Nguyen, J., "Compiler Analysis to Implement Point-to-Point Synchronization in Parallel Programs," Ph.d. Thesis MIT, MIT-LCS TR-595. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Using Prefetching to Hide Lock Acquisition Latency in.. - Karlsson, Stenström (1996) (Correct)
....Our work targets the more general locks while fuzzy barriers improve the performance of barriers only. Another approach to reduce synchronization overhead is to plainly remove synchronizations or by reducing barriers to point to point synchronizations. This has been done by for example Nguyen in [18]. His approach was to statically analyze the program with a compiler and then replace the barriers in the program with less expensive point to point synchronizations. This approach could be used in conjunction with lock prefetching. Nguyen s approach could be used to remove as many ....
Nguyen, J., "Compiler Analysis to Implement Point-to-Point Synchronization in Parallel Programs," Ph.d. Thesis MIT, MIT-LCS TR-595.
Reactive Synchronization Algorithms for Multiprocessors - Lim (1995) (40 citations) (Correct)
....the program to use fine grained synchronization instead of barriers improves performance by a factor of three due to increased parallelism. Hardware support for fine grained, data level synchronization in the form of full empty bits [53] yields an additional 40 performance improvement. Nguyen [46] used compiler analysis to transform statically partitioned DOALL loops to use point to point communication between processors instead of global barriers. Conventional implementations of DOALL loops use a barrier at the end of each DOALL loop to enforce data dependencies across DOALL loops. ....
John Nguyen. Compiler Analysis to Implement Point-To-Point Synchronization in Parallel Programs. PhD thesis, Massachusetts Institute of Technology, Cambridge, MA, August 1993.
Impact of Load Imbalance on the Design of Software Barriers - Eichenberger, Abraham (1995) (5 citations) (Correct)
....detect and increase the number of independent operations, and hence the slack time. Eichenberger and Abraham [13] characterized the performance improvements due to fuzzy barriers and showed that the expected idle time at a fuzzy barrier is inversely proportional to the slack time. Finally, Nguyen [14] investigated compiler techniques that transform synchronization barriers into point II 64 to point synchronizations, showing encouraging performance improvements. The source and extent of variation of thread (processor) execution times have been investigated in a few studies. Adve and Vernon ....
J. Nguyen, Compiler Analysis to Implement Point-to-Point Synchronization in Parallel Programs, PhD thesis, MIT, August 1993.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC