is an integer) are encoded to generate X-bit packets of encoded bits for writing to erased cells of a flash memory array, where less power is consumed to write a bit having a first value to an erased cell than to write a bit having a second value to the cell. Preferably, a count signal is generated for each packet of raw bits indicating the number of bits of the packet having the first (or second) value, the count signal is processed to generate a control signal which determines an encoding for the packet, and the raw bits of the packet are encoded according to a scheme determined by the control signal. In some embodiments, each erased cell is indicative of the binary value "1", the count signal is compared to a reference value (indicative of X/2) to generate a control signal determining whether the packet should undergo polarity inversion, and the packet is inverted (or not inverted) depending on the value of the control signal.

Abstract: The present invention provides a look ahead architecture to satisfy the retransmit recovery time constraints in a mark and retransmit system while allowing a full bitline precharge. A number of sense amplifiers are provided in the look ahead architecture that may be equipped with a "shadow latch" to store the read data when the mark pointer is asserted. As a result, the data to be retransmitted will be retrieved from the shadow latches when the retransmit is asserted, allowing a full precharge cycle before reading from the memory array.

Abstract: A memory including a plurality of input/output terminals 220 for exchanging data bits during a data access cycle and receiving command and control bits during a command and control cycle. The memory further includes an array of memory cells 201, a data input/output circuitry for transferring data between the input/output terminals and the array of memory cells during the data access cycle, and control circuitry for controlling operations of the memory in response to command and control bits received at the input/output terminals during the command and control cycle.

Abstract: A multi-processor computer system comprising a data storage device, a memory controller, and a plurality of processors. The data storage device has a plurality of memory lines, each memory line having a portion for alternatively storing data or, a set of GONE codes, a count value, and a processor identification code value. A memory controller coupled to the data storage alternatively stores and retrieves data or the GONE code, the count field value and the processor identification code value. At least one of the processors includes a cache memory and a cache memory controller. The cache memory controller compares a GONE code associated with the requested memory line with the contents of the requested memory line, and requests the contents of the requested memory line from a second of the processors in response to the comparison.

Abstract: In a computer system implementing state transitions that change logically and atomically at an address packet independently of a response, the coherence domain is extended among distributed memory. As such, memory line ownership transfers upon request, and not upon requestor receipt of data. Requestor receipt of data is rapidly implemented by providing a ReadToShareFork transaction that simultaneously causes a write-type operation that updates invalid data from a requested memory address, and provides the updated data to the requesting device. More specifically, when writing valid data to memory, the ReadToShare Fork transaction simultaneously causes reissuance of the originally requested transaction using the same memory address and ID information. The requesting device upon recognizing its transaction ID on the bus system will pull the now valid data from the desired memory location.

Abstract: A method for storing and transferring wave table audio samples from system memory to a cache unit. The method creates a linked-list of pages in system memory for storing the audio sample. The linked-list is actually a pointer list indicating the locations in system memory where the audio samples are stored. A Digital Signal Processor (DSP) is able to translate the starting address of the pointer list to retrieve a requested audio sample from the system memory. The requested audio sample is then transferred to the cache unit where the DSP is able to retrieve audio samples in a linear fashion at a rate much faster than individually fetching the required portions of the audio sample from the main memory of the system.

Abstract: A method for configuring memory in a computer system. The method calls for determining the maximum configurable size of installed memory modules for each of a plurality of interleave options. The computer system is configured to utilize the largest of the maximum configurable sizes of interleave options. The maximum configurable sizes of interleave options are further compared to determine if the memory modules are installed in a manner providing the largest amount of configurable memory. If not, an error message is sent to the user.

Abstract: A non-volatile caching system and a method for implement such a system is disclosed. The system is particularly applicable to rotating magnetic media such as hard disk drives. The system retains data even in the event of system shut-down and re-boot. The system is capable of rapidly caching data from large, randomly accessed files, such as databases, in a space-efficient manner. The cached data can be stored in nearly any standard or non-standard format on the magnetic media. A conversion routine converts CD-ROM file names or network file names to local hard disk drive file names and back. A mini-database is created for each cached file on the hard disk drive. The mini-data base maps randomly-accessed blocks of data within the cached file on the local hard disk drive.

Abstract: A method and apparatus for use in computer systems utilizes a memory chip employing multiple distributed SRAM caches directly linked to a single DRAM main memory block. Each cache is directly linked to a different bus. Each chip further contains a partially distributed arbitration and control circuit for implementing cache policy and arbitrating memory refresh cycles.

Abstract: A disk controller for headerless disk drive system contains a decoder which decodes control words from a data buffer and counts pulses of a byte clock to identify the boundaries of requested data sectors in data frames. To select the control words which correspond to the position of a read/write head relative to media containing the data, the disk controller also contains a programmable alignment processor and an EOS counter that increments in response to EOS pulses generate as the read/write head passes servo sectors. The EOS counter indicates the position of the read/write head by indicating a data frame over which the head is positioned. The alignment processor scans the control words to locate the control word corresponding the data frame indicated by the EOS counter. During the scanning, the alignment processor also aligns logical and physical data sector counters to the position of the read/write head.

Abstract: Memory space in the lower-level cache (LLC) of a computer system is allocated in cache-line sized units, while memory space in the higher-level cache (HLC) of the computer system is allocated in page sized units; with each page including two or more cache lines. Accordingly, during the execution of a program, cache-line-sized components of a page-sized block of data are incrementally stored in the cache lines of the LLCs. Subsequently, the system determines that it is time to review the allocation of cache resources, i.e., between the LLC and the HLC. The review trigger may be external to the processor, e.g., a timer interrupting the processor on a periodic basis. Alternatively, the review trigger may be from the LLC or the HLC, e.g., when the LLC is full, or when usage of the HLC drops below a certain percentage.