International Business Machines Corporation

Globalfoundries Inc.

Sanjeev Ghai

Guy L Guthrie

Ravi Kumar Arimilli

James Stephen Fields, Jr

Derek E Williams

William John Starke

Hugh Shen

Lakshminarayana Baba Arimilli

Warren Edward Maule

Praveen S Reddy

Phillip G Williams

John Steven Dodson

Jody B Joyner

Kenneth Lee Wright

Hien Le

Mark Andrew Brittain

Jeff A Stuecheli

Philip G Williams

Jeffrey A Stuecheli

Jonathan R Jackson

Timothy Jay Dell

Edgar Rolando Cordero

Geraint M North

Stephen J Powell

Scott Barnett Swaney

John Thomas Hollaway, Jr

Luiz Carlos Alves

Bernard Charles Drerup

Bradly George Frey

Cathy May

Mohit Karve

Naga P Gorti

Luke Murray

Alexander Michael Taft

Luis E De La Torre

Hung Doan

A method, programming product, processor, and/or system for prefetching data is disclosed that includes: receiving a request for data at a cache; identifying whether the request for data received at the cache is a demand request or a prefetch request; and determining, in response to identifying that the request for data received at the cache is a prefetch request, whether to terminate the prefetch request, wherein determining whether to terminate the prefetch request comprises: determining how many hits have occurred for a prefetch stream corresponding to the prefetch request received at the cache; and determining, based upon the number of hits that have occurred for the prefetch stream corresponding to the prefetch request received by the cache, whether to terminate the prefetch request.

Prefetch unit filter for microprocessor

A processing unit for a data processing system includes a processor core that issues memory access requests and a cache memory coupled to the processor core. The cache memory includes a reservation circuit that tracks reservations established by the processor core via load-reserve requests and a plurality of read-claim (RC) state machines for servicing memory access requests of the processor core. The cache memory, responsive to receipt from the processor core of a store-conditional request specifying a store target address, allocates an RC state machine among the plurality of RC state machines to process the store-conditional request and transfers responsibility for tracking a reservation for the store target address from the reservation circuit to the RC state machine.

Accelerated processing of streams of load-reserve requests

An information handling system (IHS) includes a processor with a cache memory system. The processor includes a processor core with an L1 cache memory that couples to an L2 cache memory. The processor includes an arbitration mechanism that controls load and store requests to the L2 cache memory. The arbitration mechanism includes control logic that enables a load request to interrupt a store request that the L2 cache memory is currently servicing. When the L2 cache memory finishes servicing the interrupting load request, the L2 cache memory may return to servicing the interrupted store request at the point of interruption.

Information handling system with immediate scheduling of load operations

A data processing system includes multiple processing units all having access to a shared memory. A processing unit of the data processing system includes a processor core including an upper level cache, core reservation logic that records addresses in the shared memory for which the processor core has obtained reservations, and an execution unit that executes memory access instructions including a fronting load instruction. Execution of the fronting load instruction generates a load request that specifies a load target address. The processing unit further includes lower level cache that, responsive to receipt of the load request and based on the load request indicating an address match for the load target address in the core reservation logic, protects the load target address against access by any conflicting memory access request during a protection interval following servicing of the load request.

Synchronized access to data in shared memory by protecting the load target address of a fronting load

A processing unit includes a processor core that executes a store-conditional instruction that generates a store-conditional request specifying a store target address. The processing unit further includes a reservation register that records shared memory addresses for which the processor core has obtained reservations and a cache that services the store-conditional request by conditionally updating the shared memory with the store data based on the reservation register indicating a reservation for the store target address. The cache includes a blocking state machine configured to protect the store target address against access by any conflicting memory access request snooped on a system interconnect during a protection window extension following servicing of the store-conditional request. The cache is configured to vary a duration of the protection window extension for different snooped memory access requests based on one of broadcast scopes and the relative locations of masters of the snooped memory access requests.

Synchronizing access to shared memory by extending protection for a target address of a store-conditional request

In a processing unit, a processor core executes instructions in a plurality of simultaneous hardware threads, where multiple of the plurality of hardware threads concurrently execute memory transactions. A transactional memory circuit in the processing unit tracks transaction footprints of the memory transactions of the multiple hardware thread. In response to detecting failure of a given memory transaction of one of the plurality of multiple threads due to an overflow condition, the transactional memory circuit transitions to a throttled operating mode and reduces a number of hardware threads permitted to concurrently execute memory transactions.

Dynamic transaction throttling in a data processing system supporting transactional memory

A data processing system includes multiple processing units all having access to a shared memory. A processing unit includes a processor core that executes memory access instructions including a store-conditional instruction that generates a store-conditional request specifying a store target address and store data. The processing unit further includes a reservation register that records shared memory addresses for which the processor core has obtained reservations and a cache that services the store-conditional request by conditionally updating the shared memory with the store data based on the reservation register indicating a reservation for the store target address. The processing unit additional includes a blocking state machine configured to protect the store target address against access by any conflicting memory access request during a protection window extension following servicing of the store-conditional request.

Synchronized access to shared memory by extending protection for a store target address of a store-conditional request

A data processing system includes a plurality of processor cores each having a respective store-through upper level cache and a store-in banked lower level cache. Store requests of the plurality of processor cores destined for the banked lower level cache are buffered in multiple store queues including a first store queue and a second store queue. In response to determining that the multiple store queues contain store requests targeting a common bank of the banked lower level cache, store requests from the first store queue are temporarily favored for selection for issuance to the banked lower level cache over those in the second store queue.

Banked cache temporarily favoring selection of store requests from one of multiple store queues

Temporarily favoring selection of store requests from one of multiple store queues for issuance to a bank of a banked cache

A processor core has a store-through upper level cache and a store-in lower level cache. In response to execution of a memory move instruction sequence including a plurality of copy-type instruction and a plurality of paste-type instructions, the processor core transmits a corresponding plurality of copy-type and paste-type requests to the lower level cache, where each copy-type request specifies a source real address and each paste-type request specifies a destination real address. In response to receipt of each copy-type request, the lower level cache copies a respective one of a plurality of data granules from a respective storage location specified by the source real address of that copy-type request into a non-architected buffer. In response to receipt of each paste-type request, the lower level cache writes a respective one of the plurality of data granules from the non-architected buffer to a respective storage location specified by the destination real address of that paste-type request.

Memory move instruction sequence including a stream of copy-type and paste-type instructions

A data processing system includes a processor core having a store-through upper level cache and a store-in lower level cache. In response to a first instruction, the processor core generates a copy-type request and transmits the copy-type request to the lower level cache, where the copy-type request specifies a source real address. In response to a second instruction, the processor core generates a paste-type request and transmits the paste-type request to the lower level cache, where the paste-type request specifies a destination real address. In response to receipt of the copy-type request, the lower level cache copies a data granule from a storage location specified by the source real address into a non-architected buffer, and in response to receipt of the paste-type request, the lower level cache writes the data granule from the non-architected buffer to a storage location specified by the destination real address.

Memory access in a data processing system utilizing copy and paste instructions

A technique for operating a data processing system includes determining whether a cache line that is to be victimized from a cache includes high availability (HA) data that has not been logged. In response determining that the cache line that is to be victimized from the cache includes HA data that has not been logged, an address for the HA data is written to an HA dirty address data structure, e.g., a dirty address table (DAT), in a first memory via a first non-blocking channel. The cache line that is victimized from the cache is written to a second memory via a second non-blocking channel.

Techniques for logging addresses of high-availability data via a non-blocking channel

A processor core is supported by an upper level cache and a lower level cache that receives, from an interconnect fabric, a write injection request requesting injection of a partial cache line of data into a target cache line identified by a target real address. In response to receipt of the write injection request, a determination is made that the upper level cache is a highest point of coherency for the target real address. In response to the determination, the upper level cache and lower level cache collaborate to transfer the target cache line from the upper level cache to the lower level cache. The lower level cache updates the target cache line by merging the partial cache of data into the target cache line and storing the updated target cache line in the lower level cache.

Injection of at least a partial cache line in a private multilevel cache hierarchy

A technique for operating a high-availability (HA) data processing system includes, in response to receiving an HA logout indication at a cache, initiating a walk of the cache to locate cache lines in the cache that include HA data. In response to determining that a cache line includes HA data, an address of the cache line is logged in a first portion of a buffer in the cache. In response to the first portion of the buffer reaching a determined fill level, contents of the first portion of the buffer are logged to another memory. In response to all cache lines in the cache being walked, the cache walk is terminated.

Cache configured to log addresses of high-availability data

Growing community of inventors

Round Rock, TX, United States of America

Sanjeev Ghai