Depth-first convolution in deep neural networks

    公开(公告)号:US11487998B2

    公开(公告)日:2022-11-01

    申请号:US16443695

    申请日:2019-06-17

    Abstract: In one embodiment, a depth-first deep convolutional network (DCN) having a first convolutional layer having a first first-layer kernel and adapted to convolve a first input and a second convolutional layer having a first second-layer kernel and adapted to convolve a second-layer input. A method for the DCN includes initiating convolution in the first convolution layer of the first input tensor with the first first-layer kernel to generate a value strip for the second input tensor and, prior to completion of the convolution in the first convolution layer, initiating convolution in the second convolution layer of the second input with the first second-layer kernel to generate a value strip for a third layer.

    System, apparatus, and method for decompressing data
    3.
    发明授权
    System, apparatus, and method for decompressing data 有权
    用于解压缩数据的系统,装置和方法

    公开(公告)号:US09413386B1

    公开(公告)日:2016-08-09

    申请号:US14626905

    申请日:2015-02-19

    CPC classification number: H03M7/3088 H03M7/30 H03M7/3086 H03M7/6005

    Abstract: A system for data decompression may include a processor coupled to a remote memory having a remote dictionary stored thereon and coupled to a decompression logic having a local memory with a local dictionary. The processor may decompress data during execution by accessing the local dictionary, and if necessary, the remote dictionary.

    Abstract translation: 用于数据解压缩的系统可以包括耦合到具有存储在其上的远程字典的远程存储器的处理器,并且耦合到具有本地字典的本地存储器的解压缩逻辑。 处理器可以在执行期间通过访问本地字典来解压缩数据,并且如果需要,可以对远程字典进行解压缩。

    System and method for dictionary-based cache-line level code compression for on-chip memories using gradual bit removal
    4.
    发明授权
    System and method for dictionary-based cache-line level code compression for on-chip memories using gradual bit removal 有权
    用于基于字典的缓存线级代码压缩的系统和方法,用于使用逐位删除的片上存储器

    公开(公告)号:US09300320B2

    公开(公告)日:2016-03-29

    申请号:US14318564

    申请日:2014-06-27

    Abstract: A multi-pass compression iteratively removes combinations of bits from locations in each word of a cache line of an uncompressed data stream. For each combination of removed bits, the remaining bits in the word values of the cache line are analyzed to generate a compression score. A highest compression score triggers the building of a dictionary from the remaining bits in the word values of the cache line. After a dictionary is built, the method may continue iteratively to create subsequent dictionaries from the words that remain uncompressed in the cache line. To decompress a word, a first bit section of the compressed word is used to identify a dictionary that is then queried for bits indexed in a second bit section of the compressed word. The uncompressed word is reconstructed by interleaving the queried bits with the removed combination of bits from a third bit section of the word.

    Abstract translation: 多遍压缩迭代地从未压缩的数据流的高速缓存线的每个字中的位置中去除位的组合。 对于删除位的每个组合,分析高速缓存行的字值中的剩余位以产生压缩分数。 最高的压缩分数触发从高速缓存行的单词值中的其余位构建字典。 在构建字典之后,该方法可以继续迭代地从在高速缓存行中保持未压缩的单词创建后续字典。 为了对字进行解压缩,使用压缩字的第一比特部分来识别字典,然后在压缩字的第二比特部分中对索引的比特进行查询。 通过将查询的比特与从单词的第三比特部分中去除的比特组合进行交织来重构未压缩的单词。

Patent Agency Ranking