摘要:
A method of integrating a hybrid architecture in a set associative cache having a first type of memory structure for one or more ways in each congruence class, and a second type of memory structure for the remaining ways of the congruence class, includes determining whether a memory access request results in a cache hit or a cache miss; in the event of a cache miss, determining whether LRU way of the first type memory structure is also the LRU way of the entire congruence class, and if not, then copying the contents of the LRU way of the first type memory structure into the LRU way of the entire congruence class, and filling the LRU way of the first type memory structure with a new cache line in the event of a cache miss; and updating LRU bits, depending upon the results of the memory access request.
摘要:
The invention provides a method, system, and program product for managing a connection. In particular, the invention manages connection information in memory based on an expected usage of the corresponding connection. Connection information can be stored in faster memory, such as cache memory, when the connection is expected to have numerous additional messages. Similarly, the connection information for a connection not expected to have many additional messages can be swapped out of the cache memory and stored in relatively slower memory. As a result, the connection information that is more frequently used is more likely to be available in a faster memory.
摘要:
A method and apparatus are provided for managing cache allocation for a plurality of data types in a unified cache having dynamically allocable lines for first type data and for second type data. Cache allocation is managed by counting misses to first type data and misses to second type data in the unified cache, and by determining when a difference between a number of first type data misses and a number of second type data misses crosses a preselected threshold. A replacement algorithm of the unified cache then is adjusted in response to the detected crossing of the preselected threshold, the adjusting step including increasing a replacement priority of the first type data lines in the cache. The replacement algorithm preferably is an LRU algorithm wherein the adjusting step includes incrementing an age indication of the first type data lines. Hardware for implementing the inventive cache allocation management method comprises a miss counter configured to increment its count in response to a miss to first type data signal on a first counter input and to output a first logic state on a first counter output when the counter's count exceeds a first predetermined count. A priority adjustment circuit coupled to the first counter output increases the replacement priority of the first type data relative to the replacement priority of the second type data in response to the first logic state output by the miss counter.
摘要:
A memory card adapter and method is provided which can add features or provide functions to a computer system's memory modules without having to replace and discard existing memory modules. An adapter is provided which has electrical contacts that are capable of being plugged into a memory module receiving socket of a motherboard and a memory module receiving socket capable of receiving and retaining a memory module such as a SIMM. The adapter has logic, circuitry and/or memory chips to add new function to the existing memory module and also has all information and hardware needed for proper interface with the motherboard of the computer system. The present invention can add a variety of function such as parity, error correction code and error correction code on SIMM as well as convert signals which form from the system for use on the SIMM which signals in the form generate by the computer are not compatible with the SIMM.
摘要:
The invention provides a method, system, and program product for managing a connection. In particular, the invention manages connection information in memory based on an expected usage of the corresponding connection. Connection information can be stored in faster memory, such as cache memory, when the connection is expected to have numerous additional messages. Similarly, the connection information for a connection not expected to have many additional messages can be swapped out of the cache memory and stored in relatively slower memory. As a result, the connection information that is more frequently used is more likely to be available in a faster memory.
摘要:
A multi-layer silicon stack architecture includes one or more processing layers including one or more computing elements; one or more networking layers disposed between the processing layers, the network layer includes one or more networking elements, wherein each computing element includes a plurality of network connections to adjacently disposed networking elements.
摘要:
An apparatus and method in which ECC bus protection capability can be generated on a memory card in conjunction with a computer system with a built-in ECC capability to reduce data transmission errors. Data generated by the system is transmitted to the card and stored in DRAMs. On a read cycle, the card generates a set of checkbits which are sent to the system on a checkbit bus. The system generates a set of checkbits from the data read from the memory card and compares these checkbits with those received from the memory card. A mismatch indicates transmission error on the bus(s) during a read cycle. Any correctable error is corrected. Data is invalidated if an uncorrectable error is detected. In another embodiment checkbits generated by the system during a write cycle are transmitted to the card an checkbits are generated by the card. These two sets of checkbits are compared and if there is a mismatch data is either flagged as bad or corrected. Furthermore, in one embodiment, if the memory card does "not know" in advance the type of ECC or H-matrix code resident in the computer system, the card has the capability to determine what H-matrix code is resident and set up a corresponding H-matrix.
摘要:
A high speed Random Access Memory (RAM) array device includes several logical banks, each of which can be uniquely addressed. Each of these logical banks contains a unique memory array segment and associated page register, the latter serving as a temporary storage location during high-speed page hit operations. To reduce latency during an initial page hit, further array optimization is realized by segmenting each logical bank into two segments with one, smaller segment, comprising a faster random access memory (FRAM) for storing initial data in a data stream. A high speed page register connects the FRAM directly to a multiplexer/demultiplexer connected to the device I/O ports bypassing an internal bus protocol such that the initial data can be transferred between the FRAM and the I/O ports faster thereby improving page-hit latency. Hence, segmenting the logical banks to include only a small high speed segment results in a performance gain approaching what could be achieved by implementing the entire memory device with a high speed FRAM, but at much lower cost.
摘要:
A circuit board for receiving different chip modules at each chip module site has a site for receiving a chip module having electrical connectors thereon and a first set of contacts at the chip module site having a first arrangement for receiving a chip module having an electrical connector footprint conforming to the first module contact arrangement. There is also provided a second set of contacts at the chip module site having a second arrangement for receiving a chip module having an electrical connector footprint conforming to the second module contact arrangement, the second set of contacts having a different arrangement than, and being electrically connected to, the first set of contacts.
摘要:
A system and method for connecting a short word length memory to a significantly wider bus operated in an address/data multiplexing mode. A mode of operation is defined for the bus whereby the bus lines are divided for purposes of memory accessing into a data group and an address group. The data group is operable bidirectionally to read or write memory, using the addresses provided on the group of address lines. This architecture and practice is particularly suited for a boot ROM used with processors, in that such ROMs are normally of relatively short word length while the processors are of relatively long word length and are accordingly connected to buses of similar long word length. Bridge logic interfaces the processor bus to the ROM for sequencing, timing and supplemental control in converting the data from the ROM format to the processor format.