Sungkyunkwan University

Jaewook Jeon

Jaewan Park

Doyeon Kim

Junyoung Moon

Minhee Jo

Phuoc Tien Nguyen

Jinyoung Byun

Provided is a method for arranging tasks including sequentially allocating offsets, in unit of load measurement, to tasks having a period equal to or greater than a reference value, among the tasks, and sequentially allocating the offsets to tasks having a period smaller than the reference value, among the tasks, in ascending order, starting from a task having a shortest period, the sequentially allocating the offset comprising sequentially allocating a smallest offset to a first task having a shortest period among tasks having the period smaller than the reference value and allocating the offset to a second task, which is an allocation target of the offset, among remaining tasks excluding the first task, among the tasks having the period smaller than the reference value, based on a result of comparing the period of the first task and an execution time of the second task.

Task distribution method for minimizing preemption between tasks and apparatus for performing the same

According to the disclosure, a memory allocation method and device and recording medium in a multi-core processor system are disclosed. According to an embodiment, a method for allocating a shared variable to a memory in a multi-core processor system comprises mapping each task to a core, allocating unshared variables to memories times of access, to which are sequentially minimized, in descending order of actual variable access count, calculating an actual variable access count per core, selecting a core with a highest actual variable access count, and allocating a shared variable to a memory of the selected core.

Method and apparatus for memory allocation in a multi-core processor system, and recording medium therefor

Disclosed is a stereo matching method for generating a disparity map from a stereo image. The stereo matching method may include obtaining a cost volume by applying a first convolutional neural network (CNN) and a second CNN to a left image and a right image, respectively, wherein the cost volume is determined based on feature maps extracted from the left image and the right image, respectively, performing normalization on the cost volume by applying a third CNN, up-sampling the normalized cost volume, and obtaining a disparity map by applying regression analysis to the up-sampled cost volume.

Growing community of inventors

Suwon-si, South Korea

Jaewook Jeon