Samsung Electronics Co., Ltd.

Ulsan National Institute of Science and Technology

Unist (ulsan National Institute of Science and Technology )

Hyeongseok Yu

Jongeun Lee

Hyeonuk Sim

Soojung Ryu

Yeongon Cho

Changmoo Kim

Sehwan Lee

Junwoo Jang

Donghoon Yoo

Yoojin Kim

Hyunsun Park

Yoonseo Choi

Haewoo Park

Choonki Jang

A processor-implemented method includes determining a first quantization value by performing log quantization on a parameter from one of input activation values and weight values in a layer of a neural network, comparing a threshold value with an error between a first dequantization value obtained by dequantization of the first quantization value and the parameter, determining a second quantization value by performing log quantization on the error in response to the error being greater than the threshold value as a result of the comparing; and quantizing the parameter to a value in which the first quantization value and the second quantization value are grouped.

Method and apparatus with neural network parameter quantization

A neural network device including an on-chip buffer memory that stores an input feature map of a first layer of a neural network, a computational circuit that receives the input feature map of the first layer through a single port of the on-chip buffer memory and performs a neural network operation on the input feature map of the first layer to output an output feature map of the first layer corresponding to the input feature map of the first layer, and a controller that transmits the output feature map of the first layer to the on-chip buffer memory through the single port to store the output feature map of the first layer and the input feature map of the first layer together in the on-chip buffer memory.

Method and device with neural network implementation

A method of processing data in a neural network, includes identifying a sparsity of input data, based on valid information included in the input data in which the input data includes valid values and invalid values, generate rearranged input data, based on a form of the sparsity by rearranging, in the input data, location of at least one of the valid values and the invalid values, and generating, by performing a convolution on the rearranged input data in the neural network, an output.

Method and apparatus for processing data

A neural network device includes: an on-chip buffer memory that stores an input feature map of a first layer of a neural network, a computational circuit that receives the input feature map of the first layer through a single port of the on-chip buffer memory and performs a neural network operation on the input feature map of the first layer to output an output feature map of the first layer corresponding to the input feature map of the first layer, and a controller that transmits the output feature map of the first layer to the on-chip buffer memory through the single port to store the output feature map of the first layer and the input feature map of the first layer together in the on-chip buffer memory.

A method of performing an arithmetic operation by a processing apparatus includes determining a polynomial expression approximating an arithmetic operation to be performed on a variable; adaptively determining upper bits for addressing a look-up table (LUT) according to a variable section to which the variable belongs; obtaining coefficients of the polynomial expression from the LUT by addressing the LUT using a value of the upper bits; and performing the arithmetic operation by calculating a result value of the polynomial expression using the coefficients.

Method and processing apparatus for performing arithmetic operation

A method of calculating data includes acquiring a difference between first data that is input and second data that was previously stored; determining a method of generating third data corresponding to a result of a calculation of the first data based on the difference; and performing a calculation corresponding to the determined method using a calculator.

Method and apparatus for calculating data

A power consumption control apparatus includes a resource selecting unit configured to select resources, whose power consumption levels are to be determined, from among resources of a graphic processing unit (GPU), a resource use information acquiring unit configured to determine whether the selected resources are used from a code block which is all or part of a program executed using the GPU, and a power consumption controlling unit configured to determine a power consumption level of the selected resource based on a determination result of the resource information acquiring unit and to control the power consumption level of the selected resources based on a determined power consumption level of the selected resources.

Apparatus and method of controlling power consumption of graphic processing unit (GPU) resources

Provided are a method of controlling a range of representable numbers includes receiving a floating point value represented by an exponent and a mantissa, each represented by a predetermined numbers of bits, determining a bit configuration of the exponent and the mantissa of the floating point value based on a value of a most significant bit of the exponent of the floating point value, and determining a constant required for calculation of the floating point value according to the determined bit configuration of the exponent, and an apparatus for providing such a method.

Growing community of inventors

Seoul, South Korea

Hyeongseok Yu