Using software logging to support multi-version buffering in thread-level speculation

被引:0
作者
Garzarán, MJ [1 ]
Prvulovic, M [1 ]
Viñals, V [1 ]
Llabería, JM [1 ]
Rauchwerger, L [1 ]
Torrellas, J [1 ]
机构
[1] Univ Illinois, Urbana, IL 61801 USA
来源
12TH INTERNATIONAL CONFERENCE ON PARALLEL ARCHITECTURES AND COMPILATION TECHNIQUES, PROCEEDINGS | 2003年
关键词
D O I
暂无
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
In Thread-Level Speculation (TLS), speculative tasks generate memory state that cannot simply be combined with the rest of the system because it is unsafe. One way to deal with this difficulty is to allow speculative state to merge with memory but back up in an undo log the data that will be overwritten. Such undo log can be used to roll back to a safe state if a violation occurs. This approach is said to use Future Main Memory (FMM), as memory keeps the most speculative state. While the aggressive approach of FMM systems often delivers better performance than more conservative approaches, it also requires additional hardware support. To simplify the design of FMM systems, this paper proposes a software-only design for the undo log system. We show that an FMM system with software logging is a good design point: the design has less implementation complexity than an FMM system with hardware logs, and it only reduces performance moderately. In particular, in a simulated 16-processor machine, applications take only 10% longer to execute than if the system had the logging system fully implemented in hardware.
引用
收藏
页码:170 / 181
页数:12
相关论文
共 24 条
[1]   Report of the European Association for Palliative Care [J].
Blumhuber, H ;
DeConno, F ;
Hanks, GW .
JOURNAL OF PAIN AND SYMPTOM MANAGEMENT, 1996, 12 (02) :82-84
[2]  
Cintra M, 2000, PROCEEDING OF THE 27TH INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE, P13, DOI [10.1145/342001.363382, 10.1109/ISCA.2000.854373]
[3]  
EKANADHAM K, 1998, P 4 INT S HIGH PERF
[4]  
FIGUEIREDO R, 2001, P INT C PAR PROC SEP
[5]  
FRANK M, 2001, MITLCSTM619
[6]   ARB: A hardware mechanism for dynamic reordering of memory references [J].
Franklin, M ;
Sohi, GS .
IEEE TRANSACTIONS ON COMPUTERS, 1996, 45 (05) :552-571
[7]   Tradeoffs in buffering memory state for thread-level speculation in multiprocessors [J].
Garzarán, MJ ;
Prvulovic, M ;
Llaberiá, JM ;
Viñals, V ;
Rauchwerger, L ;
Torrellas, J .
NINTH INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE COMPUTER ARCHITECTURE, PROCEEDINGS, 2003, :191-202
[8]   Speculative versioning cache [J].
Gopal, S ;
Vijaykumar, TN ;
Smith, JE ;
Sohi, GS .
1998 FOURTH INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE COMPUTER ARCHITECTURE, PROCEEDINGS, 1998, :195-205
[9]  
GUPTA M, 1998, P SUP 1998 NOV
[10]  
HAGERSTEN E, 1999, P 5 INT S HIGH PERF