发明名称 |
Fast prediction of shared memory access pattern |
摘要 |
A computer implemented method analyzes shared memory accesses during execution of an application program. The method includes instrumenting events of shared memory accesses in the application program, where the application program is to be executed on a target configuration having p nodes; executing the application program using p1 processing nodes, where p1 is less than p and satisfies a constraint. For accesses made by the executing application program, the method determines a target thread and maps determined target threads to either a remote node or a local node corresponding to a remote memory access and to a local memory access, respectively. Also disclosed is a computer-readable storage medium that stores a program of executable instructions that implements the method, and a data processing system. The invention can be implemented using a language such as Unified Parallel C (UPC) directed to a partitioned global address space (PGAS) paradigm. |
申请公布号 |
US8819346(B2) |
申请公布日期 |
2014.08.26 |
申请号 |
US201213416331 |
申请日期 |
2012.03.09 |
申请人 |
International Business Machines Corporation |
发明人 |
Cong Guojing;Tiotto Ettore;Wen Hui-Fang |
分类号 |
G06F12/00 |
主分类号 |
G06F12/00 |
代理机构 |
Harrington & Smith |
代理人 |
Harrington & Smith |
主权项 |
1. A computer implemented method to analyze shared memory accesses during execution of an application program, comprising:
instrumenting events of shared memory accesses in the application program, where the application program is to be executed on a target configuration comprising p nodes; executing the application program using p1 processing nodes, where p1 is less than p and satisfies a constraint; and for accesses made by the executing application program, determining a target thread and mapping determined target threads to either a remote node or a local node corresponding to a remote memory access and to a local memory access, respectively; where instrumenting instruments a compiled version of the application program; and where mapping comprises intercepting shared memory access function calls generated by a compiler and analyzing arguments of the function calls to determine whether an access is a remote access or a local access. |
地址 |
Armonk NY US |