中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

3 records (took 0.015 seconds)

    FIELD SPECIALIZATION SYSTEMS AND METHODS FOR IMPROVING PROGRAM PERFORMANCE
    1.
    发明申请
    FIELD SPECIALIZATION SYSTEMS AND METHODS FOR IMPROVING PROGRAM PERFORMANCE 审中-公开
    现场专业系统和改进程序性能的方法

    公开(公告)号：WO2016161130A1

    公开(公告)日：2016-10-06

    申请号：PCT/US2016/025295

    申请日：2016-03-31

    Applicant: DATAWARE VENTURES, LLC

    Inventor： SNODGRASS, Richard, T. ,  DEBRAY, Saumya, K. ,  ZHANG, Rui ,  THOMAS, Stephen ,  MASON, Sean

    IPC: G06F9/45 ,  G06F9/445 ,  G06F17/30 ,  G06F12/00

    CPC classification number: G06F17/30466 ,  G06F8/443 ,  G06F17/30474

    Abstract: Systems and methods for improving the performance of a computer program, such as a database management system (DBMS), are provided. Such methods involve identifying invariant intervals for variables in the DBMS code, based on a Program Representation (PR). Program interactions within the DBMS and domain assertions are deduced, based on the PR and an Ecosystem Specification for the DBMS. One or more candidate snippets are identified, based on the invariant intervals for variables in the DBMS code, the PR, one or more execution summaries associated with the DBMS, the deduced program interactions and the deduced domain assertions. Spiffs are then generated, based on the one or more candidate snippets. Such spiffs include predicate query spiff, hash-join query spiff, aggregate spiff, page spiff, and string matching spiff. The DBMS code is modified based on the speccode generated from these spiffs.

    Abstract translation: 提供了用于提高诸如数据库管理系统（DBMS）的计算机程序的性能的系统和方法。 这种方法涉及基于程序表示（PR）来识别DBMS代码中的变量的不变间隔。 基于PR和DBMS的生态系统规范，推导出DBMS和域断言中的程序交互。 基于DBMS代码中的变量，PR，与DBMS相关联的一个或多个执行摘要，推断的程序交互和推断的域断言来标识一个或多个候选片段。 然后根据一个或多个候选片段生成Spiff。 这样的spiff包括谓词查询spiff，哈希连接查询spiff，聚合spiff，页面spiff和字符串匹配spiff。 DBMS代码根据从这些spiff生成的规范代码进行修改。

    BROADENING FIELD SPECIALIZATION
    2.
    发明申请
    BROADENING FIELD SPECIALIZATION 审中-公开
    拓展野外专业

    公开(公告)号：WO2017106351A1

    公开(公告)日：2017-06-22

    申请号：PCT/US2016/066665

    申请日：2016-12-14

    Applicant: DATAWARE VENTURES, LLC ,  THE ARIZONA BOARD OF REGENTS ON BEHALF OF THE UNIVERSITY OF ARIZONA

    Inventor： SNODGRASS, Richard, T. ,  DEBRAY, Saumya, K. ,  ZHANG, Rui ,  LIU, Yang

    IPC: G06F7/00 ,  G06F9/445 ,  G06F9/45 ,  G06F12/00 ,  G06F17/30

    CPC classification number: G06F7/00 ,  G06F8/443 ,  G06F9/44521 ,  G06F12/00

    Abstract: Four extensions to the conventional field specialization process are disclosed. The first extension is cross-application value flows, where a value transfers out of one application and subsequently into another application. The second extension is an inter-application analysis. Static and dynamic analysis is performed by a Spiff Toolset not just on the source code of a single application, but also across the data read and written by that application. The third extension is invariant cross-application termination, verifying the possibility of an invariant originating in an application and terminating in a specialization opportunity in a separate application. The fourth extension relates to run-time code placement algorithms to mitigate an increase in I-cache pressure and L2-cache pressure. A MaxResidency algorithm uses information from dynamic analysis of provided workloads before DBMS compilation and specific structure of the query evaluation plan to place run-time code, thus retaining possible run time improvement of that code.

    Abstract translation: 公开了传统现场专业化过程的四个扩展。 第一个扩展是跨应用程序价值流，其中一个值从一个应用程序转移出来，然后转移到另一个应用程序。 第二个扩展是一个应用程序间分析。 静态和动态分析由Spiff工具集执行，不仅在单个应用程序的源代码上，而且在由该应用程序读取和写入的数据中执行。 第三个扩展是不变的跨应用程序终止，验证源自应用程序的不变量的可能性，并终止在单独应用程序中的专门化机会。 第四个扩展涉及运行时代码放置算法，以缓解I缓存压力和L2缓存压力的增加。 MaxResidency算法在DBMS编译之前使用来自提供的工作负载的动态分析的信息，并使用查询评估计划的特定结构来放置运行时代码，从而保持该代码可能的运行时间改进。

Patent Agency Ranking