Techniques for implementing and/or operating an apparatus, which includes a memory system coupled to a processing system via a memory bus. The memory system includes hierarchical memory levels and a memory controller. The memory controller receives a memory access request at least in part by receiving an address parameter indicative of a memory address associated with a data block from the memory bus during a first clock cycle and receiving a context parameter indicative of context information associated with current targeting of the data block from the memory bus during a second clock cycle, instructs the memory system to output the data block to the memory bus based on the memory address indicated in the address parameter, and predictively controls data storage in the hierarchical memory levels based at least in part on the context information indicated in the context parameter of the memory access request.

In one embodiment, a cache memory includes: a plurality of data banks, each of the plurality of data banks having a plurality of entries each to store a portion of a cache line distributed across the plurality of data banks; and a plurality of tag banks decoupled from the plurality of data banks, wherein a tag for a cache line is to be assigned to one of the plurality of tag banks. Other embodiments are described and claimed.

A first master receives a first virtual address in a virtual memory, the first virtual address in the virtual memory corresponding, according to a mapping function, to a first physical address of a first physical memory bank which is to be accessed by the first master. The first master accesses the first physical address to perform a first memory operation in the first memory bank. A second master receives a second virtual address in a virtual memory, the second virtual address in the virtual memory corresponding, according to the mapping function, to a second physical address of a second physical memory bank which is to be accessed by the second master. Concurrently with access by the first master to the first physical address, the second master accesses the second physical address to perform a second memory operation in the second physical memory bank.

A control method of a storage device wherein a host cannot transfer a command to the storage device when the storage device transfers data to the host, after which there is a data transfer delay time period and no data is transferred to the host until a read command is received from the host, the control method comprising the steps of: detecting, by a memory controller of the storage device, a host delay time of the host each time a read command is received from the host during the data transfer delay time period; and adjusting, by the memory controller, the data transfer delay time period based on one or more of the detected host delay times.

A cache memory for a processor including an arbiter, a tag array and a request queue. The arbiter arbitrates among multiple memory access requests and provides a selected memory access request. The tag array has a first read port receiving the selected memory access request and has a second read port receiving a prefetch request from a prefetcher. The tag array makes a hit or miss determination of whether data requested by the selected memory access request or the prefetch request is stored in a corresponding data array. The request queue has a first write port for receiving the selected memory access request when it misses in the tag array, and has a second write port for receiving the prefetch request when it misses in the tag array. The additional read and write ports provide a separate and independent pipeline path for handing prefetch requests.

Bus protocol features are provided for chaining memory access requests on a high speed interconnect bus, allowing for reduced signaling overhead. Multiple memory request messages are received over a bus. A first message has a source identifier, a target identifier, a first address, and first payload data. The first payload data is stored in a memory at locations indicated by the first address. Within a selected second one of the request messages, a chaining indicator is received associated with the first request message and second payload data. The second request message does not include an address. Based on the chaining indicator, a second address for which memory access is requested is calculated based on the first address. The second payload data is stored in the memory at locations indicated by the second address.

A scheduler of a memory system is provided. The scheduler may include a pattern storage part and a pattern selector. The pattern storage part may have a plurality of storage patterns, each of the storage patterns provide for a process sequence for a plurality of instructions. The pattern selector may be configured to select one of the plurality of storage patterns in the pattern storage part and generate a schedule such that external instructions are executed in the process sequence set by the selected storage pattern.

A high-bandwidth memory (HBM) system includes an HBM device and a logic circuit. The logic circuit includes a first interface coupled to a host device and a second interface coupled to the HBM device. The logic circuit receives a first command from the host device through the first interface and converts the received first command to a first processing-in-memory (PIM) command that is sent to the HBM device through the second interface. The first PIM command has a deterministic latency for completion. The logic circuit further receives a second command from the host device through the first interface and converting the received second command to a second PIM command that is sent to the HBM device through the second interface. The second PIM command has a non-deterministic latency for completion.

A technique for scheduling processing tasks having different latencies is provided. The technique involves identifying one or more available requests in a request queue, where each request queue corresponds to a different latency. A request arbiter examines a shift register to determine whether there is an available slot for the one or more requests. A slot is available for a request if there is a slot that is a number of slots from the end of the shift register equal to the number of cycles the request takes to complete processing in a corresponding processing pipeline. If a slot is available, the request is scheduled for execution and the slot is marked as being occupied. If a slot is not available, the request is not scheduled for execution on the current cycle. On transitioning to a new cycle, the shift register is shifted towards its end and the technique repeats.

A high-bandwidth memory (HBM) system includes an HBM device and a logic circuit. The logic circuit includes a first interface coupled to a host device and a second interface coupled to the HBM device. The logic circuit receives a first command from the host device through the first interface and converts the received first command to a first processing-in-memory (PIM) command that is sent to the HBM device through the second interface. The first PIM command has a deterministic latency for completion. The logic circuit further receives a second command from the host device through the first interface and converting the received second command to a second PIM command that is sent to the HBM device through the second interface. The second PIM command has a non-deterministic latency for completion.