東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Dynamic translation of runtime envir...

Dominguez, Rodrigo.

Linked to FindBook

Google Book

Amazon

博客來

Dynamic translation of runtime environments for heterogeneous computing.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Dynamic translation of runtime environments for heterogeneous computing./
Author:	Dominguez, Rodrigo.
Description:	88 p.
Notes:	Source: Dissertation Abstracts International, Volume: 74-09(E), Section: B.
Contained By:	Dissertation Abstracts International74-09B(E).
Subject:	Engineering, Computer. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3563295
ISBN:	9781303115158

Dynamic translation of runtime environments for heterogeneous computing.
Dominguez, Rodrigo.

Dynamic translation of runtime environments for heterogeneous computing. - 88 p.

Source: Dissertation Abstracts International, Volume: 74-09(E), Section: B.

Thesis (Ph.D.)--Northeastern University, 2013.

The recent move toward heterogeneous computer architectures calls for a global rethinking of current software and hardware paradigms. Researchers are exploring new parallel programming models, advanced compiler designs, and novel resource management techniques to exploit the features of many-core processor architectures. Graphics Processing Units (GPUs) have become the platform of choice in this area for accelerating a large range of data-parallel and task-parallel applications. The rapid adoption of GPU computing has been greatly aided by the introduction of high-level programming environments such as CUDA C and OpenCL. However, each vendor implements these programming models differently and we must analyze the internals in order to get a better understanding of the performance results.

ISBN: 9781303115158Subjects--Topical Terms:

1669061
Engineering, Computer.

Dynamic translation of runtime environments for heterogeneous computing.
LDR:03589nam a2200313 4500 001 1960323
005 20140611111835.5
008 150210s2013 ||||||||||||||||| ||eng d
020 $a 9781303115158
035 $a (MiAaPQ)AAI3563295
035 $a AAI3563295
040 $a MiAaPQ $c MiAaPQ
100 1 $a Dominguez, Rodrigo. $3 2095954
245 1 0 $a Dynamic translation of runtime environments for heterogeneous computing.
300 $a 88 p.
500 $a Source: Dissertation Abstracts International, Volume: 74-09(E), Section: B.
500 $a Adviser: David R. Kaeli.
502 $a Thesis (Ph.D.)--Northeastern University, 2013.
520 $a The recent move toward heterogeneous computer architectures calls for a global rethinking of current software and hardware paradigms. Researchers are exploring new parallel programming models, advanced compiler designs, and novel resource management techniques to exploit the features of many-core processor architectures. Graphics Processing Units (GPUs) have become the platform of choice in this area for accelerating a large range of data-parallel and task-parallel applications. The rapid adoption of GPU computing has been greatly aided by the introduction of high-level programming environments such as CUDA C and OpenCL. However, each vendor implements these programming models differently and we must analyze the internals in order to get a better understanding of the performance results.
520 $a This dissertation presents the design of Caracal, a dynamic compiler that can translate between runtime environments used in heterogeneous computing. A major challenge of developing such a translation system is the inherent difference in both the underlying instruction set architecture, as well as the runtime system. One of the more challenging questions across different runtime environments is the handling of program control flow by the compiler and the hardware. Some implementations can support unstructured control flow based on branches and labels, while others are based on structured control flow relying solely on if-then and while constructs.
520 $a Caracal can translate applications based on unstructured control flow so they can run on hardware that requires structured programs. In order to accomplish this, Caracal builds the control tree of the program and creates single-entry, single-exit regions called hammock graphs. We use Caracal to analyze the performance differences between NVIDIA's implementation of CUDA C and AMD's implementation of OpenCL. Our results show that the requirement for structured control flow can impact register pressure, and can degrade performance by as much as 2.5X.
520 $a We also explore the differences between heterogeneous parallel processors architectures that will impact the design of a translator when attempting to optimize code. We evaluate this issue for two specific compiler optimizations that are highly sensitive to the underlying architecture. We use Caracal to evaluate 1) vectorization and 2) loop unrolling. Our translation system needs to capture rich semantics to be able to effectively translate code to different systems that will exploit these two optimization techniques differently. Our experiments show that by tuning these two optimizations in our translation system, we can reduce execution time by up to 90%.
590 $a School code: 0160.
650 4 $a Engineering, Computer. $3 1669061
650 4 $a Computer Science. $3 626642
690 $a 0464
690 $a 0984
710 2 $a Northeastern University. $b Electrical and Computer Engineering. $3 1018491
773 0 $t Dissertation Abstracts International $g 74-09B(E).
790 $a 0160
791 $a Ph.D.
792 $a 2013
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3563295