August D. I., "Hyperblock performance optimizations for ILP processors", M.S. thesis, Department of Electrical and ComputerEngineering, University of Illinois, Urbana, IL, 1996. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
A Technique for Variable Dependence Driven Loop Peeling - Song, Kavi (2002) (Correct)
....techniques, loop peeling and loop invariant code motion have widely been studied and used in compilers. For a survey of these and other source level optimization can be found in [3] Loop peeling was originally mentioned in [12] and automatic loop peeling techniques were discussed in [13] August [2] showed how loop peeling could be applied in practice, and elucidated how this optimization alone may not increase program performance, but may expose opportunities for other optimization leading to performance improvements. August used only heuristic loop peeling techniques. We feel that when ....
August D. I., "Hyperblock performance optimizations for ILP processors", M.S. thesis, Department of Electrical and ComputerEngineering, University of Illinois, Urbana, IL, 1996.
Structural And Static Analysis Techniques For Enhancing Compiler.. - Crozier (1999) Self-citation (August) (Correct)
....loops that execute infrequently are one important exception. These loops do not contain enough iterations to achieve the desired level of ILP using the unrolling techniques. Loop peeling is an optimization that peels off iterations of an inner loop that are nested within the selected blocks [47]. An algorithm for performing loop peeling is given in Figure 3.3. These iterations can then be overlapped with the instructions from the surrounding blocks to increase ILP. The loop is usually peeled enough times so that the majority of the invocations just execute the peeled code. The peeled ....
D. I. August, "Hyperblock performance optimizations for ILP processors," M.S. thesis, Dept. of Electrical and Computer Engineering, University of Illinois, Urbana, IL, 1996.
Condition Awareness Support For Predicate Analysis And Optimization - Sias (1999) (1 citation) Self-citation (August) (Correct)
....reduction, loop induction variable elimination, and loop induction variable reassociation [16] Also during the Lcode phase, code is restructured into superblocks [17] which are not predicated, or into hyperblocks [18] which use predication. Superblock specific [16] and hyperblock specific [19] transformations are performed to exploit the code optimization and scheduling benefits of these two ILP compilation models. Lcode based classical optimizations are also performed iteratively after the formation of superblocks and hyperblocks, since more opportunities for their application are ....
.... This is convenient, since entry of the hyperblock region guarantees that the execution conditions of each instruction will be determined solely by guarding predicates (with the slight qualification that in partially resolved hyperblocks, side exit branches also guard instruction execution) [19]. Thus, discerning the logical relationships between predicates in the hyperblock does not involve control flow analysis outside the hyperblock region or analysis around backedges, which would dilute accuracy and require some level of iteration. There exists, however, the possibility of extending ....
D. I. August, "Hyperblock performance optimizations for ILP processors," M.S. thesis, Dept. of Electrical and Computer Engineering, University of Illinois, Urbana, IL, 1996.
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