公开(公告)号：US07669194B2

公开(公告)日：2010-02-23

申请号：US10926595

申请日：2004-08-26

申请人： Roch Georges Archambault ,  Robert James Blainey ,  Yaoqing Gao ,  Allan Russell Martin ,  James Lawrence McInnes ,  Francis Patrick O'Connell

发明人： Roch Georges Archambault ,  Robert James Blainey ,  Yaoqing Gao ,  Allan Russell Martin ,  James Lawrence McInnes ,  Francis Patrick O'Connell

IPC分类号： G06F9/44 ,  G06F9/45 ,  G06F9/30

CPC分类号： G06F8/4442

摘要： A mechanism for minimizing effective memory latency without unnecessary cost through fine-grained software-directed data prefetching using integrated high-level and low-level code analysis and optimizations is provided. The mechanism identifies and classifies streams, identifies data that is most likely to incur a cache miss, exploits effective hardware prefetching to determine the proper number of streams to be prefetched, exploits effective data prefetching on different types of streams in order to eliminate redundant prefetching and avoid cache pollution, and uses high-level transformations with integrated lower level cost analysis in the instruction scheduler to schedule prefetch instructions effectively.

摘要翻译： 提供了一种通过使用集成高级和低级代码分析和优化的细粒度软件导向数据预取来最小化有效存储器延迟而不需要成本的机制。 该机制识别和分类流，识别最可能引起缓存未命中的数据，利用有效的硬件预取来确定要预取的流的适当数量，利用不同类型的流上的有效数据预取，以消除冗余预取和 避免高速缓存污染，并在指令调度程序中使用集成较低级别成本分析的高级转换，有效地调度预取指令。

公开(公告)号：US20100095271A1

公开(公告)日：2010-04-15

申请号：US12644756

申请日：2009-12-22

申请人： Roch Georges Archambault ,  Robert James Blainey ,  Yaoqing Gao ,  Allan Russell Martin ,  James Lawrence McInnes ,  Francis Patrick O'Connell

发明人： Roch Georges Archambault ,  Robert James Blainey ,  Yaoqing Gao ,  Allan Russell Martin ,  James Lawrence McInnes ,  Francis Patrick O'Connell

IPC分类号： G06F9/44

CPC分类号： G06F8/4442

摘要： A mechanism for minimizing effective memory latency without unnecessary cost through fine-grained software-directed data prefetching using integrated high-level and low-level code analysis and optimizations is provided. The mechanism identifies and classifies streams, identifies data that is most likely to incur a cache miss, exploits effective hardware prefetching to determine the proper number of streams to be prefetched, exploits effective data prefetching on different types of streams in order to eliminate redundant prefetching and avoid cache pollution, and uses high-level transformations with integrated lower level cost analysis in the instruction scheduler to schedule prefetch instructions effectively.

摘要翻译： 提供了一种通过使用集成高级和低级代码分析和优化的细粒度软件导向数据预取来最小化有效存储器延迟而不需要成本的机制。 该机制识别和分类流，识别最可能引起缓存未命中的数据，利用有效的硬件预取来确定要预取的流的适当数量，利用不同类型的流上的有效数据预取，以消除冗余预取和 避免高速缓存污染，并在指令调度程序中使用集成较低级别成本分析的高级转换，有效地调度预取指令。

公开(公告)号：US07506331B2

公开(公告)日：2009-03-17

申请号：US10930042

申请日：2004-08-30

申请人： Roch Georges Archambault ,  Robert Frederick Enenkel ,  Robert William Hay ,  Allan Russell Martin ,  James Lawrence McInnes ,  Ronald Ian McIntosh ,  Mark Peter Mendell

发明人： Roch Georges Archambault ,  Robert Frederick Enenkel ,  Robert William Hay ,  Allan Russell Martin ,  James Lawrence McInnes ,  Ronald Ian McIntosh ,  Mark Peter Mendell

IPC分类号： G06F9/45

CPC分类号： G06F8/443

摘要： A method, apparatus, and computer instructions for processing instructions. A data dependency graph is built. The data dependency graph is analyzed for recurrences, and unpipelined instructions that lie outside of the recurrences are expanded.

摘要翻译： 一种用于处理指令的方法，装置和计算机指令。 构建数据依赖图。 分析数据依赖关系图以进行复现，扩展位于复发之外的无关注指令。

公开(公告)号：US07962907B2

公开(公告)日：2011-06-14

申请号：US11840371

申请日：2007-08-17

申请人： Allan Russell Martin ,  James Lawrence McInnes

发明人： Allan Russell Martin ,  James Lawrence McInnes

IPC分类号： G06F9/44

CPC分类号： G06F8/4452 ,  G06F9/3838

摘要： An improved scheduling technique for software pipelining is disclosed which is designed to find schedules requiring fewer processor clock cycles and reduce register pressure hot spots when scheduling multiple groups of instructions (e.g. as represented by multiple sub-graphs of a DDG) which are independent, and substantially identical. The improvement in instruction scheduling and reduction of hot spots is achieved by evenly distributing such groups of instructions around the schedule for a given loop.

摘要翻译： 公开了一种用于软件流水线的改进的调度技术，其被设计为在调度多个独立的指令组（例如，由DDG的多个子图表示）时，需要更少的处理器时钟周期并减少寄存器压力热点，以及 基本相同。 指令调度的改善和热点的减少是通过围绕给定循环的时间表均匀分布这些指令组来实现的。

公开(公告)号：US07331045B2

公开(公告)日：2008-02-12

申请号：US10835129

申请日：2004-04-29

申请人： Allan Russell Martin ,  James Lawrence McInnes

发明人： Allan Russell Martin ,  James Lawrence McInnes

IPC分类号： G06F9/45

CPC分类号： G06F8/4452 ,  G06F9/3838

摘要： An improved scheduling technique for software pipelining is disclosed which is designed to find schedules requiring fewer processor clock cycles and reduce register pressure hot spots when scheduling multiple groups of instructions (e.g. as represented by multiple sub-graphs of a DDG) which are independent, and substantially identical. The improvement in instruction scheduling and reduction of hot spots is achieved by evenly distributing such groups of instructions around the schedule for a given loop.

摘要翻译： 公开了一种用于软件流水线的改进的调度技术，其被设计为在调度多个独立的指令组（例如，由DDG的多个子图表示）时，需要更少的处理器时钟周期并减少寄存器压力热点，以及 基本相同。 指令调度的改善和热点的减少是通过围绕给定循环的时间表均匀分布这些指令组来实现的。

公开(公告)号：US07493611B2

公开(公告)日：2009-02-17

申请号：US10929193

申请日：2004-08-30

申请人： Allan Russell Martin

发明人： Allan Russell Martin

IPC分类号： G06F9/45

CPC分类号： G06F8/4452

摘要： A scheduling algorithm is provided for selecting the placement of instructions with internal slack into a schedule of instructions within a loop. The algorithm achieves this by pinning nodes with internal slack to corresponding nodes on the critical path of the code that have similar properties in terms of the data dependency graph, such as earliest time and latest time. The effect is that nodes with internal slack are more often optimally placed in the schedule, reducing the need for rotating registers or register copy instructions. The benefit of the present invention can primarily be seen when performing instruction scheduling or software pipelining on loop code, but can also apply to other forms of instruction scheduling when greater control of placement of nodes with internal slack is desired.

摘要翻译： 提供了一种调度算法，用于选择具有内部松弛的指令的放置到循环内的指令调度。 该算法通过将具有内部松弛的节点固定到具有数据依赖图方面具有类似属性的代码的关键路径上的相应节点（诸如最早的时间和最近的时间）来实现。 效果是内部松弛的节点通常被最佳地放置在时间表中，减少了对旋转寄存器或寄存器复制指令的需要。 当在循环码上执行指令调度或软件流水线时，主要可以看到本发明的好处，但是当期望具有内部松弛的节点的放置的更大控制时也可以应用于其他形式的指令调度。

公开(公告)号：US07444628B2

公开(公告)日：2008-10-28

申请号：US10930040

申请日：2004-08-30

申请人： Allan Russell Martin

发明人： Allan Russell Martin

IPC分类号： G06F9/45

CPC分类号： G06F8/4452

摘要： A method, apparatus, and computer instructions for scheduling instructions for execution. Identify a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency. Determine whether the series of instructions is a uniform series of instructions. Schedule execution of the uniform series of instructions within the loop to optimize execution of the loop in response to the identified series of instructions being the uniform series of instructions.

摘要翻译： 一种用于调度执行指令的方法，装置和计算机指令。 识别循环中的一系列指令，其中该系列指令具有循环数据依赖性。 确定一系列指令是否是统一的指令系列。 计划执行循环内的统一指令序列，以响应于所指定的一系列指令是均匀的一系列指令来优化循环的执行。

公开(公告)号：US08387035B2

公开(公告)日：2013-02-26

申请号：US12353154

申请日：2009-01-13

申请人： Allan Russell Martin

发明人： Allan Russell Martin

IPC分类号： G06F9/45

CPC分类号： G06F8/4452

摘要： A scheduling algorithm is provided for selecting the placement of instructions with internal slack into a schedule of instructions within a loop. The algorithm achieves this by pinning nodes with internal slack to corresponding nodes on the critical path of the code that have similar properties in terms of the data dependency graph, such as earliest time and latest time. The effect is that nodes with internal slack are more often optimally placed in the schedule, reducing the need for rotating registers or register copy instructions. The benefit of the present invention can primarily be seen when performing instruction scheduling or software pipelining on loop code, but can also apply to other forms of instruction scheduling when greater control of placement of nodes with internal slack is desired.

公开(公告)号：US08266610B2

公开(公告)日：2012-09-11

申请号：US12233895

申请日：2008-09-19

申请人： Allan Russell Martin

发明人： Allan Russell Martin

IPC分类号： G06F9/45

CPC分类号： G06F8/4452

摘要： A method, apparatus, and computer instructions for scheduling instructions for execution. Identify a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency. Determine whether the series of instructions is a uniform series of instructions. Schedule execution of the uniform series of instructions within the loop to optimize execution of the loop in response to the identified series of instructions being the uniform series of instructions.

摘要翻译： 一种用于调度执行指令的方法，装置和计算机指令。 识别循环中的一系列指令，其中该系列指令具有循环数据依赖性。 确定一系列指令是否是统一的指令系列。 计划执行循环内的统一指令序列，以响应于所指定的一系列指令是均匀的一系列指令来优化循环的执行。

公开(公告)号：US20090013316A1

公开(公告)日：2009-01-08

申请号：US12233895

申请日：2008-09-19

申请人： Allan Russell Martin

发明人： Allan Russell Martin

IPC分类号： G06F9/44

CPC分类号： G06F8/4452

摘要： A method, apparatus, and computer instructions for scheduling instructions for execution. Identify a series of instructions in a loop, wherein the series of instructions has a cyclic data dependency. Determine whether the series of instructions is a uniform series of instructions. Schedule execution of the uniform series of instructions within the loop to optimize execution of the loop in response to the identified series of instructions being the uniform series of instructions.

摘要翻译： 一种用于调度执行指令的方法，装置和计算机指令。 识别循环中的一系列指令，其中该系列指令具有循环数据依赖性。 确定一系列指令是否是统一的指令系列。 计划执行循环内的统一指令序列，以响应于所指定的一系列指令是均匀的一系列指令来优化循环的执行。