Samsung Electronics Co., Ltd.

Dongyan Jiang

Hongzhong Zheng

Qiang Peng

Dimin Niu

Krishna Teja Malladi

Jongmin Gim

Changhui Lin

Andrew Zhenwen Chang

Mu-Tien Chang

Mu Tien Chang

A transaction manager for use with memory is described. The transaction manager can include a write data buffer to store outstanding write requests, a read data multiplexer to select between data read from the memory and the write data buffer, a command queue and a priority queue to store requests for the memory, and a transaction table to track outstanding write requests, each write request associated with a state that is Invalid, Modified, or Forwarded.

Effective transaction table with page bitmap

An adaptive matrix multiplier. In some embodiments, the matrix multiplier includes a first multiplying unit a second multiplying unit, a memory load circuit, and an outer buffer circuit. The first multiplying unit includes a first inner buffer circuit and a second inner buffer circuit, and the second multiplying unit includes a first inner buffer circuit and a second inner buffer circuit. The memory load circuit is configured to load data from memory, in a single burst of a burst memory access mode, into the first inner buffer circuit of the first multiplying unit; and into the first inner buffer circuit of the second multiplying unit.

Adaptive matrix multiplication accelerator for machine learning and deep learning applications

A memory system provides deduplication of user data in the physical memory space of the system for user data that is duplicated in the virtual memory space of a host system. A transaction manager (TM) uses a transaction table to maintain data coherency and data concurrency for the virtual memory space. A write data engine manager (WDEM) uses an outstanding bucket number and command queues to maintain data coherency and data concurrency for the physical memory space. The WDEM receives data write requests from the TM and sends a corresponding write command to a selected command queue. A write data engine responds to a write command in a command queue by storing the data in an overflow memory region if the data is not duplicated in the virtual memory space, or by incrementing a reference counter for the data if the data is duplicated in the virtual memory space.

Scalable architecture enabling large memory system for in-memory computations

A memory system is disclosed. The memory system may include a Big Hash Table and a Little Hash Table. The memory system may also include an Overflow Region and a Translation Table to map a logical address to a Physical Line Identifier (PLID), which may include a region identifier and a physical address.

Method and apparatus for maximized dedupable memory

A deduplication memory system includes a virtual memory space, a physical memory space and a memory manager. The memory manager generates a user data entry that is stored in the physical memory space. The user data entry represents a unique user data of a predetermined granularity appearing in the virtual memory space, and includes first and second portions. The first portion includes information relating to a number of duplication times the unique user data corresponding to the user data entry is duplicated in the virtual memory space, and the second portion includes a selected part of the unique user data from which the unique user data may be reconstructed. The first portion may include an index to an extended reference counter table or a special data pattern table if the number of duplication times of the unique user data is greater than or equal to a predetermined number.

Embedded reference counter and special data pattern auto-detect

A method of storing data in a memory module including an in-module prefetcher, an in-module prefetch buffer, memory, and a memory controller, the method including sending address information from the in-module prefetcher to the memory controller and to the prefetch buffer, determining prefetch accuracy based on a comparison of the address information sent to the memory controller and the address information sent to the prefetch buffer, determining a prefetch mode based on the prefetch accuracy, and storing the data in the memory based on the prefetch mode.

Byte-addressable flash-based memory module with prefetch mode that is adjusted based on feedback from prefetch accuracy that is calculated by comparing first decoded address and second decoded address, where the first decoded address is sent to memory controller, and the second decoded address is sent to prefetch buffer

A dedupable cache is disclosed. The dedupable cache may include cache memory including both a dedupable read cache and a non-dedupable write buffer. The dedupable cache may also include a deduplication engine to manage reads from and writes to the dedupable read cache, and may return a write status signal indicating whether a write to the dedupable read cache was successful or not. The dedupable cache may also include a cache controller, which may include: a cache hit/miss check to determine whether an address in a request may be found in the dedupable read cache; a hit block to manage data accesses when the requested data may be found in the dedupable read cache; a miss block to manage data accesses when the requested data is not found in the dedupable read cache; and a history storage to store information about accesses to the data in the dedupable read cache.

Dedupe DRAM cache

A method of retrieving data stored in a memory associated with a dedupe module is provided. The method includes: identifying a logical address of the data; identifying a physical line ID of the data in accordance with the logical address by looking up at least a portion of the logical address in a translation table; locating a respective physical line, the respective physical line corresponding to the PLID; and retrieving the data from the respective physical line, the retrieving including copying a respective hash cylinder to the read cache, the respective hash cylinder including: a respective hash bucket, the respective hash bucket including the respective physical line; and a respective reference counter bucket, the respective reference counter bucket including a respective reference counter associated with the respective physical line.

Method and apparatus for enabling larger memory capacity than physical memory size

A dedupe module is provided. The dedupe module includes: a host interface; a dedupe engine to receive a data request from a host system via the host interface; a memory controller; a plurality of memory modules, each memory module being coupled to the memory controller; and a read cache for caching data from the memory controller for use by the dedupe engine.

Growing community of inventors

San Jose, CA, United States of America

Dongyan Jiang