The mechanical parallelization of loop nests containing while loops
نویسنده
چکیده
Acknowledgments Nobody can write a thesis without help form others, and it is usually impossible to express one's gratitude for this immense amount of help. The least I can do is to devote the rst pages of my thesis to all these wonderful people, and thank them all for their precious support and individual help. I want to mention some people explicitly, even knowing that my list must be inclomplete. First of all, I want to thank Professor Christian Lengauer who has been an excellent advisor to me. Thank you for my position, for your liberality concerning working modes, for uncountably many fruitful discussions with you (oocial and private), for your indefatigability in improving my English, for multiple detailed proof readings of this thesis, for always having time for my problems, ...; short, thank you for having been a real \Doktorvater", which, to me, is more than an advisor. In addition, I am grateful to Professor Paul Feautrier: thank you that you have accepted to review this thesis, and took the time to give me detailed comments on my draft of this thesis. I also want to thank Professor P. Kleinschmidt, Professor F.-J. Brandenburg and Professor W. Hahn for having agreed to being my examiners and for their helpfulness. Furthermore, I would like to thank Professor N. Schwartz who always helps at the formal aspects on the way to a Ph. D. However, there are also helpful persons outside of my dissertation committee. First of all, I want to thank my French colleague and friend Jean-Frann cois Collard. Thank you for your cooperation already at the beginning of this thesis, when we did not yet know each other. Because of your open-minded way, we succeeded in working together instead of being competitors. This led to many fruitful discussions and a deep friendship. Thanks a lot for that. Furthermore, I would like to thank the members of the Lehrstuhl f ur Programmierung for the good working climate and for some helpful hints, and esp. Christoph Herrmann for his excellent proof reading. In addition, there is another member of the group I want to mention speciically: Ulrike Lechner. I would call her \the good soul of our group". Thank you for sharing the ooce and some work, and for the wonderful climate in our ooce, not only due to your owers. A-pro-pos climate: one of the most agreeable teams I have ever …
منابع مشابه
On the Parallelization of Loop Nests Containing while Loops
Recently, eeorts have been made to devise automatic methods, based on a mathematical model, for the par-allelization of loop nests with while loops. These methods are extensions of methods for the parallelization of nested for loops.
متن کاملA Model for Systolic Parallelization of Multiple Loop Programs
The model presented here for systolic parallelization of programs with multiple loops aims at compilation of sequential programs for parallel computer architectures which support ne-grained communication. Regular loop nests in programs can be subjected to space-time loop transformations as has been examined in the area of systolic parallelization. The aim of this work is to show how this kind o...
متن کاملAffine Transformations for Communication Minimized Parallelization and Locality Optimization of Arbitrarily Nested Loop Sequences
A long running program often spends most of its time in nested loops. The polyhedral model provides powerful abstractions to optimize loop nests with regular accesses for parallel execution. Affine transformations in this model capture a complex sequence of execution-reordering loop transformations that improve performance by parallelization as well as better locality. Although a significant am...
متن کاملAffine Transformations for Communication Minimal Parallelization and Locality Optimization of Arbitrarily Nested Loop Sequences
A long running program often spends most of its time in nested loops. The polyhedral model provides powerful abstractions to optimize loop nests with regular accesses for parallel execution. Affine transformations in this model capture a complex sequence of execution-reordering loop transformations that improve performance by parallelization as well as better locality. Although a significant am...
متن کاملGeneration of Synchronous Code for Automatic Parallelization of while Loops
Automatic parallelization of imperative programs has focused on nests of do loops with aane bounds and aane dependences, because in this case execution domains and dependences are precisely known at compile-time. Parallelization can then be done using a suitable space-time transformation , yielding a logically synchronous program. Code generation consists of scanning the transformed execution d...
متن کاملCascaded Execution: Speeding Up Unparallelized Execution on Shared-Memory Multiprocessors
Both inherently sequential code and limitations of analysis techniques prevent full parallelization of many applications by parallelizing compilers. Amdahl’s Law tells us that as parallelization becomes increasingly effective, any unparallelized loop becomes an increasingly dominant performance bottleneck. We present a technique for speeding up the execution of unparallelized loops by cascading...
متن کامل