International Business Machines Corporation

Qualcomm Incorporated

Garrett M Drapala

Pak-kin Mak

Timothy Carl Bronson

Craig Raymond Walters

Michael Andrew Blake

Robert J Sonnelitter, Iii

Michael F Fee

Vesselina K Papazova

Ekaterina M Ambroladze

Gary E Strait

William J Lewis

Matthias Klein

Thomas Philip Speier

Eric Francis Robinson

Arthur J O'Neill

Kenneth Douglas Klapproth

Joseph Gerald McDonald

Hieu Trong Huynh

Hanno Ulrich

Vijayalakshmi Srinivasan

Norbert Hagspiel

Arthur J O'Neill, Jr

Sascha Junghans

Deanna P Berger

Lawrence D Curley

Rebecca Marie Gott

Deanna Postles Dunn

Nikhil Narendradev Sharma

Timothy John Slegel

Mark S Farrell

Patrick James Meaney

Harold Wade Cain, Iii

Richard Gerard Hofmann

Brian Stempel

Deanna Postles Dunn Berger

David Scott Ray

Diana Lynn Orf

Christine Comins Jones

Kevin Neal Magill

Perry Willmann Remaklus, Jr

Rebecca S Wisniewski

Tim Bronson

James F Driftmyer

Niket Choudhary

Edward J Kaminski

Derrin M Berger

An apparatus, method, and system for memory bandwidth aware data prefetching is presented. The method may comprise monitoring a number of request responses received in an interval at a current prefetch request generation rate, comparing the number of request responses received in the interval to at least a first threshold, and adjusting the current prefetch request generation rate to an updated prefetch request generation rate by selecting the updated prefetch request generation rate from a plurality of prefetch request generation rates, based on the comparison. The request responses may be NACK or RETRY responses. The method may further comprise either retaining a current prefetch request generation rate or selecting a maximum prefetch request generation rate as the updated prefetch request generation rate in response to an indication that prefetching is accurate.

Method, apparatus, and system for memory bandwidth aware data prefetching

Disclosed are ticketed flow control mechanisms in a processing system with one or more masters and one or more slaves. In an aspect, a targeted slave receives a request from a requesting master. If the targeted slave is unavailable to service the request, a ticket for the request is provided to the requesting master. As resources in the targeted slave become available, messages are broadcasted for the requesting master to update the ticket value. When the ticket value has been updated to a final value, the requesting master may re-transmit the request.

Ticket based request flow control

Selective honoring of speculative memory-prefetch requests based on bandwidth constraint of a memory access path component(s) in a processor-based system. To reduce memory access latency, a CPU includes a request size in a memory read request of requested data to be read from memory and a request mode of the requested data as required or preferred. A memory access path component includes a memory read honor circuit configured to receive the memory read request and consult the request size and request mode of requested data in the memory read request. If the selective prefetch data honor circuit determines that bandwidth of the memory system is less than a defined bandwidth constraint threshold, then the memory read request is forwarded to be fulfilled, otherwise, the memory read request is downgraded to only include any requested required data.

Selectively honoring speculative memory prefetch requests based on bandwidth state of a memory access path component(s) in a processor-based system

A cache coherency management facility to reduce latency in granting exclusive access to a cache in certain situations. A node requests exclusive access to a cache line of the cache. The node is in one region of nodes of a plurality of regions of nodes. The one region of nodes includes the node requesting exclusive access and another node of the computing environment, in which the node and the another node are local to one another as defined by a predetermined criteria. The node requesting exclusive access checks a locality cache coherency state of the another node, the locality cache coherency state being specific to the another node and indicating whether the another node has access to the cache line. Based on the checking indicating that the another node has access to the cache line, a determination is made that the node requesting exclusive access is to be granted exclusive access to the cache line. The determining being independent of transmission of information relating to the cache line from one or more other nodes of the one or more other regions of nodes.

Granting exclusive cache access using locality cache coherency state

A method, computer program product, and system for maintaining a proper ordering of a data steam that includes two or more sequentially ordered stores, the data stream being moved to a destination memory device, the two or more sequentially ordered stores including at least a first store and a second store, wherein the first store is rejected by the destination memory device. A computer-implemented method includes sending the first store to the destination memory device. A conditional request is sent to the destination memory device for approval to send the second store to the destination memory device, the conditional request dependent upon successful completion of the first store. The second store is cancelled responsive to receiving a reject response corresponding to the first store.

Increased bandwidth of ordered stores in a non-uniform memory subsystem

A computer system comprising multiple nodes, each node comprising a plurality of processors and a local cache hierarchy, suppresses local cache coherency of a node operations or global cache coherency operations between nodes based on the coherency request being a global or local request, and the state of the cache line at the node.

Nested cache coherency protocol in a tiered multi-node computer system

Maintaining store order with high throughput in a distributed shared memory system. A request is received for a first ordered data store and a coherency check is initiated. A signal is sent that pipelining of a second ordered data store can be initiated. If a delay condition is encountered during the coherency check for the first ordered data store, rejection of the first ordered data store is signaled. If a delay condition is not encountered during the coherency check for the first ordered data store, a signal is sent indicating a readiness to continue pipelining of the second ordered data store.

Maintaining order with parallel access data streams

In one embodiment, a computer-implemented method includes instructing two or more processors that are operating in a normal state of a symmetric multiprocessing (SMP) network to transition from the normal state to a slow state. The two or more processors reduce their frequencies to respective target frequencies in a transitional state when transitioning from the normal state to the slow state. It is determined that the two or more processors have achieved their respective target frequencies for the slow state. The slow state is entered, responsive to this determination. Responsive to entering the slow state, a first processor of the two or more processors is instructed to send empty packets across an interconnect to compensate for a first greatest potential rate differential between the first processor and a remainder of the two or more processors during the slow state.

Dynamic synchronous to asynchronous frequency transitions in high-performance symmetric multiprocessing

In one embodiment, a computer-implemented method includes detecting a cache miss for a cache line. A resource is reserved on each of one or more remote computing nodes, responsive to the cache miss. A request for a state of the cache line on the one or more remote computing nodes is broadcast to the one or more remote computing nodes, responsive to the cache miss. A resource credit is received from a first remote computing node of the one or more remote computing nodes, responsive to the request. The resource credit indicates that the first remote computing node will not participate in completing the request. The resource on the first remote computing node is released, responsive to receiving the resource credit from the first remote computing node.

Early shared resource release in symmetric multiprocessing computer systems

In an approach for backing up designated data located in a cache, data stored within an index of a cache is identified, wherein the data has an associated designation indicating that the data is applicable to be backed up to a higher level memory. It is determined that the data stored to the cache has been updated. A status associated with the data is adjusted, such that the adjusted status indicates that the data stored to the cache has not been changed. A copy of the data is created. The copy of the data is stored to the higher level memory.

Designated cache data backup during system operation

Growing community of inventors

Poughkeepsie, NY, United States of America

Garrett M Drapala