Techniques for computer memory management are disclosed herein. In one embodiment, a method includes in response to receiving a request for allocation of memory, determining whether the request is for allocation from a first memory region or a second memory region of the physical memory. The first memory region has first memory subregions of a first size and the second memory region having second memory subregions of a second size larger than the first size of the first memory region. The method further includes in response to determining that the request for allocation of memory is for allocation from the first or second memory region, allocating a portion of the first or second multiple memory subregions of the first or second memory region, respectively, in response to the request.

According to one embodiment, a memory system includes a non-volatile semiconductor memory, a block management unit, and a transcription unit. The semiconductor memory includes a plurality of blocks to which data can be written in both the first mode and the second mode. The block management unit manages a block that stores therein no valid data as a free block. When the number of free blocks managed by the block management unit is smaller than or equal to a predetermined threshold value, the transcription unit selects one or more used blocks that stores therein valid data as transcription source blocks and transcribes valid data stored in the transcription source blocks to free blocks in the second mode.

The embodiments of this application provide a method and device for optimizing neural network. The method includes: binarizing and bit-packing input data of a convolution layer along a channel direction, and obtaining compressed input data; binarizing and bit-packing respectively each convolution kernel of the convolution layer along the channel direction, and obtaining each corresponding compressed convolution kernel; dividing the compressed input data sequentially in a convolutional computation order into blocks of the compressed input data with the same size of each compressed convolution kernel, wherein the data input to one time convolutional computation form a data block; and, taking a convolutional computation on each block of the compressed input data and each compressed convolution kernel sequentially, obtaining each convolutional result data, and obtaining multiple output data of the convolution layer according to each convolutional result data.

A method and solid-state storage device are disclosed for validating erasure status of data blocks on a solid-state drive. The method includes assigning each data block of a plurality of data blocks on the solid-state drive, a block identifier and an erasure status, the block identifier being system data, user data, or unmapped data, and the erasure status being erased or not erased.

This disclosure relates to a method for situation-dependent storage of data of a system, in which data of the system is detected, is amalgamated in at least one data block and is stored in a volatile memory, and in which, in response to the occurrence of at least one predefined trigger event in the at least one data block, amalgamated data are transferred from the volatile memory into a read-only memory, and in which a time window, in which the data for the at least one data block is captured, is selected automatically and dynamically according to the at least one trigger event.

Various embodiments set forth techniques for free space management in a block store. The techniques include receiving a request to allocate one or more blocks in a block store, accessing a sparse hierarchical data structure to identify an allocator page identifying a region of a backing store having a greatest number of free blocks, and allocating the one or more blocks.

Methods, systems, and devices for maintenance command interfaces for a memory system are described. A host system and a memory system may be configured according to a shared protocol that supports enhanced management of maintenance operations between the host system and memory system, such as maintenance operations to resolve error conditions at a physical address of a memory system. In some examples, a memory system may initiate maintenance operations based on detections performed at the memory system, and the memory system may provide a maintenance indication for the host system. In some examples, a host system may initiate maintenance operations based on detections performed at the host system. In various examples, the described maintenance signaling may include capability signaling between the host system and memory system, status indications between the host system and memory system, and other maintenance management techniques.

A storage controller including: a host interface circuit receiving first, second, third and fourth requests corresponding to first, second, third and fourth logical addresses; a memory interface circuit communicating with first nonvolatile memories through a first channel and second nonvolatile memories through a second channel; a first flash translation layer configured to manage the first nonvolatile memories; and a second flash translation layer configured to manage the second nonvolatile memories, the first flash translation layer outputs commands corresponding to the first and fourth requests through the first channel, and the second flash translation layer outputs commands respectively corresponding to the second and third requests through the second channel, and a value of the first logical address is smaller than a value of the second logical address, and a value of the third logical address is smaller than a value of the fourth logical address.

A memory device comprises a memory bank comprising a plurality of addressable memory cells, wherein the memory bank is divided into a plurality of segments. Further, the device comprises a cache memory operable for storing a second plurality of data words, wherein each data word of the second plurality of data words is either awaiting write verification associated with the memory bank or is to be re-written into the memory bank. The cache memory is divided into a plurality of primary segments, wherein each primary segment of the cache memory is direct mapped to a corresponding segment of the plurality of segments, wherein each primary segment is sub-divided into a plurality of secondary segments, and wherein each of the plurality of secondary segments comprises at least one counter for tracking a number of entries stored therein.

A method can include identifying a first key value of a first cell of a first grid of grids of cells to which a first feature maps, embedding the first grid into each cell of a second grid, identifying a second key value of a second cell of the second grid to which a second feature maps, the second key value representative of the first and second key values, comparing the identified key value to the key values of a memory, in response to determining the identified key value is in the memory, and providing data indicating a class associated with the identified key value in the memory.