发明名称 |
Method and apparatus for transforming program code |
摘要 |
Provided is a method of transforming program code written such that a plurality of work-items are allocated respectively to and concurrently executed on a plurality of processing elements included in a computing unit. A program code translator may identify, in the program code, two or more code regions, which are to be enclosed by work-item coalescing loops (WCLs), based on a synchronization barrier function contained in the program code, such that the work-items are serially executable on a smaller number of processing elements than a number of the processing elements, and may enclose the identified code regions with the WCLs, respectively. |
申请公布号 |
US9015683(B2) |
申请公布日期 |
2015.04.21 |
申请号 |
US201012977786 |
申请日期 |
2010.12.23 |
申请人 |
Samsung Electronics Co., Ltd.;SNU R&DB Foundation |
发明人 |
Cho Seung-Mo;Choi Jong-Deok;Lee Jaejin |
分类号 |
G06F9/44;G06F9/45 |
主分类号 |
G06F9/44 |
代理机构 |
NSIP Law |
代理人 |
NSIP Law |
主权项 |
1. A method of transforming program code written such that a plurality of work-items are concurrently executed on a plurality of processing elements included in a computing unit, the method comprising:
identifying, in the program code executed at a computing unit, two or more code regions that are to be enclosed by work-item coalescing loops (WCLs) based on a synchronization barrier function contained in the program code, such that the work-items are serially executable on a smaller number of processing elements than a number of the work-items; enclosing the identified code regions with the WCLs; and in response to a private variable contained in the program code being defined in one of the identified code regions and being used in another identified code region, expanding the private variable according to a number of dimensions in identifiers of the work-items. |
地址 |
Suwon-si KR |