Filters








3,830 Hits in 4.0 sec

Loop transformations

Louis-Noël Pouchet, Uday Bondhugula, Cédric Bastoul, Albert Cohen, J. Ramanujam, P. Sadayappan, Nicolas Vasilache
2011 Proceedings of the 38th annual ACM SIGPLAN-SIGACT symposium on Principles of programming languages - POPL '11  
However, determining appropriate compositions of loop transformations to achieve this remains a significantly challenging task; current compilers may achieve significantly lower performance than hand-optimized  ...  To address this fundamental challenge, we first present a convex characterization of all distinct, semantics-preserving, multidimensional affine transformations.  ...  Acknowledgments We are grateful to Paul Feautrier for providing the foundations for multidimensional affine scheduling, and the simplification of Vasilache's original convex formulation of all semantics-preserving  ... 
doi:10.1145/1926385.1926449 dblp:conf/popl/PouchetBBCRSV11 fatcat:h5y7lpv26fdhjcrmemweqffnxu

Combined loop transformation and hierarchy allocation for data reuse optimization

Jason Cong, Peng Zhang, Yi Zou
2011 2011 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)  
This paper presents a combined approach which optimizes loop transformation and memory hierarchy allocation simultaneously to achieve global optimal results on external memory bandwidth and on-chip data  ...  High-level synthesis; loop transformation; memory hierarchy optimization; data reuse I.  ...  Data reuse optimization finds the optimal loop transformation and on-chip buffer allocation.  ... 
doi:10.1109/iccad.2011.6105324 dblp:conf/iccad/CongZZ11 fatcat:4gmi6ouh5vgzdljdkk5nddggee

Optimizing memory hierarchy allocation with loop transformations for high-level synthesis

Jason Cong, Peng Zhang, Yi Zou
2012 Proceedings of the 49th Annual Design Automation Conference on - DAC '12  
This paper presents an efficient approach for optimizing the on-chip memory allocation by loop transformations in the imperfectly nested loops.  ...  We analytically model the on-chip buffer size and off-chip bandwidth after affine loop transformation, loop fusion/distribution and code motion.  ...  , which uses the branch-and-bound approach to prune the sub-optimal transformation space and the knapsack reuse technique to reduce the complexity of each transformation.  ... 
doi:10.1145/2228360.2228586 dblp:conf/dac/CongZZ12 fatcat:hzsuwniftjbs3j3tq3rxfwyhf4

Simplification and runtime resolution of data dependence constraints for loop transformations

Diogo N. Sampaio, Louis-Noël Pouchet, Fabrice Rastello
2017 Proceedings of the International Conference on Supercomputing - ICS '17  
a test to be evaluated at runtime. is test allows to determine whether the transformation is valid, and if so triggers the execution of the transformed code, falling back to the original code otherwise  ...  Loop transformations such as tiling, parallelization or vectorization are essential tools in the quest for high-performance program execution.  ...  Prior to performing FML, a conjunctive system undergoes a few transformations: (1) Convex Hull: e system is enriched with an octagon convex hull.  ... 
doi:10.1145/3079079.3079098 dblp:conf/ics/SampaioPR17 fatcat:cd7ar6k465aitbx7jtm6h5odye

Generation of Distributed Loop Control [chapter]

Marcus Bednara, Frank Hannig, Jürgen Teich
2002 Lecture Notes in Computer Science  
Our methodology is applicable to one-and multi-dimensional processor arrays and is based on local control signal propagation.  ...  Each control element thereby propagates a "start" and a "stop execution" signal to its neighbors.  ...  Similar approaches are also used in the area of parallelizing compilers for supercomputers with linear transformations such as loop skewing, loop tiling, and loop permutation [19] .  ... 
doi:10.1007/3-540-45874-3_9 fatcat:qplvjhoqcvhw5hmuqvbihr23qe

Accelerating Human-in-the-loop Machine Learning

Doris Xin, Litian Ma, Jialin Liu, Stephen Macke, Shuchen Song, Aditya Parameswaran
2018 Proceedings of the Second Workshop on Data Management for End-To-End Machine Learning - DEEM'18  
We propose Helix, a machine learning system that optimizes the execution across iterationsintelligently caching and reusing, or recomputing intermediates as appropriate.  ...  Machine learning workflow development is a process of trial-and-error: developers iterate on workflows by testing out small modifications until the desired accuracy is achieved.  ...  data via loop fusion.  ... 
doi:10.1145/3209889.3209897 dblp:conf/sigmod/XinMLMSP18 fatcat:3arqg55lyvfmdngqy4cupmyimm

Topological Transformation Approaches to Database Query Processing

Alok Watve, Sakti Pramanik, Salman Shahid, Chad R. Meiners, Alex X. Liu
2015 IEEE Transactions on Knowledge and Data Engineering  
For larger than two dimensions, we propose a space transformation scheme based on disjoint planer rotation and a new type of query, pruning box query, to get the precise results.  ...  Experimental results with large synthetic databases and some real databases show the effectiveness of the proposed transformation scheme.  ...  Chad Meiners worked on this project when he was a PostDoc in the Department of Computer Science and Engineering, Michigan State University. Professor Sakti Pramanik is the corresponding author.  ... 
doi:10.1109/tkde.2014.2363658 fatcat:zjy3lnb3czfo3movgechueryfi

Loop Leaping with Closures [chapter]

Sebastian Biallas, Jörg Brauer, Andy King, Stefan Kowalewski
2012 Lecture Notes in Computer Science  
This paper proposes a simple and an efficient approach for loop leaping for these domains based on viewing loops as closure operators.  ...  Loop leaping is the colloquial name given to a form of program analysis in which summaries are derived for nested loops starting from the innermost loop and proceeding in a bottom-up fashion considering  ...  We conclude that these invariants appear to be optimal even though the closure semantics can, in principle, lead to a sub-optimal transformer for loops.  ... 
doi:10.1007/978-3-642-33125-1_16 fatcat:5o4ac2b3kbfi3a4hovidbc5qju

Data-Driven Loop Bound Learning for Termination Analysis

Rongchen Xu, Jianhui Chen, Fei He0001
2022 International Conference on Software Engineering  
We present efficient data-driven algorithms for inferring various kinds of loop bounds, including simple loop bounds, conjunctive loop bounds, and lexicographic loop bounds.  ...  This paper employs a reinforced black-box learning approach for termination proving, consisting of a loop bound learner and a validation checker.  ...  Convex Optimization and Set Covering Problem Given an objective function f (x), the convex optimization problem [5] is to find an optimum x * that minimizes f (x) and satisfies all constraints, i.e.,  ... 
doi:10.1145/3510003.3510220 dblp:conf/icse/XuCH22 fatcat:6kwuh7e2ijbenmjyb7blp24sha

When polyhedral transformations meet SIMD code generation

Martin Kong, Richard Veras, Kevin Stock, Franz Franchetti, Louis-Noël Pouchet, P. Sadayappan
2013 SIGPLAN notices  
Data locality and parallelism are critical optimization objectives for performance on modern multi-core machines.  ...  ., multi-core) and fine-grain parallelism (e.g., vector SIMD) must be effectively exploited, but despite decades of progress at both ends, current compiler optimization schemes that attempt to address  ...  Army through contract W911NF-10-1-0004, and by Intel ECG.  ... 
doi:10.1145/2499370.2462187 fatcat:dzup6ptj6fdexpizsuow5xtpbi

When polyhedral transformations meet SIMD code generation

Martin Kong, Richard Veras, Kevin Stock, Franz Franchetti, Louis-Noël Pouchet, P. Sadayappan
2013 Proceedings of the 34th ACM SIGPLAN conference on Programming language design and implementation - PLDI '13  
Data locality and parallelism are critical optimization objectives for performance on modern multi-core machines.  ...  ., multi-core) and fine-grain parallelism (e.g., vector SIMD) must be effectively exploited, but despite decades of progress at both ends, current compiler optimization schemes that attempt to address  ...  Army through contract W911NF-10-1-0004, and by Intel ECG.  ... 
doi:10.1145/2491956.2462187 dblp:conf/pldi/KongVSFPS13 fatcat:2cqpl2tqobcjvbdlpoarmfbx3u

Neural architecture search as program transformation exploration

Jack Turner, Elliot J. Crowley, Michael F. P. O'Boyle
2021 Proceedings of the 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems  
This allows them to be combined with existing transformations into a unified optimization framework.  ...  We prototyped the combined framework in TVM and were able to find optimizations across different DNNs, that significantly reduce inference time -over 3× in the majority of cases.  ...  Optimization Description Program Transformations reorder Interchange nested loops tile Cache and register blocking unroll Loop unrolling prefetch Memory coalescing between threads split Divide  ... 
doi:10.1145/3445814.3446753 fatcat:mchoutirjja3nerrrkipt2vvxa

Complete maps of molecular-loop conformational spaces

Josep M. Porta, Lluís Ros, Federico Thomas, Francesc Corcho, Josep Cantó, Juan Jesús Pérez
2007 Journal of Computational Chemistry  
Experiments are included that show the method's performance on rigid loops, mobile loops, and multi-loop molecules.  ...  loop closure.  ...  Cajal and I3 program funds supporting the first two authors.  ... 
doi:10.1002/jcc.20863 fatcat:ca57cwpe3vdgbedq7tzokbztxq

Complete maps of molecular-loop conformational spaces

Josep M. Porta, Lluís Ros, Federico Thomas, Francesc Corcho, Josep Cantó, Juan Jesús Pérez
2007 Journal of Computational Chemistry  
Experiments are included that show the method's performance on rigid loops, mobile loops, and multi-loop molecules.  ...  loop closure.  ...  Cajal and I3 program funds supporting the first two authors.  ... 
doi:10.1002/jcc.20733 pmid:17450561 fatcat:tmdxi45425gadkefs4dw7vhmki

Human-in-the-loop Techniques in Machine Learning

Chengliang Chai, Guoliang Li
2020 IEEE Data Engineering Bulletin  
In this paper, we first summarize the human-in-the-loop techniques in machine learning, including: (1) Data Extraction: Non-structured data always needs to be transformed to structured data for feature  ...  Human-in-the-loop techniques are playing more and more significant roles in the machine learning pipeline, which consists of data preprocessing, data labeling, model training and inference.  ...  Second, optimization methods focus on designing optimization functions that capture the relations between humans' qualities and tasks' truth, and then provide an iterative method to compute these two sets  ... 
dblp:journals/debu/Chai020 fatcat:okdxp6ml2fcmvnkha36i2lcm6u
« Previous Showing results 1 — 15 out of 3,830 results