In one embodiment, a node coupled to solid state drives (SSDs) of a plurality of storage arrays executes a storage input/output (I/O) stack having a plurality of layers. The node includes a non-volatile random access memory (NVRAM). A first portion of the NVRAM is configured as a write-back cache to store write data associated with a write request and a second portion of the NVRAM is configured as one or more non-volatile logs (NVLogs) to record metadata associated with the write request. The write data is passed from the write-back cache over a first path of the storage I/O stack for storage on a first storage array and the metadata is passed from the one or more NVLogs over a second path of the storage I/O stack for storage on a second storage array, wherein the first path is different from the second path.

Apparatuses and methods related to providing coherent memory access. An apparatus for providing coherent memory access can include a memory array, a first processing resource, a first cache line and a second cache line coupled to the memory array, a first cache controller, and a second cache controller. The first cache controller coupled to the first processing resource and to the first cache line can be configured to provide coherent access to data stored in the second cache line and corresponding to a memory address. A second cache controller coupled through an interface to a second processing resource external to the apparatus and coupled to the second cache line can be configured to provide coherent access to the data stored in the first cache line and corresponding to the memory address. Coherent access can be provided using a first cache line address register of the first cache controller which stores the memory address and a second cache line address register of the second cache controller which also stores the memory address.

A queuing requester for access to a memory system is provided. Transaction requests are received from two or more requestors for access to the memory system. Each transaction request includes an associated priority value. A request queue of the received transaction requests is formed in the queuing requester. Each transaction request includes an associated priority value. A highest priority value of all pending transaction requests within the request queue is determined. An elevated priority value is selected when the highest priority value is higher than the priority value of an oldest transaction request in the request queue; otherwise the priority value of the oldest transaction request is selected. The oldest transaction request in the request queue with the selected priority value is then provided to the memory system. An arbitration contest with other requesters for access to the memory system is performed using the selected priority value.

An example memory sub-system includes: a plurality of bank groups, wherein each bank group comprises a plurality of memory banks; a plurality of row buffers, wherein two or more row buffers of the plurality of row buffers are associated with each memory bank; a cache comprising a plurality of cache lines; a processing logic communicatively coupled to the plurality of bank groups and the plurality of row buffers, the processing logic to perform operations comprising: receiving an activate command specifying a row of a memory bank of the plurality of memory banks; fetching data from the specified row to a row buffer of the plurality of row buffers; and copying the data to a cache line of the plurality of cache lines.

A system includes a first memory device of a first memory type, a second memory device of a second memory type, and a third memory device of a third memory type. The system further includes a processing device to retrieve one or more sections of data from the first memory device comprising a first memory type, and retrieve one or more remaining sections of data from the second memory device comprising a second memory type, wherein the one or more remaining sections of data from the second memory device are associated with the one or more sections of data from the first memory device. The processing device is further to combine the one or more sections of data from the first memory device comprising the first memory type with the one or more remaining sections of each of data from the second memory device comprising the second memory type into a contiguous page, and copy the contiguous page to a third memory device comprising a third memory type.

A system includes a first memory component of a first memory type, a second memory component of a second memory type with a higher access latency than the first memory component, and a third memory component of a third memory type with a higher access latency than the first and second memory components. The system further includes a processing device to identify a section of a data page stored in the first memory component, and access patterns associated with the data page and the section of the data page. The processing device determines to cache the data page at the second memory component based on the access patterns, copying the section of the data page stored in the first memory component to the second memory component. The processing device then copies additional sections of the data page stored at the third memory component to the second memory component.

A three-dimensional (3D) memory device includes a 3D NAND memory array, an on-die static random-access memory (SRAM), and peripheral circuits formed on the same chip with the on-die SRAM. The peripheral circuits include a page buffer coupled to the on-die SRAM and a controller coupled to the on-die SRAM and the page buffer. The controller may be configured to load program data into the page buffer and cache the program data into the on-die SRAM as a backup copy of the program data. In response to a status of programming the program data from the page buffer into the 3D NAND memory array being failed, the controller may be further configured to transmit the backup copy of the program data in the on-die SRAM to the page buffer, and program the backup copy of the program data in the page buffer into the 3D NAND memory array.

A system includes a first memory device of a first memory type, a second memory device of a second memory type, and a third memory device of a third memory type. The system further includes a processing device to retrieve one or more sections of data from the first memory device comprising a first memory type, and retrieve one or more remaining sections of data from the second memory device comprising a second memory type, wherein the one or more remaining sections of data from the second memory device are associated with the one or more sections of data from the first memory device. The processing device is further to combine the one or more sections of data from the first memory device comprising the first memory type with the one or more remaining sections of each of data from the second memory device comprising the second memory type into a contiguous page, and copy the contiguous page to a third memory device comprising a third memory type.

Apparatuses and methods related to providing coherent memory access. An apparatus for providing coherent memory access can include a memory array, a first processing resource, a first cache line and a second cache line coupled to the memory array, a first cache controller, and a second cache controller. The first cache controller coupled to the first processing resource and to the first cache line can be configured to provide coherent access to data stored in the second cache line and corresponding to a memory address. A second cache controller coupled through an interface to a second processing resource external to the apparatus and coupled to the second cache line can be configured to provide coherent access to the data stored in the first cache line and corresponding to the memory address. Coherent access can be provided using a first cache line address register of the first cache controller which stores the memory address and a second cache line address register of the second cache controller which also stores the memory address.

A memory mapping method includes storing a feature map including a plurality of sets of data used for a neural network operation in a memory, shifting a position of a portion of the data included in the feature map that is stored in the memory based on a parameter of the neural network operation, and outputting data requested by the neural network operation from the feature map in which the position of the portion is shifted based on a memory bandwidth.