國立高雄大學圖資館 |

Language: English

Back

The Jrpm system for dynamically parallelizing sequential Java programs

Record Type:	Electronic resources : Monograph/item
Title/Author:	The Jrpm system for dynamically parallelizing sequential Java programs
Author:	Chen, Michael Kerby.
Description:	196 p.
Notes:	Adviser: Kunle Olukotun.
Notes:	Source: Dissertation Abstracts International, Volume: 65-04, Section: B, page: 2005.
Contained By:	Dissertation Abstracts International65-04B.
Subject:	Engineering, Electronics and Electrical.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3128629
ISBN:	0496758918

The Jrpm system for dynamically parallelizing sequential Java programs
Chen, Michael Kerby.

The Jrpm system for dynamically parallelizing sequential Java programs [electronic resource] - 196 p.

Adviser: Kunle Olukotun.

Thesis (Ph.D.)--Stanford University, 2004.

As the limits of instruction-level parallelism (1--10s of instructions) with a single thread of control are approached, we must look elsewhere for architectural improvements that can speedup program execution. Coarser grained parallelism, like fine-grained thread-level parallelism (10s--1,000s of instructions), is a potential area for exploration. This type of parallelism is not exploited today due to limitations of current multiprocessor architectures and automatic parallelizing compilers.

ISBN: 0496758918Subjects--Topical Terms:

226981
Engineering, Electronics and Electrical.

The Jrpm system for dynamically parallelizing sequential Java programs
LDR:03052nmm _2200301 _450 001 162627
005 20051017073506.5
008 230606s2004 eng d
020 $a 0496758918
035 $a 00149128
035 $a 162627
040 $a UnM $c UnM
100 0 $a Chen, Michael Kerby. $3 227771
245 1 4 $a The Jrpm system for dynamically parallelizing sequential Java programs $h [electronic resource]
300 $a 196 p.
500 $a Adviser: Kunle Olukotun.
500 $a Source: Dissertation Abstracts International, Volume: 65-04, Section: B, page: 2005.
502 $a Thesis (Ph.D.)--Stanford University, 2004.
520 # $a As the limits of instruction-level parallelism (1--10s of instructions) with a single thread of control are approached, we must look elsewhere for architectural improvements that can speedup program execution. Coarser grained parallelism, like fine-grained thread-level parallelism (10s--1,000s of instructions), is a potential area for exploration. This type of parallelism is not exploited today due to limitations of current multiprocessor architectures and automatic parallelizing compilers.
520 # $a Experimental results demonstrate that Jrpm can exploit thread-level parallelism with minimal effort from the programmer. On four processors, I achieve speedups of 3 to 4 for floating point applications, 2 to 3 on multimedia applications, and between 1.5 and 2.5 on integer applications. Performance resulted from automatic selection of thread decompositions by the hardware profiler, intra-procedural optimizations on code compiled dynamically into speculative threads, and some minor programmer transformations for exposing parallelism that cannot be performed automatically.
520 # $a This dissertation describes the Java runtime parallelizing machine (Jrpm), a complete system for parallelizing sequential programs automatically. Jrpm is based on a CMP (chip multiprocessor) with TLS (thread-level speculation) support. CMPs, systems with tightly coupled processor cores implemented together on a single die, have low sharing and communication costs relative to traditional multiprocessors. TLS simplifies program parallelization by allowing one to parallelize optimistically without violating correct sequential program behavior. Using a Java virtual machine with dynamic compilation support coupled with a hardware profiler, both speculative buffer requirements and inter-thread dependencies of prospective STLs (speculative thread loops) are analyzed in real-time to identify the best loops to parallelize. Once sufficient data has been collected to make a reasonable decision, selected loops are recompiled dynamically to run in parallel.
590 $a School code: 0212.
650 # 0 $a Engineering, Electronics and Electrical. $3 226981
650 # 0 $a Computer Science. $3 212513
690 $a 0544
690 $a 0984
710 0 # $a Stanford University. $3 212607
773 0 # $g 65-04B. $t Dissertation Abstracts International
790 $a 0212
790 1 0 $a Olukotun, Kunle, $e advisor
791 $a Ph.D.
792 $a 2004
856 4 0 $u http://libsw.nuk.edu.tw/login?url=http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3128629 $z http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3128629