A method and system for managing a buffer device in a storage system. The method comprising determining a first priority for a first queue included in the buffer device, the first queue comprising at least one data page associated with a first storage device in the storage system; in at least one round, in response to the first priority not satisfying a first predetermined condition, updating the first priority according to a first updating rule, the first updating rule making the updated first priority much closer to the first predetermined condition than the first priority; and in response to the first priority satisfying the first predetermined condition, flushing data in a data page in the first queue to the first storage device.

A data access system including a processor and a storage system including a main memory and a cache module. The cache module includes a FLC controller and a cache. The cache is configured as a FLC to be accessed prior to accessing the main memory. The processor is coupled to levels of cache separate from the FLC. The processor generates, in response to data required by the processor not being in the levels of cache, a physical address corresponding to a physical location in the storage system. The FLC controller generates a virtual address based on the physical address. The virtual address corresponds to a physical location within the FLC or the main memory. The cache module causes, in response to the virtual address not corresponding to the physical location within the FLC, the data required by the processor to be retrieved from the main memory.

A read method executed by a computing system includes a processor, at least one nonvolatile memory, and at least one cache memory performing a cache function of the at least one nonvolatile memory. The method includes receiving a read request regarding a critical word from the processor. A determination is made whether a cache miss is generated, through a tag determination operation corresponding to the read request. Page data corresponding to the read request is received from the at least one nonvolatile memory in a wraparound scheme when a result of the tag determination operation indicates that the cache miss is generated. The critical word is output to the processor when the critical word of the page data is received.

A tensor computation dataflow accelerator semiconductor circuit is disclosed. The data flow accelerator includes a DRAM bank and a peripheral array of multiply-and-add units disposed adjacent to the DRAM bank. The peripheral array of multiply-and-add units are configured to form a pipelined dataflow chain in which partial output data from one multiply-and-add unit from among the array of multiply-and-add units is fed into another multiply-and-add unit from among the array of multiply-and-add units for data accumulation. Near-DRAM-processing dataflow (NDP-DF) accelerator unit dies may be stacked atop a base die. The base die may be disposed on a passive silicon interposer adjacent to a processor or a controller. The NDP-DF accelerator units may process partial matrix output data in parallel. The partial matrix output data may be propagated in a forward or backward direction. The tensor computation dataflow accelerator may perform a partial matrix transposition.

A general matrix-matrix multiplication (GEMM) dataflow accelerator circuit is disclosed that includes a smart 3D stacking DRAM architecture. The accelerator circuit includes a memory bank, a peripheral lookup table stored in the memory bank, and a first vector buffer to store a first vector that is used as a row address into the lookup table. The circuit includes a second vector buffer to store a second vector that is used as a column address into the lookup table, and lookup table buffers to receive and store lookup table entries from the lookup table. The circuit further includes adders to sum the first product and a second product, and an output buffer to store the sum. The lookup table buffers determine a product of the first vector and the second vector without performing a multiply operation. The embodiments include a hierarchical lookup architecture to reduce latency. Accumulation results are propagated in a systolic manner.

A method and system for managing a buffer device in a storage system. The method comprising determining a first priority for a first queue included in the buffer device, the first queue comprising at least one data page associated with a first storage device in the storage system; in at least one round, in response to the first priority not satisfying a first predetermined condition, updating the first priority according to a first updating rule, the first updating rule making the updated first priority much closer to the first predetermined condition than the first priority; and in response to the first priority satisfying the first predetermined condition, flushing data in a data page in the first queue to the first storage device.

A read method executed by a computing system includes a processor, at least one nonvolatile memory, and at least one cache memory performing a cache function of the at least one nonvolatile memory. The method includes receiving a read request regarding a critical word from the processor. A determination is made whether a cache miss is generated, through a tag determination operation corresponding to the read request. Page data corresponding to the read request is received from the at least one nonvolatile memory in a wraparound scheme when a result of the tag determination operation indicates that the cache miss is generated. The critical word is output to the processor when the critical word of the page data is received.

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.

Provided are a computer program product, system, and method for considering a frequency of access to groups of tracks and density of the groups to select groups of tracks to destage. One of a plurality of densities for one of a plurality of groups of tracks is incremented in response to determining at least one of that the group is not ready to destage and that one of the tracks in the group in the cache transitions to being ready to destage. A determination is made of a group frequency indicating a frequency at which tracks in the group are modified. At least one of the density and the group frequency is used for each of the groups to determine whether to destage the group. The tracks in the group in the cache are destaged to the storage in response to determining to destage the group.

A Green NAND SSD (GNSD) controller receives reads and writes from a host and writes to flash memory. A SSD DRAM has a DRAM Translation Layer (ETL) with buffers managed by the GNSD controller. The GNSD controller performs deduplication, compression, encryption, high-level error-correction, and grouping of host data writes, and manages mapping tables to store host write data in the SSD DRAM to reduce writes to flash memory. The GNSD controller categorizes host writes as data types for paging files, temporary files, meta-data, and user data files, using address ranges and file extensions read from meta-data tables. Paging files and temporary files are optionally written to flash. Full-page and partial-page data are grouped into multi-page meta-pages by data type before storage by the GNSD controller. Status bits include two overwrite bits indicating frequently-written data that is retained in the SSD DRAM rather than being flushed to flash and re-allocated.