Abstract: A method and processor chip design for enabling a processor core to continue sending store operations speculatively to the store queue after the core receives indication that the store queue is full. The processor core is configured with speculative store logic that enables the processor core to continue issuing store operations while the store queue full signal is asserted. A copy of the speculatively issued store operation is placed within a speculative store buffer. The core waits for a signal from the store queue indicating the store operation was accepted into the store queue. When the speculatively-issued store operation is accepted within the store queue, the copy is discarded from the buffer. However, when the store operation is rejected, the speculative store logic re-issues the store operation ahead of normal store operations.

Abstract: In a cache coherent data processing system including at least first and second coherency domains, a memory block is stored in a system memory in association with a domain indicator indicating whether or not the memory block is cached, if at all, only within the first coherency domain. A master in the first coherency domain determines whether or not a scope of broadcast transmission of an operation should extend beyond the first coherency domain by reference to the domain indicator stored in the cache and then performs a broadcast of the operation within the cache coherent data processing system in accordance with the determination.

Abstract: In a cache coherent data processing system including at least first and second coherency domains, a master performs a first broadcast of an operation within the cache coherent data processing system that is limited in scope of transmission to the first coherency domain. The master receives a response of the first coherency domain to the first broadcast of the operation. If the response indicates the operation cannot be serviced in the first coherency domain alone, the master increases the scope of transmission by performing a second broadcast of the operation in both the first and second coherency domains. If the response indicates the operation can be serviced in the first coherency domain, the master refrains from performing the second broadcast.

Abstract: A cache memory which loads two memory values into two cache lines by receiving separate portions of a first requested memory value from a first data bus over a first time span of successive clock cycles and receiving separate portions of a second requested memory value from a second data bus over a second time span of successive clock cycles which overlaps with the first time span. In the illustrative embodiment a first input line is used for loading both a first byte array of the first cache line and a first byte array of the second cache line, a second input line is used for loading both a second byte array of the first cache line and a second byte array of the second cache line, and the transmission of the separate portions of the first and second memory values is interleaved between the first and second data busses.

Abstract: A processor includes at least one instruction execution unit that executes store instructions to obtain store operations and a store queue coupled to the instruction execution unit. The store queue includes a queue entry in which the store queue gathers multiple store operations during a store gathering window to obtain a data portion of a write transaction directed to lower level memory. In addition, the store queue includes dispatch logic that varies a size of the store gathering window to optimize store performance for different store behaviors and workloads.

Abstract: A cache memory which loads two memory values into two cache lines by receiving separate portions of a first requested memory value from a first data bus over a first time span of successive clock cycles and receiving separate portions of a second requested memory value from a second data bus over a second time span of successive clock cycles which overlaps with the first time span. In the illustrative embodiment a first input line is used for loading both a first byte array of the first cache line and a first byte array of the second cache line, a second input line is used for loading both a second byte array of the first cache line and a second byte array of the second cache line, and the transmission of the separate portions of the first and second memory values is interleaved between the first and second data busses.

Abstract: A cache coherent data processing system includes at least first and second coherency domains each including at least one processing unit and a cache memory. The cache memory includes a cache controller, a data array including a data storage location for caching a memory block, and a cache directory. The cache directory includes a tag field for storing an address tag in association with the memory block and a coherency state field associated with the tag field and the data storage location. The coherency state field has a plurality of possible states including a state that indicates that the address tag is valid, that the storage location does not contain valid data, and that the memory block is possibly cached outside of the first coherency domain.

Abstract: A cache coherent data processing system includes a plurality of processing units each having at least an associated cache, a system memory, and a memory controller that is coupled to and controls access to the system memory. The system memory includes a plurality of storage locations for storing a memory block of data, where each of the plurality of storage locations is sized to store a sub-block of data. The system memory further includes metadata storage for storing metadata, such as a domain indicator, describing the memory block. In response to a failure of a storage location for a particular sub-block among the plurality of sub-blocks, the memory controller overwrites at least a portion of the metadata in the metadata storage with the particular sub-block of data.

Abstract: In a cache coherent data processing system including at least first and second coherency domains, a memory block is stored in a system memory in association with a domain indicator indicating whether or not the memory block is cached, if at all, only within the first coherency domain. A master in the first coherency domain determines whether or not a scope of broadcast transmission of an operation should extend beyond the first coherency domain by reference to the domain indicator stored in the cache and then performs a broadcast of the operation within the cache coherent data processing system in accordance with the determination.

Abstract: In a microprocessor having a load/store unit and prefetch hardware, the prefetch hardware includes a prefetch queue containing entries indicative of allocated data streams. A prefetch engine receives an address associated with a store instruction executed by the load/store unit. The prefetch engine determines whether to allocate an entry in the prefetch queue corresponding to the store instruction by comparing entries in the queue to a window of addresses encompassing multiple cache blocks, where the window of addresses is derived from the received address. The prefetch engine compares entries in the prefetch queue to a window of 2M contiguous cache blocks. The prefetch engine suppresses allocation of a new entry when any entry in the prefetch queue is within the address window. The prefetch engine further suppresses allocation of a new entry when the data address of the store instruction is equal to an address in a border area of the address window.

Abstract: A method and apparatus for enabling protection of a particular member of a cache during LRU victim selection. LRU state array includes additional “protection” bits in addition to the state bits. The protection bits serve as a pointer to identify the location of the member of the congruence class that is to be protected. A protected member is not removed from the cache during standard LRU victim selection, unless that member is invalid. The protection bits are pipelined to MRU update logic, where they are used to generate an MRU vector. The particular member identified by the MRU vector (and pointer) is protected from selection as the next LRU victim, unless the member is Invalid. The make MRU operation affects only the lower level LRU state bits arranged a tree-based structure and thus only negates the selection of the protected member, without affecting LRU victim selection of the other members.

Abstract: Memory modules are designed with multiple write buffers utilized to temporarily hold write data. “Write-to-buffer” operations moves write data from the memory controller to the write buffers while the memory module is busy processing read operations. Then, address-only “write” commands are later issued to write the buffered write data to the memory device. The write commands targeting idle DIMMs are issued in sequence ahead of writes targeting busy DIMMs (or soon to be busy). Moving the data via a background write-to-buffer operation increases the efficiency of the common write data channel and allows the write data bus to reach maximum bandwidth during periods of heavy read activity. The actual write operations, deferred to periods of when the negative affects of the write can be completely/mostly hidden. In periods of light read activity or when there are no reads pending, buffering data in the memory module enables the buffered data to be written in parallel across multiple memory modules simultaneously.

Abstract: A cache coherent data processing system includes at least first and second coherency domains each including at least one processing unit and a cache memory. The cache memory includes a cache controller, a data array including a data storage location for caching a memory block, and a cache directory. The cache directory includes a tag field for storing an address tag in association with the memory block and a coherency state field associated with the tag field and the data storage location. The coherency state field has a plurality of possible states including a state that indicates that the address tag is valid, that the storage location does not contain valid data, and that the memory block is possibly cached outside of the first coherency domain.

Abstract: A processor includes at least one instruction execution unit that executes store instructions to obtain store operations and a store queue coupled to the instruction execution unit. The store queue includes a queue entry in which the store queue gathers multiple store operations during a store gathering window to obtain a data portion of a write transaction directed to lower level memory. In addition, the store queue includes dispatch logic that varies a size of the store gathering window to optimize store performance for different store behaviors and workloads.

Abstract: A processor includes at least one instruction execution unit that executes store instructions to obtain store operations and a store queue coupled to the instruction execution unit. The store queue includes a queue entry in which the store queue gathers multiple store operations during a store gathering window to obtain a data portion of a write transaction directed to lower level memory. In addition, the store queue includes dispatch logic that varies a size of the store gathering window to optimize store performance for different store behaviors and workloads.

Abstract: A set of N copies of bank control logic are provided for tracking the banks within the memory modules (DRAMS). When the total number of banks within the memory module(s) is greater than N, the addresses of particular banks are folded into a single grouping. The banks are represented by the N copies of the bank control logic even when the total number of banks is greater than N. Each bank within the group is tagged as being busy when any one of the banks in the group is the target of a memory access request. The algorithm folds the addresses of the banks in an order that substantially minimizes the likelihood that a bank that is in a busy or false busy state will be the target of another memory access request. Power and logic savings are recognized as only N copies of bank control logic have to be supported.

Abstract: In a microprocessor having a load/store unit and prefetch hardware, the prefetch hardware includes a prefetch queue containing entries indicative of allocated data streams. A prefetch engine receives an address associated with a store instruction executed by the load/store unit. The prefetch engine determines whether to allocate an entry in the prefetch queue corresponding to the store instruction by comparing entries in the queue to a window of addresses encompassing multiple cache blocks, where the window of addresses is derived from the received address. The prefetch engine compares entries in the prefetch queue to a window of 2M contiguous cache blocks. The prefetch engine suppresses allocation of a new entry when any entry in the prefetch queue is within the address window. The prefetch engine further suppresses allocation of a new entry when the data address of the store instruction is equal to an address in a border area of the address window.

Abstract: A cache memory which loads two memory values into two cache lines by receiving separate portions of a first requested memory value from a first data bus over a first time span of successive clock cycles and receiving separate portions of a second requested memory value from a second data bus over a second time span of successive clock cycles which overlaps with the first time span. In the illustrative embodiment a first input line is used for loading both a first byte array of the first cache line and a first byte array of the second cache line, a second input line is used for loading both a second byte array of the first cache line and a second byte array of the second cache line, and the transmission of the separate portions of the first and second memory values is interleaved between the first and second data busses.

Abstract: A cache memory which loads two memory values into two cache lines by receiving separate portions of a first requested memory value from a first data bus over a first time span of successive clock cycles and receiving separate portions of a second requested memory value from a second data bus over a second time span of successive clock cycles which overlaps with the first time span. In the illustrative embodiment a first input line is used for loading both a first byte array of the first cache line and a first byte array of the second cache line, a second input line is used for loading both a second byte array of the first cache line and a second byte array of the second cache line, and the transmission of the separate portions of the first and second memory values is interleaved between the first and second data busses.

Abstract: A cache memory which loads two memory values into two cache lines by receiving separate portions of a first requested memory value from a first data bus over a first time span of successive clock cycles and receiving separate portions of a second requested memory value from a second data bus over a second time span of successive clock cycles which overlaps with the first time span. In the illustrative embodiment a first input line is used for loading both a first byte array of the first cache line and a first byte array of the second cache line, a second input line is used for loading both a second byte array of the first cache line and a second byte array of the second cache line, and the transmission of the separate portions of the first and second memory values is interleaved between the first and second data busses. The first data bus can be one of a plurality of data busses in a first data bus set, and the second data bus can be one of a plurality of data busses in a second data bus set.