-
1.发明公开
公开(公告)号：EP3220277A1
公开(公告)日：2017-09-20
申请号：EP15875107.3
申请日：2015-12-18
发明人： YANG, Renhua , ZHAO, Junfeng , YANG, Wei , WANG, Yuangang , LIN, Yinyin
IPC分类号： G06F13/28
CPC分类号： G06F12/0868 , G06F11/14 , G06F12/0246 , G06F12/0802 , G06F12/109 , G06F13/28 , G11C11/40615 , G11C11/40622 , Y02D10/13 , Y02D10/14 , Y02D10/151
摘要： Embodiments of the present invention provide a memory access method, a storage-class memory, and a computer system. The computer system includes a memory controller and a hybrid memory, and the hybrid memory includes a dynamic random access memory DRAM and a storage-class memory SCM. The memory controller is configured to send a first access instruction to the DRAM and the SCM. When determining that a first memory cell set that is of the DRAM and to which a first address in the received first access instruction points includes a memory cell whose retention time is shorter than a refresh cycle of the DRAM, the SCM may obtain a second address having a mapping relationship with the first address. Further, the SCM converts, according to the second address, the first access instruction into a second access instruction for accessing the SCM, to implement access to the SCM. The computer system provided in the embodiments of the present invention can ensure correctness of data while reducing power consumption for DRAM refresh.
摘要翻译： 本发明的实施例提供一种存储器存取方法，存储级存储器和计算机系统。 计算机系统包括存储器控制器和混合存储器，并且混合存储器包括动态随机存取存储器DRAM和存储级存储器SCM。 存储器控制器被配置为向DRAM和SCM发送第一访问指令。 当确定DRAM的第一存储单元组和第一存取指令点中的第一地址包括保持时间短于DRAM的刷新周期的存储单元时，SCM可以获得第二地址 与第一地址具有映射关系。 进一步地，SCM根据第二地址将第一访问指令转换为访问SCM的第二访问指令，以实现对SCM的访问。 本发明实施例提供的计算机系统可以保证数据的正确性，同时降低DRAM刷新的功耗。
-
公开(公告)号：EP3848855A1
公开(公告)日：2021-07-14
申请号：EP18934368.4
申请日：2018-09-19
发明人： ZHANG, Qing , YANG, Wei , XIAO, Yifan , ZHANG, Lianghe , JUI, Shangling
摘要： Embodiments of this application provide an intention identification model learning method and an apparatus, and a device, and relates to the field of communications technologies, to help improve accuracy of the intention identification model in a human-machine dialog system, improve accuracy of task execution in the human-machine dialog system, and improve user experience. The method includes: receiving, by a server, positive data (S101) that corresponds to a first skill and that is entered by a skill developer; generating, by the server based on the positive data that corresponds to the first skill, negative data (S102) that corresponds to the first skill; determining, by the server, a second skill (S103) similar to the first skill; obtaining, by the server, data (S104) that corresponds to each second skill; generating, by the server, a second base model (S105) based on the data that corresponds to the second skill and a first base model stored on the server; and performing learning (S 106), by the server, based on the second base model and the positive data and the negative data that correspond to the first skill, and generating an intention identification model.
-
公开(公告)号：EP3822798A2
公开(公告)日：2021-05-19
申请号：EP20195180.3
申请日：2017-06-23
发明人： XIAO, Shihai , LONGNOS, Florian , YANG, Wei
IPC分类号： G06F13/16
摘要： This application provides a memory access technology and a computer system. The computer system includes a memory controller, a media controller, and a non-volatile memory NVM connected to the media controller. After receiving a first read command that is sent by the memory controller, the media controller may read first data from the NVM based on a first address in the first read command. Then the media controller can return, to the memory controller, at least two fixed-length data subblocks and metadata of the at least two data subblocks based on at least two send commands sent by the memory controller. The metadata includes a location identifier that is used to indicate an offset of a corresponding data subblock in the first data. The memory controller combines the at least two data subblocks into the first data based on location identifiers in the metadata of the at least two data subblocks.
-
公开(公告)号：EP3389051B1
公开(公告)日：2020-09-09
申请号：EP16885510.4
申请日：2016-01-18
发明人： YU, Hao , WANG, Yuhao , ZHAO, Junfeng , YANG, Wei , XIAO, Shihai , NI, Leibin
-
公开(公告)号：EP3040994B1
公开(公告)日：2018-03-07
申请号：EP14847885.2
申请日：2014-09-24
发明人： LIN, Yinyin , FU, Yarong , YANG, Kai , YANG, Wei , WANG, Yuangang , ZHAO, Junfeng
CPC分类号： G11B5/09 , G11B5/02 , G11B2005/0002 , G11C11/02 , G11C11/15 , G11C11/161 , G11C11/1673 , G11C11/1675 , G11C11/5607 , G11C19/0808 , G11C19/0841
摘要： Embodiments of the present invention provide an information storage apparatus and method, and the apparatus includes: a magnetic track, where the magnetic track consists of multiple magnetic domains, and each magnetic domain is divided into at least two magnetic regions; a writing unit, disposed on the magnetic track, where information is written to the at least two magnetic regions of each magnetic domain by the writing unit; and a reading unit, disposed on the magnetic track, where the information written to the at least two magnetic regions is read by the reading unit; therefore, multiple pieces of valid information are written to one magnetic domain of the magnetic track, which increases storage density of the magnetic track, and expands a storage capacity of the storage apparatus.
-
6.发明公开IMAGE RECOGNITION ACCELERATOR, TERMINAL DEVICE AND IMAGE RECOGNITION METHOD 审中-公开图像识别加速器，终端装置和图像识别方法
公开(公告)号：EP3244345A1
公开(公告)日：2017-11-15
申请号：EP16761030.2
申请日：2016-02-22
发明人： YU, Hao , WANG, Yuhao , NI, Leibin , YANG, Wei , ZHAO, Junfeng , XIAO, Shihai
IPC分类号： G06K9/00
CPC分类号： G06K9/00973 , G06K9/00 , G06K9/6201
摘要： This application discloses an image recognition accelerator, a terminal device, and an image recognition method. The image recognition accelerator includes a dimensionality-reduction processing module, an NVM, and an image matching module. During an image recognition process of the image recognition accelerator, the dimensionality-reduction processing module first reduces a dimensionality of first image data according to a specified dimensionality-reduction parameter γ. The NVM writes, into a first storage area of the NVM according to a specified first current I, ω low-order bits of each numeric value of the first image data on which dimensionality reduction has been performed, and writes, into a second storage area of the NVM according to a specified second current, (N-ω) high-order bits of each numeric value of the first image data on which dimensionality reduction has been performed. The first current is lower than the second current. In this way, the matching module may determine whether an image library stored in the NVM includes image data matching the first image data on which dimensionality reduction has been performed. The image recognition accelerator provided in embodiments of the present invention can ensure accuracy of image recognition while reducing a system power consumption of the terminal device.
摘要翻译： 该申请公开了图像识别加速器，终端装置和图像识别方法。 图像识别加速器包括维度降低处理模块，NVM和图像匹配模块。 在图像识别加速器的图像识别过程中，维数降低处理模块首先根据指定的维数降低参数γ降低第一图像数据的维数。 NVM根据指定的第一电流I，将已经执行了降维的第一图像数据的每个数值的ω个低位位写入NVM的第一存储区域，并写入第二存储区域 根据指定的第二电流确定已经执行了降维的第一图像数据的每个数值的（N-ω）高阶位。 第一个电流低于第二个电流。 以这种方式，匹配模块可以确定存储在NVM中的图像库是否包括与已经执行了降维的第一图像数据相匹配的图像数据。 本发明实施例提供的图像识别加速器可以保证图像识别的准确性，同时降低终端设备的系统功耗。
-
公开(公告)号：EP3136390A1
公开(公告)日：2017-03-01
申请号：EP15783125.6
申请日：2015-04-22
发明人： FU, Yarong , ZHAO, Junfeng , WANG, Yuangang , YANG, Wei , LIN, Yinyin , YANG, Kai
IPC分类号： G11B5/012
CPC分类号： G11C11/1675 , G11B5/012 , G11B5/653 , G11B5/656 , G11C11/16 , G11C19/0841
摘要： Embodiments of the present invention provide a write apparatus and a magnetic memory, where the write apparatus includes: a first drive port, a second drive port, a first information storage area, a second information storage area, and an information buffer, where there is a first area between the first information storage area and the information buffer, there is a second area between the second information storage area and the information buffer, the first information storage area, the second information storage area, and the information buffer are made of a first magnetic material, the first area and the second area are made of a second magnetic material, and magnetic energy of the first magnetic material is higher than magnetic energy of the second magnetic material; the first information storage area is configured to write first data to the information buffer; the second information storage area is configured to write second data to the information buffer; and the information buffer is configured to buffer data written from the first information storage area or the second information storage area, and write the buffered data to a magnetic domain of the magnetic memory, which can ensure write stability of the magnetic memory.
摘要翻译： 本发明的实施例提供一种写入装置和磁性存储器，其中写入装置包括：第一驱动端口，第二驱动端口，第一信息存储区域，第二信息存储区域和信息缓冲区，其中存在 在第一信息存储区域和信息缓冲区之间的第一区域中，在第二信息存储区域和信息缓冲区之间存在第二区域，第一信息存储区域，第二信息存储区域和信息缓冲区由 第一磁性材料，第一区域和第二区域由第二磁性材料制成，并且第一磁性材料的磁能高于第二磁性材料的磁能; 所述第一信息存储区域被配置为将第一数据写入所述信息缓冲器; 第二信息存储区域被配置为将第二数据写入信息缓冲器; 信息缓冲器用于缓存从第一信息存储区或第二信息存储区写入的数据，并将缓存的数据写入磁存储器的磁畴，可以保证磁存储器的写入稳定性。
-
公开(公告)号：EP3089043A1
公开(公告)日：2016-11-02
申请号：EP14891927.7
申请日：2014-05-16
发明人： CHEN, Shaojie , WEI, Zhulin , ZHAO, Junfeng , HE, Rui , DONG, Xiaowen , LIN, Peng , YANG, Wei
IPC分类号： G06F13/14
CPC分类号： H01Q1/2258 , G06F13/4063 , H01Q1/2291 , H04W84/10
摘要： A cabinet server and a data center based on the cabinet server are disclosed. The cabinet server includes at least two function nodes, multiple intra-cabinet antennas, and multiple inter-cabinet antennas. The function nodes are vertically arranged, to form a server core; and the intra-cabinet antennas are vertically arranged, disposed at a side of the server core, and electrically connected to corresponding function nodes, and adjacent intra-cabinet antennas are wirelessly connected. When a radio signal is transmitted within the cabinet server, a transmission path is formed of the vertically arranged intra-cabinet antennas. Because the intra-cabinet antennas are disposed at the side of the server core, electromagnetic radiation generated by the radio signal in a transmission process has a relatively small effect on the function nodes, thereby reducing the effect of the electromagnetic radiation on various electronic devices in the function nodes, improving service lives of the electronic devices, and improving transmission quality of the radio signal.
-
9.发明公开METHOD AND DEVICE FOR WRITING DATA IN STORAGE MEDIUM 审中-公开VERFAHREN UND VORRICHTUNG ZUM SCHREIBEN VON DATEN IN EINEM SPEICHERMEDIUM
公开(公告)号：EP3076282A1
公开(公告)日：2016-10-05
申请号：EP14876033.3
申请日：2014-12-22
发明人： XU, Ronggang , XU, Jun , YANG, Wei
IPC分类号： G06F3/06
CPC分类号： G06F3/061 , G06F3/0653 , G06F3/0659 , G06F3/0673 , G06F11/3034 , G06F11/3452 , G06F11/3485 , G06F11/3495 , G06F2201/88
摘要： The present invention relates to a method and apparatus for writing data to a storage medium, where the method for writing data to a storage medium includes: writing data to a storage unit of a storage medium according to a write cycle of the storage medium, where the storage unit is a read/write unit of the storage medium; after write operations are concurrently executed on n storage units in a first write cycle, counting a quantity k of storage units in which writing succeeds among the n storage units in the first write cycle; and starting, in a second write cycle, write operations on subsequent data whose data amount is m storage units in addition to performing write operations on n-k storage units in which writing is not successful. Statistics about a case in which writing to storage units in the first write cycle succeeds are collected, and when write operations on the k storage units are completed in the first write cycle, write operations on m subsequent storage units are started in the second write cycle, which can improve efficiency in writing data to the storage medium.
摘要翻译： 本发明涉及将数据写入存储介质的方法和装置，其中将数据写入存储介质的方法包括：根据存储介质的写周期将数据写入存储介质的存储单元，其中 存储单元是存储介质的读/写单元; 在第一写入周期中对n个存储单元并行执行写入操作之后，对第一写入周期中的n个存储单元中的写入成功的存储单元的数量进行计数; 并且在第二写入周期中，除了对写入不成功的n-k个存储单元执行写入操作之外，对数据量为m个存储单元的后续数据进行写操作。 收集关于在第一写入周期中写入存储单元成功的情况的统计信息，并且当在第一写周期中完成对k个存储单元的写入操作时，在第二写入周期中开始对m个后续存储单元的写入操作 ，这可以提高将数据写入存储介质的效率。
-
公开(公告)号：EP3051536A1
公开(公告)日：2016-08-03
申请号：EP14856096.4
申请日：2014-10-15
发明人： LIN, Yinyin , WEI, Zhulin , ZHAO, Junfeng , YANG, Wei , FU, Yarong , YANG, Kai
IPC分类号： G11C11/02
CPC分类号： G06F3/0625 , G06F3/0658 , G06F3/0676 , G11C8/16 , G11C11/161 , G11C11/1675 , G11C19/0808 , G11C19/0841
摘要： Embodiments of the present invention provide a storage unit, a memory, and a method for controlling a storage unit, to increase storage density and reduce power consumption. The storage unit includes: a U-shaped magnetic track, a first drive circuit, a second drive circuit, a first drive port, and a second drive port, where the U-shaped magnetic track includes a first port, a second port, a first storage area, and a second storage area; the first drive circuit is configured to drive the first storage area, and the second drive circuit is configured to drive the second storage area; by controlling input voltages of the first port, the second port, the first drive port, and the second drive port and driving the first drive circuit, a current pulse is generated in the first storage area, and a magnetic domain in the first storage area is driven to move; and by controlling the input voltages of the first port, the second port, the first drive port, and the second drive port and driving the second drive circuit, a current pulse is generated in the second storage area, and a magnetic domain in the second storage area is driven to move.
摘要翻译： 本发明的实施例提供一种存储单元，存储器以及用于控制存储单元的方法，以增加存储密度并降低功耗。 所述存储单元包括：U形磁道，第一驱动电路，第二驱动电路，第一驱动端口和第二驱动端口，所述U形磁道包括第一端口，第二端口， 第一存储区域和第二存储区域; 所述第一驱动电路用于驱动所述第一存储区域，所述第二驱动电路用于驱动所述第二存储区域; 通过控制第一端口，第二端口，第一驱动端口和第二驱动端口的输入电压并驱动第一驱动电路，在第一存储区域中产生电流脉冲，并且第一存储区域中的磁畴 被驱动移动; 并且通过控制第一端口，第二端口，第一驱动端口和第二驱动端口的输入电压并驱动第二驱动电路，在第二存储区域中产生电流脉冲，并且在第二存储区域中产生磁畴 存储区域被驱动移动。
-
-
-
-
-
-
-
-
-
搜索结果
35 条记录 (耗时 0.028 秒)
