Prefetching with multiple processors and threads via a coherency bus
    1.
    发明授权
    Prefetching with multiple processors and threads via a coherency bus 失效
    通过一个一致性总线预取多个处理器和线程

    公开(公告)号:US08543767B2

    公开(公告)日:2013-09-24

    申请号:US13488215

    申请日:2012-06-04

    IPC分类号: G06F13/00

    摘要: A processing system includes a memory and a first core configured to process applications. The first core includes a first cache. The processing system includes a mechanism configured to capture a sequence of addresses of the application that miss the first cache in the first core and to place the sequence of addresses in a storage array; and a second core configured to process at least one software algorithm. The at least one software algorithm utilizes the sequence of addresses from the storage array to generate a sequence of prefetch addresses. The second core issues prefetch requests for the sequence of the prefetch addresses to the memory to obtain prefetched data and the prefetched data is provided to the first core if requested.

    摘要翻译: 处理系统包括被配置为处理应用的存储器和第一核心。 第一个核心包括第一个缓存。 处理系统包括被配置为捕获错过第一核心中的第一高速缓存的应用程序的地址序列并将地址序列放置在存储阵列中的机制; 以及被配置为处理至少一个软件算法的第二核心。 所述至少一个软件算法利用来自存储阵列的地址序列来生成预取地址序列。 第二个核心将预取地址序列的预取请求发送到存储器以获得预取数据,并且如果请求,则将预取数据提供给第一核。

    Effective prefetching with multiple processors and threads
    2.
    发明授权
    Effective prefetching with multiple processors and threads 失效
    有效的预取与多个处理器和线程

    公开(公告)号:US08200905B2

    公开(公告)日:2012-06-12

    申请号:US12192072

    申请日:2008-08-14

    IPC分类号: G06F13/00

    摘要: A processing system includes a memory and a first core configured to process applications. The first core includes a first cache. The processing system includes a mechanism configured to capture a sequence of addresses of the application that miss the first cache in the first core and to place the sequence of addresses in a storage array; and a second core configured to process at least one software algorithm. The at least one software algorithm utilizes the sequence of addresses from the storage array to generate a sequence of prefetch addresses. The second core issues prefetch requests for the sequence of the prefetch addresses to the memory to obtain prefetched data and the prefetched data is provided to the first core if requested.

    摘要翻译: 处理系统包括被配置为处理应用的存储器和第一核心。 第一个核心包括第一个缓存。 处理系统包括被配置为捕获错过第一核心中的第一高速缓存的应用程序的地址序列并将地址序列放置在存储阵列中的机制; 以及被配置为处理至少一个软件算法的第二核心。 所述至少一个软件算法利用来自存储阵列的地址序列来生成预取地址序列。 第二个核心将预取地址序列的预取请求发送到存储器以获得预取数据,并且如果请求,则将预取数据提供给第一核。

    EFFECTIVE PREFETCHING WITH MULTIPLE PROCESSORS AND THREADS
    3.
    发明申请
    EFFECTIVE PREFETCHING WITH MULTIPLE PROCESSORS AND THREADS 失效
    有效的预处理与多个处理器和螺纹

    公开(公告)号:US20120246406A1

    公开(公告)日:2012-09-27

    申请号:US13488215

    申请日:2012-06-04

    IPC分类号: G06F12/08

    摘要: A processing system includes a memory and a first core configured to process applications. The first core includes a first cache. The processing system includes a mechanism configured to capture a sequence of addresses of the application that miss the first cache in the first core and to place the sequence of addresses in a storage array; and a second core configured to process at least one software algorithm. The at least one software algorithm utilizes the sequence of addresses from the storage array to generate a sequence of prefetch addresses. The second core issues prefetch requests for the sequence of the prefetch addresses to the memory to obtain prefetched data and the prefetched data is provided to the first core if requested.

    摘要翻译: 处理系统包括被配置为处理应用的存储器和第一核心。 第一个核心包括第一个缓存。 处理系统包括被配置为捕获错过第一核心中的第一高速缓存的应用程序的地址序列并将地址序列放置在存储阵列中的机制; 以及被配置为处理至少一个软件算法的第二核心。 所述至少一个软件算法利用来自存储阵列的地址序列来生成预取地址序列。 第二个核心将预取地址序列的预取请求发送到存储器以获得预取数据,并且如果请求,则将预取数据提供给第一核。

    DATA REORGANIZATION IN NON-UNIFORM CACHE ACCESS CACHES
    4.
    发明申请
    DATA REORGANIZATION IN NON-UNIFORM CACHE ACCESS CACHES 有权
    非均匀缓存访问缓存中的数据重组

    公开(公告)号:US20100274973A1

    公开(公告)日:2010-10-28

    申请号:US12429754

    申请日:2009-04-24

    IPC分类号: G06F12/08 G06F12/00

    CPC分类号: G06F12/0846 G06F12/0811

    摘要: Embodiments that dynamically reorganize data of cache lines in non-uniform cache access (NUCA) caches are contemplated. Various embodiments comprise a computing device, having one or more processors coupled with one or more NUCA cache elements. The NUCA cache elements may comprise one or more banks of cache memory, wherein ways of the cache are horizontally distributed across multiple banks. To improve access latency of the data by the processors, the computing devices may dynamically propagate cache lines into banks closer to the processors using the cache lines. To accomplish such dynamic reorganization, embodiments may maintain “direction” bits for cache lines. The direction bits may indicate to which processor the data should be moved. Further, embodiments may use the direction bits to make cache line movement decisions.

    摘要翻译: 预期在非均匀缓存访问(NUCA)高速缓存中动态地重组高速缓存线的数据的实施例。 各种实施例包括具有与一个或多个NUCA高速缓存元件耦合的一个或多个处理器的计算设备。 NUCA高速缓存元件可以包括一个或多个高速缓冲存储器组,其中高速缓存的方式在多个存储体之间水平分布。 为了改善处理器对数据的访问等待时间,计算设备可以使用高速缓存行来将缓存线路动态地传播到更靠近处理器的存储体中。 为了实现这种动态重组,实施例可以保持高速缓存行的“方向”位。 方向位可以指示哪个处理器应该移动数据。 此外,实施例可以使用方向位来进行高速缓存行移动决定。

    Data reorganization in non-uniform cache access caches
    5.
    发明授权
    Data reorganization in non-uniform cache access caches 有权
    非均匀缓存访问缓存中的数据重组

    公开(公告)号:US08140758B2

    公开(公告)日:2012-03-20

    申请号:US12429754

    申请日:2009-04-24

    IPC分类号: G06F15/163

    CPC分类号: G06F12/0846 G06F12/0811

    摘要: Embodiments that dynamically reorganize data of cache lines in non-uniform cache access (NUCA) caches are contemplated. Various embodiments comprise a computing device, having one or more processors coupled with one or more NUCA cache elements. The NUCA cache elements may comprise one or more banks of cache memory, wherein ways of the cache are horizontally distributed across multiple banks. To improve access latency of the data by the processors, the computing devices may dynamically propagate cache lines into banks closer to the processors using the cache lines. To accomplish such dynamic reorganization, embodiments may maintain “direction” bits for cache lines. The direction bits may indicate to which processor the data should be moved. Further, embodiments may use the direction bits to make cache line movement decisions.

    摘要翻译: 预期在非均匀缓存访问(NUCA)高速缓存中动态地重组高速缓存线的数据的实施例。 各种实施例包括具有与一个或多个NUCA高速缓存元件耦合的一个或多个处理器的计算设备。 NUCA高速缓存元件可以包括一个或多个高速缓冲存储器组,其中高速缓存的方式在多个存储体之间水平分布。 为了改善处理器对数据的访问等待时间,计算设备可以使用高速缓存行来将缓存线路动态地传播到更靠近处理器的存储体中。 为了实现这种动态重组,实施例可以保持高速缓存行的“方向”位。 方向位可以指示哪个处理器应该移动数据。 此外,实施例可以使用方向位来进行高速缓存行移动决定。

    Active flow management with hysteresis
    6.
    发明授权
    Active flow management with hysteresis 失效
    主动流量管理带滞后

    公开(公告)号:US07453798B2

    公开(公告)日:2008-11-18

    申请号:US10782617

    申请日:2004-02-19

    IPC分类号: H04L12/00

    摘要: The present invention provides for a computer network method and system that applies “hysteresis” to an active queue management algorithm. If a queue is at a level below a certain low threshold and a burst of packets arrives at a network node, then the probability of dropping the initial packets in the burst is recalculated, but the packets are not dropped. However, if the queue level crosses beyond a hysteresis threshold, then packets are discarded pursuant to a drop probability.Also, according to the present invention, queue level may be decreased until it becomes less than the hysteresis threshold, with packets dropped per the drop probability until the queue level decreases to at least a low threshold. In one embodiment, an adaptive algorithm is also provided to adjust the transmit probability for each flow together with hysteresis to increase the packet transmit rates to absorb bursty traffic.

    摘要翻译: 本发明提供一种向活动队列管理算法应用“迟滞”的计算机网络方法和系统。 如果队列处于低于某个低阈值的水平,并且一​​群数据包到达网络节点,则重新计算突发中丢弃初始数据包的概率,但不会丢弃数据包。 然而,如果队列级别超过滞后阈值,则根据丢弃概率丢弃数据包。 此外,根据本发明,可以减少队列级别,直到其变得小于滞后阈值,其中每个丢弃概率的分组丢弃,直到队列级别降低到至少低阈值。 在一个实施例中,还提供自适应算法来调整每个流的发送概率以及迟滞以增加分组传输速率以吸收突发业务。

    Store-to-load forwarding mechanism for processor runahead mode operation
    7.
    发明授权
    Store-to-load forwarding mechanism for processor runahead mode operation 失效
    存储到负载转发机制,用于处理器跑头模式操作

    公开(公告)号:US08639886B2

    公开(公告)日:2014-01-28

    申请号:US12364984

    申请日:2009-02-03

    IPC分类号: G06F12/08

    摘要: A system and method to optimize runahead operation for a processor without use of a separate explicit runahead cache structure. Rather than simply dropping store instructions in a processor runahead mode, store instructions write their results in an existing processor store queue, although store instructions are not allowed to update processor caches and system memory. Use of the store queue during runahead mode to hold store instruction results allows more recent runahead load instructions to search retired store queue entries in the store queue for matching addresses to utilize data from the retired, but still searchable, store instructions. Retired store instructions could be either runahead store instructions retired, or retired store instructions that executed before entering runahead mode.

    摘要翻译: 一种用于在不使用单独的显式跑道缓存结构的情况下优化处理器的跑步头操作的系统和方法。 尽管存储指令不允许更新处理器缓存和系统存储器,但存储指令将其结果写入现有的处理器存储队列中,而不是简单地将存储指令放在处理器跑头模式中。 在跑步模式期间使用存储队列来保存存储指令结果允许更多的最新跑步加载指令来搜索存储队列中的退出存储队列条目以匹配地址以利用来自已退休但仍可搜索的存储指令的数据。 退休存储指令可以是退出存储指令退出,或退出存储指令,在进入排头模式之前执行。

    STORE-TO-LOAD FORWARDING MECHANISM FOR PROCESSOR RUNAHEAD MODE OPERATION
    8.
    发明申请
    STORE-TO-LOAD FORWARDING MECHANISM FOR PROCESSOR RUNAHEAD MODE OPERATION 失效
    用于处理器RUNAHEAD模式操作的存储加载转发机制

    公开(公告)号:US20100199045A1

    公开(公告)日:2010-08-05

    申请号:US12364984

    申请日:2009-02-03

    IPC分类号: G06F12/08 G06F9/312

    摘要: A system and method to optimize runahead operation for a processor without use of a separate explicit runahead cache structure. Rather than simply dropping store instructions in a processor runahead mode, store instructions write their results in an existing processor store queue, although store instructions are not allowed to update processor caches and system memory. Use of the store queue during runahead mode to hold store instruction results allows more recent runahead load instructions to search retired store queue entries in the store queue for matching addresses to utilize data from the retired, but still searchable, store instructions. Retired store instructions could be either runahead store instructions retired, or retired store instructions that executed before entering runahead mode.

    摘要翻译: 一种用于在不使用单独的显式跑道缓存结构的情况下优化处理器的跑步头操作的系统和方法。 尽管存储指令不允许更新处理器缓存和系统存储器,但存储指令将其结果写入现有的处理器存储队列中,而不是简单地将存储指令放在处理器跑头模式中。 在跑步模式期间使用存储队列来保存存储指令结果允许更多的最新跑步加载指令来搜索存储队列中的退出存储队列条目以匹配地址以利用来自已退休但仍可搜索的存储指令的数据。 退休存储指令可以是退出存储指令退出,或退出存储指令,在进入排头模式之前执行。