-
公开(公告)号：KR101533820B1
公开(公告)日：2015-07-09
申请号：KR1020090091306
申请日：2009-09-25
Applicant: 삼성전자주식회사
Abstract: 멀티프로세싱에서의메모리관리장치및 그방법이개시된다. 본발명의일실시예에따른메모리관리장치는정보를생성하는정보생성프로세스와상기생성된정보를소비하는정보소비프로세스가서로공유하며상기공유하는정보를저장하는공유큐의상태에기초하여정보소비프로세스에할당된프로세서의수나정보할당단위를조정한다. 이에따라, 정보소비프로세스가사용하는작업큐의접근을적절히제한함으로써, 멀티코어나멀티프로세서환경에서멀티프로세싱을효율적으로수행할수 있다.
-
公开(公告)号：KR1020110089748A
公开(公告)日：2011-08-09
申请号：KR1020100009283
申请日：2010-02-01
Applicant: 삼성전자주식회사
CPC classification number: G06F9/50 , G06F9/4893 , G06F2209/5019 , G06F9/5061 , G06F9/468 , G06F9/5083
Abstract: PURPOSE: An apparatus and method for processing data are provided to determine the number of processes in which data is allocated and allocate the data to the determine processor. CONSTITUTION: An estimator part predicts the amount of data work. A processor(150) processes the data by the estimated data work amount. The estimation part predicts the process result by the number of processors. A determination part(120) determines the number of processors in which data will be allocated by the predicted processing result.
Abstract translation: 目的：提供一种处理数据的装置和方法，以确定数据分配的处理次数，并将数据分配给确定处理器。 构成：估计部分预测数据的工作量。 处理器（150）通过估计的数据工作量处理数据。 估计部分通过处理器数量预测过程结果。 确定部分（120）根据预测的处理结果确定数据将被分配的处理器的数量。
-
公开(公告)号：KR1020070003502A
公开(公告)日：2007-01-05
申请号：KR1020050082858
申请日：2005-09-06
Applicant: 삼성전자주식회사
Inventor： 백현기
IPC: H04N19/895
CPC classification number: H04N19/895 , H04N19/142 , H04N19/176
Abstract: A method and an apparatus for concealing an error are provided to perform the error concealing work for each frame effectively when the loss of frame occurs in an apparatus for receiving and outputting a compressed moving image. A reference direction of a loss frame is determined through switching of the scene between frames based on the loss frame(S408). Concealing reference points for pixels constituting the loss frame are extracted by using one or more frames except the loss frame according to the determined reference direction(S410,S412,S414). Concealing pixel values for the pixels constituting the loss frame are obtained by providing the weights to the extracted concealing reference points(S411,S413,S415). The loss frame is restored by performing the concealing process for the pixels using the obtained concealing pixel values(S416).
Abstract translation: 提供一种用于隐藏错误的方法和装置，用于当在用于接收和输出压缩运动图像的装置中发生帧丢失时，有效地执行每帧的错误隐藏工作。 通过基于丢失帧通过帧之间的场景切换来确定丢失帧的参考方向（S408）。 通过根据确定的参考方向（S410，S412，S414）使用除了丢失帧之外的一个或多个帧来提取构成丢失帧的像素的隐藏参考点。 通过对提取的隐藏参考点提供权重来获得构成丢失帧的像素的隐藏像素值（S411，S413，S415）。 通过使用所获得的隐藏像素值执行对像素的隐藏处理来恢复丢失帧（S416）。
-
公开(公告)号：KR100195178B1
公开(公告)日：1999-06-15
申请号：KR1019920027557
申请日：1992-12-31
Applicant: 삼성전자주식회사
IPC: G06F7/52
CPC classification number: G06F7/552 , G06F2207/5523
Abstract: 본 발명은 제곱 계산회로를 제공하는데 있다. 2비트에서 6비트까지의 입력데이타를 입력하여 각각 4비트에서 12비트까지의 데이타를 출력하는 제곱 계산회로를 구성하기 위하여 입력에 대한 출력의 관계식을 구하고 그 식을 최대한 간략화하여 각 해당하는 비트의 제곱 계산회로를 구성한 것이다. 따라서, 종래의 곱셈기를 이용하여 제곱을 계산하는 경우보다 회로 구성이 간단화되고 처리속도가 빨라지게 된다.
-
公开(公告)号：KR101355375B1
公开(公告)日：2014-01-22
申请号：KR1020070126891
申请日：2007-12-07
Applicant: 삼성전자주식회사
Abstract: 본 발명의 멀티미디어 복호화 방법은, 중앙 프로세서 및 복수 개의 연산 프로세서를 포함하는 멀티코어 플랫폼 기반의 멀티미디어 복호화 방법에 있어서, 입력된 멀티미디어 데이터에 대해, 중앙 프로세서에서 중앙 프로세서 및 연산 프로세서가 수행할 적어도 하나 이상의 연산 작업에 대한 큐를 생성하고, 연산 프로세서 중 적어도 하나 이상의 일부 연산 프로세서에서 멀티미디어 데이터가 분할된 데이터 영역 별로 움직임 보상 작업을 수행하며, 연산 프로세서 중 하나의 연산 프로세서가 멀티미디어 데이터에 대한 디블로킹 작업을 수행함으로써, 멀티코어 플랫폼 기반에서 데이터량이 큰 멀티미디어 데이터를 분할하여 멀티프로세서의 자원을 효율적으로 활용한다.
멀티코어 플랫폼, 멀티프로세서, 멀티미디어 복호화, 데이터 분할, 기능적 분할-
Patent Agency Ranking
公开(公告)号：KR1020120069364A
公开(公告)日：2012-06-28
申请号：KR1020100130880
申请日：2010-12-20
Applicant: 삼성전자주식회사
CPC classification number: G09G5/363 , G09G5/393 , G09G2330/021 , G09G2354/00 , G09G2360/06 , G09G2360/08 , G09G2360/16 , G06F9/5094 , Y02D10/14 , Y02D10/151 , Y02D10/22
Abstract: PURPOSE: A frame processing apparatus considering of processing power for cores in multicore environment and method thereof are provided to control user request response speed by measuring drawing loads of an input frame. CONSTITUTION: A load calculation unit(310) calculates system loads for processing operations which constructs frames displayed on an UI(User Interface) by referring a display area and features of the operations. A core selecting unit(320) selects a core which including low power consumption by considering of the ability/power processing rates of the core. A storage unit(300) successively stores the operations which constructs the frames.
Abstract translation: 目的：提供考虑多核环境中的核心处理能力的帧处理装置及其方法，通过测量输入帧的绘制负载来控制用户请求响应速度。 构成：负载计算单元（310）通过参考操作的显示区域和特征来计算构造在UI（用户界面）上显示的帧的处理操作的系统负载。 核心选择单元（320）通过考虑核心的能力/功率处理速率来选择包括低功耗的核心。 存储单元（300）连续地存储构成帧的操作。
-
公开(公告)号：KR1020120055353A
公开(公告)日：2012-05-31
申请号：KR1020100117050
申请日：2010-11-23
Applicant: 삼성전자주식회사
CPC classification number: G06F9/4881 , G06F2209/483 , G06F9/50
Abstract: PURPOSE: A device for optimizing data under different multi processors and a method thereof are provided to eliminate a problem that different processors are in a standby state due to operation dependability while a specific process performs an operation under different multi processors. CONSTITUTION: An analysis unit(110) analyzes operations for data processing and the types and number of available processors. A partition unit(120) dynamically partitions data into data areas whose sizes are different using operations analyzed by the analysis unit and processor preference information of each operation. A scheduling unit(130) allocates operations to each available processor analyzed by the analysis unit and schedules the operations wherein the operations are performed about each data area partitioned by the partition unit.
Abstract translation: 目的：提供一种用于在不同多处理器下优化数据的设备及其方法，以消除由于操作可靠性而使不同处理器处于待机状态的问题，而特定处理在不同的多处理器下执行操作。 构成：分析单元（110）分析数据处理的操作以及可用处理器的类型和数量。 分区单元（120）使用由分析单元分析的操作和每个操作的处理器偏好信息来动态地将数据分割成尺寸不同的数据区域。 调度单元（130）向由分析单元分析的每个可用处理器分配操作，并调度对由分区单元划分的每个数据区域执行操作的操作。
-
公开(公告)号：KR1020080061744A
公开(公告)日：2008-07-03
申请号：KR1020060136801
申请日：2006-12-28
Applicant: 삼성전자주식회사
IPC: H04N19/103 , H04N19/577
CPC classification number: H04N19/105 , H04N19/114 , H04N19/146 , H04N19/172 , H04N19/577
Abstract: A method and an apparatus for coding and decoding images are provided to improve a coding efficiency by minimizing a prediction error between a predicted image and a coded image according to image characteristics. A sub-group generating unit(210) divides a picture group including a plurality of image pictures provided in an image sequence based on a successive bi-directional picture to generate subgroups. A correlation calculating unit(220) calculates correlation between a reference picture and the bi-directional pictures by a plurality of coding modes which are classified according to whether the bi-directional picture can be used as a reference picture and according to the reference picture referred to by the bi-directional pictures. A coding mode selecting unit(230) selects one of coding modes by using the correlation. A coding unit(240) performs coding by subgroups according to the selected coding mode.
Abstract translation: 提供了一种用于对图像进行编码和解码的方法和装置，以通过根据图像特性使预测图像和编码图像之间的预测误差最小化来提高编码效率。 子组生成单元（210）基于连续的双向图像来分割包括在图像序列中提供的多个图像图像的图像组，以生成子组。 相关计算单元（220）通过多个编码模式来计算参考图像和双向图像之间的相关性，所述多个编码模式根据双向图像是否可以用作参考图像并根据参考图像被分类 通过双向图片。 编码模式选择单元（230）通过使用相关性来选择编码模式之一。 编码单元（240）根据选择的编码模式进行子组编码。
-
公开(公告)号：KR1020060130488A
公开(公告)日：2006-12-19
申请号：KR1020060035906
申请日：2006-04-20
Applicant: 삼성전자주식회사
IPC: H04N19/137 , H04N19/51
CPC classification number: H04N21/47202 , H04N19/102 , H04N19/105 , H04N19/137 , H04N19/139 , H04N19/176 , H04N19/523 , H04N21/2318 , H04N21/2343 , H04N21/23602 , H04N21/2387 , H04N21/2662
Abstract: A video signal encoding apparatus for increasing the compression efficiency through the model switching when performing a sub pixel motion estimation and a method thereof are provided to encode a video signal by the selection of a model among plural models. A motion vector of a reference pixel is calculated first(S310). When calculating a motion vector of a half pixel, a motion vector of an integer pixel as the reference at the circumference is calculated. An estimation model is produced with x and y values as the motion vector calculation result(S320). After producing two or more models, a motion vector of a sub pixel is estimated. A process for comparing the calculated value with an error of a sub pixel is performed(S330). The motion estimation is performed by the application of the model with a small error(S350).
Abstract translation: 提供一种用于在执行子像素运动估计时通过模型切换来提高压缩效率的视频信号编码装置及其方法，用于通过在多个模型中选择模型对视频信号进行编码。 首先计算参考像素的运动矢量（S310）。 计算半像素的运动矢量时，计算出整数像素的运动矢量作为圆周的基准。 产生具有x和y值作为运动矢量计算结果的估计模型（S320）。 在产生两个或更多个模型之后，估计子像素的运动矢量。 执行用于将计算值与子像素的误差进行比较的处理（S330）。 通过应用具有小误差的模型来执行运动估计（S350）。
-
公开(公告)号：KR101703328B1
公开(公告)日：2017-02-07
申请号：KR1020100117050
申请日：2010-11-23
Applicant: 삼성전자주식회사
CPC classification number: G06F9/4881 , G06F2209/483
Abstract: 이종멀티프로세서환경에서의데이터처리최적화기술을개시한다. 이종멀티프로세서환경에서의데이터처리최적화장치를통해처리성능이상이한이종의프로세서의특성을고려해데이터를다수의상이한크기로분할하여가용한프로세서별로연산을처리함으로써연산의존성을최소화할수 있으므로, 데이터처리성능을향상시킬수 있다.
Abstract translation: 提供了一种用于异构多处理器环境中的数据处理的装置。 该装置包括：分析单元，被配置为分析1）与数据处理相关的运行操作; 2）可用于数据处理的类型和多个处理器;分配单元，被配置为动态地将数据分割成多个数据区， 基于在运行操作之前存储的分析操作和操作特定处理器优先级信息的不同大小，以及调度单元，被配置为通过分配要在可用处理器之间的数据区域中运行的操作来执行调度。
-
-
-
-
-
-
-
-
-
Search Results
30
records
(took 0.023 seconds)
