Data distribution fabric in scalable GPUs
    52.
    发明授权
    Data distribution fabric in scalable GPUs 有权
    可扩展GPU中的数据分发结构

    公开(公告)号:US09330433B2

    公开(公告)日:2016-05-03

    申请号:US14320478

    申请日:2014-06-30

    摘要: In on embodiment, a hybrid fabric interconnects multiple graphics processor cores within a processor. The hybrid fabric interconnect includes multiple data channels, including programmable virtual data channels. The virtual data channels carry multiple traffic classes of packet-based messages. The virtual data channels and multiple traffic classes may be assigned one of multiple priorities. The virtual data channels may be arbitrated independently. The hybrid fabric is scalable and can support multiple topologies, including multiple stacked integrated circuit topologies.

    摘要翻译: 在实施例中,混合架构互连处理器内的多个图形处理器核心。 混合网络互连包括多个数据信道,包括可编程虚拟数据信道。 虚拟数据信道承载多个基于分组的消息的业务类别。 可以将虚拟数据信道和多个业务类别分配为多个优先级之一。 虚拟数据通道可以独立地进行仲裁。 混合架构是可扩展的,可以支持多种拓扑结构,包括多个堆叠集成电路拓扑。

    SYSTEM COHERENCY IN A DISTRIBUTED GRAPHICS PROCESSOR HIERARCHY
    53.
    发明申请
    SYSTEM COHERENCY IN A DISTRIBUTED GRAPHICS PROCESSOR HIERARCHY 有权
    分布式图形处理器层次分析中的系统概念

    公开(公告)号:US20150278984A1

    公开(公告)日:2015-10-01

    申请号:US14227525

    申请日:2014-03-27

    IPC分类号: G06T1/60

    摘要: Methods and systems may provide for executing, by a physically distributed set of compute slices, a plurality of work items. Additionally, the coherency of one or more memory lines associated with the plurality of work items may be maintained, by a cache fabric, across a graphics processor, a system memory and one or more host processors. In one example, a plurality of crossbar nodes track the one or more memory lines, wherein the coherency of the one or more memory lines is maintained across a plurality of level one (L1) caches and a physically distributed cache structure. Each L1 cache may be dedicated to an execution block of a compute slice and each crossbar node may be dedicated to a compute slice.

    摘要翻译: 方法和系统可以提供通过物理分布的计算片段来执行多个工作项目。 此外,与多个工作项相关联的一个或多个存储器线的一致性可以由缓存结构跨图形处理器,系统存储器和一个或多个主机处理器来维护。 在一个示例中,多个交叉开关节点跟踪一个或多个存储器线,其中一个或多个存储器线的一致性被保持在多个一级(L1)高速缓存和物理分布的高速缓存结构上。 每个L1高速缓存可以专用于计算片的执行块,并且每个交叉节点可以专用于计算片。

    Dynamically Rebalancing Graphics Processor Resources
    55.
    发明申请
    Dynamically Rebalancing Graphics Processor Resources 有权
    动态重新平衡图形处理器资源

    公开(公告)号:US20140125679A1

    公开(公告)日:2014-05-08

    申请号:US13669576

    申请日:2012-11-06

    IPC分类号: G06T1/00

    CPC分类号: G06T1/20

    摘要: According to some embodiments, performance bottlenecks that arise in particular resources within a graphic processor unit may be alleviated by dynamically rebalancing workloads among the resources, with the goal of removing the current performance bottleneck, while at the same time maintaining power dissipation within a currently allocated power budget. In some embodiments this may be achieved by defining a separate clock domain for each of the plurality of graphics processor resources whose performance may then be rebalanced.

    摘要翻译: 根据一些实施例,可以通过动态地重新平衡资源之间的工作负载来缓解在图形处理器单元内的特定资源中出现的性能瓶颈,目的是消除当前的性能瓶颈,同时保持当前分配的 电力预算 在一些实施例中,这可以通过为多个图形处理器资源中的每个图形处理器资源定义单独的时钟域来实现,其中性能可以被重新平衡。

    Dynamic squelch detection power control
    56.
    发明授权
    Dynamic squelch detection power control 失效
    动态静噪检测功率控制

    公开(公告)号:US08352764B2

    公开(公告)日:2013-01-08

    申请号:US12286188

    申请日:2008-09-29

    IPC分类号: G06F1/32 G06F1/26

    摘要: In one embodiment, the present invention includes power control logic for squelch detection circuitry to enable selective enabling of one or more squelch detection circuits of an interconnect interface in a low power mode. The logic may include a squelch mode control register to select a first mode or a second mode of power control, a second register coupled to the squelch mode control register to receive software settings to indicate which squelch detect circuit(s) to disable in a low power state of the interconnect, and a detector to dynamically detect a logical lane zero of the interconnect in the second mode. Other embodiments are described and claimed.

    摘要翻译: 在一个实施例中,本发明包括用于静噪检测电路的功率控制逻辑,以使得能够以低功率模式选择性地启用互连接口的一个或多个静噪检测电路。 逻辑可以包括静噪模式控制寄存器以选择第一模式或第二模式的功率控制;第二寄存器,耦合到静噪模式控制寄存器以接收软件设置,以指示哪个静噪检测电路在低电平中禁用 互连的功率状态,以及用于在第二模式中动态地检测互连的逻辑通道零点的检测器。 描述和要求保护其他实施例。

    METHOD AND SYSTEM FOR SAFE ENQUEUING OF EVENTS
    57.
    发明申请
    METHOD AND SYSTEM FOR SAFE ENQUEUING OF EVENTS 有权
    安全事件的方法和系统

    公开(公告)号:US20130007751A1

    公开(公告)日:2013-01-03

    申请号:US13175493

    申请日:2011-07-01

    IPC分类号: G06F9/46

    CPC分类号: G06F9/544 G06F9/545

    摘要: A method and system to facilitate a user level application executing in a first processing unit to enqueue work or task(s) safely for a second processing unit without performing any ring transition. For example, in one embodiment of the invention, the first processing unit executes one or more user level applications, where each user level application has a task to be offloaded to a second processing unit. The first processing unit signals the second processing unit to handle the task from each user level application without performing any ring transition in one embodiment of the invention.

    摘要翻译: 一种促进在第一处理单元中执行的用户级应用程序以对第二处理单元安全地排队工作或任务而不执行任何环转换的方法和系统。 例如,在本发明的一个实施例中,第一处理单元执行一个或多个用户级应用,其中每个用户级应用具有卸载到第二处理单元的任务。 在本发明的一个实施例中,第一处理单元用信号通知第二处理单元来处理来自每个用户级应用的任务而不执行任何环转移。

    ENSURING COHERENCE BETWEEN GRAPHICS AND DISPLAY DOMAINS
    58.
    发明申请
    ENSURING COHERENCE BETWEEN GRAPHICS AND DISPLAY DOMAINS 有权
    保持图形和显示域之间的一致性

    公开(公告)号:US20100235320A1

    公开(公告)日:2010-09-16

    申请号:US12401499

    申请日:2009-03-10

    IPC分类号: G06F12/08 G06F17/30

    摘要: A platform may comprise a core coherency domain, graphics coherency domain and a non-coherent domain. A graphics acceleration unit (GAU) of the graphics coherency domain may generate data units from an application and the data units may comprise display data units. The GAU may annotate the display data units with an annotation value before flushing the display data units to an on-die cache. The GAU may identify modified display data units among the display data units stored in the on-die cache and issue flush commands to cause flushing of the modified display data units from the on-die cache to a main memory. The display engine of the non-coherent domain may use the modified display data units stored in the main memory to render a display on a display device.

    摘要翻译: 平台可以包括核心一致性域,图形一致性域和非相干域。 图形一致性域的图形加速单元(GAU)可以从应用产生数据单元,并且数据单元可以包括显示数据单元。 GAU可以在将显示数据单元刷新到片上高速缓存之前用注释值来注释显示数据单元。 GAU可以识别存储在片上高速缓存中的显示数据单元中的修改的显示数据单元,并发出刷新命令,以使经修改的显示数据单元从模块缓存刷新到主存储器。 非相干域的显示引擎可以使用存储在主存储器中的修改的显示数据单元在显示设备上呈现显示。

    Method and apparatus for reordering data in X86 ordering
    60.
    发明授权
    Method and apparatus for reordering data in X86 ordering 有权
    用于在X86订购中重新排序数据的方法和装置

    公开(公告)号:US06457121B1

    公开(公告)日:2002-09-24

    申请号:US09270981

    申请日:1999-03-17

    IPC分类号: G06F1200

    CPC分类号: G06F13/4054

    摘要: A device to change the ordering of datums in a packet from a storage device to a pre-determined ordering according to their addresses. The device has a first circuit to receive and process address information to determine a data ordering of data associated with the address information; and a second circuit to reorder the data into ordered packets in the predetermined ordering. This device can be used to efficiently transfer graphic data through the AGP bus in a computer.

    摘要翻译: 根据其地址,将存储设备中的数据包中的基准的顺序改变为预定的顺序的设备。 该设备具有接收和处理地址信息以确定与地址信息相关联的数据的数据顺序的第一电路; 以及第二电路,用于以预定顺序将数据重新排序成有序数据包。 该设备可用于通过计算机中的AGP总线有效传输图形数据。