Samsung Electronics Co., Ltd.

Heonsoo Lee

Junseok Park

Jaechul Park

Hong-mook Choi

Hyukjae Jang

Kyungah Jeong

Sungho Jun

Sangmin Suh

Byungchul Hong

Jaehun Chung

Sang-Hyun Park

Ihor Vasyltsov

Bohdan Karpinskyy

Jonghoon Shin

Yunhyeok Choi

Yun-Ho Youm

Mijung Noh

Youngjin Chung

In some embodiments, a multi-port queueing cache includes a plurality of first ports, a plurality of second ports, a plurality of request handlers respectively coupled to the plurality of first ports, a cache storage unit coupled to the plurality of second ports, a reserve interface configured to exchange at least one address and at least one reserved cache line number, and a request interface configured to exchange the at least one reserved cache line number and at least one data. The reserve interface and the request interface are disposed between the plurality of request handlers and the cache storage unit. The cache storage unit includes a plurality of cache lines configured to store the plurality of data. The cache storage unit is configured to output a portion of the plurality of addresses, and receive a portion of the plurality of data corresponding to the portion of the plurality of addresses.

Multi-port queueing cache and data processing system including the same

A device configured to compress a tensor including a plurality of cells includes: a quadtree generator configured to generate a quadtree searching for a non-zero cell included in the tensor and extract at least one parameter value from the quadtree; a mode selector configured to determine a compression mode based on the at least one parameter; and a bitstream generator configured to generate a bitstream by compressing the tensor based on the compression mode.

Compressing device and method using parameters of quadtree method

Provided is a neural network device including at least one processor configured to implement an arithmetic circuit configured to generate third data including a plurality of pixels based on a neural network configured to perform an arithmetic operation on first data and second data, and a compressor configured to generate compressed data by compressing the third data, wherein the compressor is further configured to generate, as the compressed data, bitmap data comprising location information about a non-zero pixel having a non-zero data value among the plurality of pixels based on a quad-tree structure.

Neural network processor for compressing featuremap data and computing system including the same

A path for transmitting encrypted data is completely separated from a path for transmitting unencrypted data. To this end, a virtual secure memory is created on an address space. If a central processing unit (CPU) writes data in the virtual secure memory, hardware stores the data in a specific area of a dynamic random access memory (DRAM) after automatically encrypting the data. In the case where the CPU intents to read data, the hardware sends the data to the CPU after automatically decrypting the data read from a specific area of the DRAM.

Computing system having an on-the-fly encryptor and an operating method thereof

A method for encryption, decryption, or encryption and decryption of data in a crypto device having at least one crypto core may include: generating a tweak value corresponding to block data, which is placed at a random position from which the encryption, decryption, or encryption and decryption starts, from among sequential block data; and/or performing the encryption, decryption, or encryption and decryption from the block data using the tweak value. A method for encryption, decryption, or encryption and decryption of block data may include: generating a tweak value corresponding to the block data at a random position; and/or performing the encryption, decryption, or encryption and decryption of the block data using the tweak value.

Crypto devices, storage devices having the same, and encryption and decryption methods thereof

A random number generator includes a first oscillator configured to output a first oscillating signal having a first frequency. A second oscillator is configured to output a second oscillating signal having a second frequency different from the first frequency. A sampling unit is configured to receive the first and second oscillating signals. The sampling unit is configured to generate at least one entropy source by combining the received first and second oscillating signals. The sampling unit is configured to generate a random bit corresponding to the generated entropy source using a third oscillating signal. A third oscillator & control unit is configured to control the first and second oscillators and to generate the third oscillating signal. A frequency of the third oscillating signal is lower than the first and second frequencies.

Random number generator

An access control method of a semiconductor device includes providing an inputted password as an input of a hash operator; performing a hash operation in the hash operator and outputting a first hash value; controlling the hash operator so that the hash operation is repeatedly performed in the hash operator by providing the first hash value as an input of the hash operator when the first hash value and a second hash value stored in a nonvolatile memory do not coincide; and setting an access level with respect to the inner circuit according to the repetition number of times of the hash operation of the hash operator when the first and second hash values coincide.

Growing community of inventors

Hwaseong-si, South Korea

Heonsoo Lee