International Business Machines Corporation

Vern A Norton

Gregory Francis Pfister

Kevin Patrick McAuliffe

William C Brantley, Jr

Joseph Weiss

Evelyn A Melton

Kimming K So

Bharat D Rathi

Rory D Jackson

Wayne S Groh

Gregoty F Pfister

Keven P McAuliffe

Scott P Wakefield

John H Anthony

An electronic computer system including a central processor and a hierarchical memory system having a large relatively low speed random access system memory and a small high speed set-associative cache memory including a data store section for storing lines of data from the system memory and a cache directory for indicating, by means of line identifier fields at any time, the lines of the system memory data currently resident in cache, is provided with a way to improve the distribution of data across the congruence classes within the cache. A mechanism is provided for performing a permutation operation on an M bit portion (X) of the system memory address, which permutation determines the congruence class into which the address will map. The permutation mechanism performs a bit-matrix multiplication of said M-bit address with an M.times.M matrix (where M is a real positive integer greater than 1) to produce a permuted M-bit address (X'). The directory controls utilize the permuted M-bit address (X') to determine the congruence class of any given memory access and automatically access the congruence class of the permuted address (X') subsequent to the permutation operation to determine if one of the line identifiers which identifies, every member of a congruence class currently stored in the directory, matches an identifier field from the memory access request from the CPU. If the match is successful the data store portion of the cache is accessed at the permuted M-bit address (X') and the requested data line is accessed at the address field specified by the CPU.

Mechanism for improving the randomization of cache accesses utilizing

Packet switch protocol and circuitry for implementing it are disclosed. According to this protocol, a message transmitter of a first node in the network may send data through a data transmission link at a predetermined rate until it is signalled, via a control signal generated by a message receiver in a second node, to suspend its transmissions. The message transmitter may also be signalled to resume transmitting data. The message receiver includes a buffer memory in which messages are temporarily stored if their selected path is blocked as they pass through the network. When the amount of available space in the buffer is less than a preprogrammed threshold value, the message receiver generates the control signal to suspend message transmission. This threshold value leaves sufficient space in the buffer to store any data which may be in the pipeline between the transmitter and the receiver. When the amount of available space rises above this threshold, the message receiver indicates to the transmitter to resume transmission. A message spanning an interface during transmission may be suspended without losing spatial contiguousness. The message transmitter also transmits a signal which marks the last portion of a message. This signal is stored in the buffer with the message data and is used to delimit messages in the buffer, allowing multiple messages to be stored in one buffer without extra delimiting values.

System for implementing to a packet switch protocol for a multinode data

An aperiodic mapping procedure for the mapping of logical to physical addresses is defined as a permutation function for generating optimized stride accesses in an interleaved multiple device system such as a large, parallel processing shared memory system wherein the function comprises a bit-matrix multiplication of a presented first (logical) address with a predetermined matrix to produce a second (physical) address. The permutation function maps the address from a first to a second address space for improved memory performance in such an interleaved memory system. Assuming that the memory has n logical address bits and 2.sub.d separately accessible memory devices (where d.ltoreq.n) and a second address that utilizes n-d bits of the first address as the offset within the referenced device node. The procedure includes performing a bit matrix multiplication between successive roows of the said matrix and bits of the first address to produce successive d bits of the second address.

Aperiodic mapping system using power-of-two stride access to interleaved

A multiprocessing system is presented having a plurality of processing nodes interconnected together by a communication network, each processing node including a processor, responsive to user software running on the system, and an associated memory module, and capable under user control of dynamically partitioning each memory module into a global storage efficiently accessible by a number of processors connected to the network, and local storage efficiently accessible by its associated processor.

Multiprocessing system having nodes containing a processor and an

An interconnection network management architecture for use with a large shared memory multiprocessor computing system including a plurality of processors and a plurality of separately addressable main memory modules. Two parallel, interconnection networks are provided each capable of interconnecting any processor to any memory module, and each having different latency characteristics. A Hot-Spot detection mechnaism is associated with each main memory module for detecting when a particular address in that module has become a Hot Spot and includes a first memory for storing all detected Hot Spots. A diverter element is associated with each processor for selectively routing memory requests over either the first or second memory network contingent on its status as a Hot Spot. A second memory is included in each diverter element for storing all Hot Spots detected by the detector elements. A control mechanism determines if any current main memory address is a listed Hot Spot within the second memory and, if so, causes the main memory reference to be transmitted to the memory system over a selected interconnection network. Another component of this mechanism determines if a particular address has been accessed a sufficient number of times within a given timeframe to be deemed a Hot Spot and removes Hot Spots from both the first and second memories when necessary.

Hardware mechanism for automatically detecting hot-spot references and

A method and apparatus for marking data that is temporarily cacheable to facilitate the efficient management of said data. A bit in the segment and/or page descriptor of the data called the marked data bit (MDB) is generated by the compiler and included in a request for data from memory by the processor in the form of a memory address and will be stored in the cache directory at a location related to the particular line of data involved. The bit is passed to the cache together with the associated real address after address translation (in the case of a real cache). when the cache controls load the address of the data in the directory it is also stored the marked data bit (MDB) in the directory with the address. When the cacheability of the temporarily cacheable data changes from cacheable to non-cacheable, a single instruction is issued to cause the cache to invalidate all marked data. When an 'invalidate marked data' instruction is received, the cache controls sweep through the entire cache directory and invalidate any cache line which has the 'marked data bit' set in a single pass. An extension of the invention involves using a multi-bit field rather than a single bit to provide a more versatile control of the temporary cacheability of data.

Method and apparatus for efficiently handling temporarily cacheable data

A multiprocessing system is presented for dynamically partitioning a storage module into a global storage efficiently accessible by a number of processors connected to a network, and local storage efficiently accessible by individual processors, including the interleaving of storage references output by a processor, under the control of that processor, and dynamically directing the storage references to first or second portions of the storage module.

Growing community of inventors

Croton-on-Hudson, NY, United States of America

Vern A Norton