Achilleas Tzenetopoulos, C. Marantos, Giannos Gavrielides, S. Xydis, Dimitrios Soudris
{"title":"FADE","authors":"Achilleas Tzenetopoulos, C. Marantos, Giannos Gavrielides, S. Xydis, Dimitrios Soudris","doi":"10.2307/j.ctt20q1xt6.20","DOIUrl":"https://doi.org/10.2307/j.ctt20q1xt6.20","url":null,"abstract":"","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132926541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Achilleas Tzenetopoulos, C. Marantos, Giannos Gavrielides, S. Xydis, D. Soudris
{"title":"FADE: FaaS-inspired application decomposition and Energy-aware function placement on the Edge","authors":"Achilleas Tzenetopoulos, C. Marantos, Giannos Gavrielides, S. Xydis, D. Soudris","doi":"10.1145/3493229.3493306","DOIUrl":"https://doi.org/10.1145/3493229.3493306","url":null,"abstract":"Lately, more and more applications are deployed on heterogeneous, power-constrained edge-computing devices. Bringing computation closer to the data, contributes both to latency and energy consumption reduction due to the elimination of excessive data transfers. However, while the main concern in such environments is the minimization of energy consumption, the heterogeneity in compute resources found at the edge may lead to Quality of Service (QoS) violations. At the same time, Serverless computing, the next frontier of Cloud computing has emerged to offer unprecedented elasticity by utilizing fine-grained, stateless functions. The reduction in the execution time and the modest memory footprint of such decomposed applications, allow for fine-grained resource multiplexing. In this work, we propose a methodology for application decomposition into fine-grained functions and energy-aware function placement on a cluster of edge devices subject to user-specified QoS guarantees.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"125 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124199766","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Efficient Application of Tensor Core Units for Convolving Images","authors":"Stefan Groth, Jürgen Teich, Frank Hannig","doi":"10.1145/3493229.3493305","DOIUrl":"https://doi.org/10.1145/3493229.3493305","url":null,"abstract":"Tensor Core Units (TCUs) are a domain-specific architecture capable of executing small Matrix Multiply-Accumulates (MMAs) in a single clock cycle, showing significant performance improvements over other optimized implementations. When Convolutional Neural Networks (CNNs) are accelerated using TCUs, the layout of the input image is transformed to allow large amounts of filters to be applied to an image using a single large matrix-matrix multiplication. However, there are applications in other domains that only require a small number of filters. To accommodate such applications, we first show the inappropriateness of this standard technique of transforming the data layout. Subsequently, we propose an approach that uses TCUs to convolve one filter with an image. Further, we introduce several optimizations of this method. Finally, we evaluate the performance of our approach and its optimizations by comparing it to code generated using a state-of-the-art image processing language.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129264316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"So Far So Good: Self-Adaptive Dynamic Checkpointing for Intermittent Computation based on Self-Modifying Code","authors":"Bahram Yarahmadi, Erven Rohou","doi":"10.1145/3493229.3493300","DOIUrl":"https://doi.org/10.1145/3493229.3493300","url":null,"abstract":"Recently, different software and hardware based checkpointing strategies have been proposed to ensure forward progress toward execution for energy harvesting IoT devices. In this work, inspired by the idea used in dynamic compilers, we propose SFSG: a dynamic strategy, which shifts checkpoint placement and specialization to the runtime and takes decisions based on the past power failures and execution paths taken before each power failure. The goal of SFSG is to provide forward progress and to avoid facing non-termination without using hardware features or programmer intervention. We evaluate SFSG on a TI MSP430 device, with different types of benchmarks as well as different uninterrupted intervals, and we evaluate it in terms of the number of checkpoints and its runtime overhead.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116929716","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Lorenzo Chelini, Martin Kong, T. Grosser, Henk Corporaal
{"title":"LoopOpt","authors":"Lorenzo Chelini, Martin Kong, T. Grosser, Henk Corporaal","doi":"10.1145/3493229.3493301","DOIUrl":"https://doi.org/10.1145/3493229.3493301","url":null,"abstract":"Despite years of research, the optimization strategy of loop-level optimization frameworks remains fragile when addressing modern and heterogeneous architectures. Furthermore, optimizers act as an opaque operation, a black-box, to the users, forcing them to tedious an error-prone manual optimization if imprecise cost models are used. Current solutions, to drive loop-level optimizers rely on pragmas or bake the transformations recipes in the source code using imperative embedded scripting. But, the optimization of programs via a sequence of imperative directives is unlikely to solve this problem fully as expressing optimization is still an error-prone and time-consuming task for the users. The ideal solution would be a declarative approach that allows the users to opt-in if the optimizer has made a poor optimization decision but avoid baking the transformation script within a given application or bind it to a particular loop nest. Based on such an idea, we propose LoopOpt, an interactive tool that enables users to design optimizations in partnership with the compiler in a declarative way. Thus, our approach opens the polyhedral black-box allowing users to design complex optimizations sequences in a declarative way.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"226 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115378460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"How to exploit sparsity in RNNs on event-driven architectures","authors":"J. Brils, Luc Waeijen, Arash Pourtaherian","doi":"10.1145/3493229.3493302","DOIUrl":"https://doi.org/10.1145/3493229.3493302","url":null,"abstract":"Event-driven architectures have been shown to provide low-power, low-latency artificial neural network (ANN) inference. This is especially beneficial on Edge devices, particularly when combined with sparse execution. Recurrent neural networks (RNNs) are ANNs that emulate memory. Their recurrent connection enables the reuse of previous output for the generation of new output. However, when trying to use RNNs in a sparse context on event-driven architectures, novel challenges in synchronization and the usage of sparse data are encountered. In this work, these challenges are systematically analyzed, and mechanisms to overcome them are proposed. Experimental results of a monocular depth estimation use case on the NeuronFlow architecture show that sparsity in RNNs can be exploited effectively on event-driven architectures.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133732595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Predicting Objectives on a Reduced Search Space of Multiobjective Function Inlining","authors":"Kateryna Muts, H. Falk","doi":"10.1145/3493229.3493303","DOIUrl":"https://doi.org/10.1145/3493229.3493303","url":null,"abstract":"The Worst-Case Execution Time (WCET), energy consumption, and code size are among the most important criteria of hard real-time systems. To estimate the WCET and energy consumption at compile time, static analyzers are often used: they estimate the objectives by invoking time-consuming microarchitecture, data flow, and control flow analyses. The expensive analyses make it almost infeasible to use evolutionary algorithms for solving multiobjective problems with these two objectives at compile time, since any evolutionary algorithm extensively evaluates objectives to find solutions. We propose a method that speeds up an evolutionary algorithm supplying it with a reduced search space and prediction model fitted on the reduced search space, so the algorithm needs to explore a smaller search space and can use fast predictions instead of time-consuming estimations to evaluate the WCET and energy consumption. The proposed approach is general enough to be used for any compiler-based optimization. We demonstrate the advantages of it solving a multiobjective function inlining problem at compile time.","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130153679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"lospre in linear time","authors":"P. K. Krause","doi":"10.1145/3493229.3493304","DOIUrl":"https://doi.org/10.1145/3493229.3493304","url":null,"abstract":"Lifetime-optimal speculative partial redundancy elimination (lospre) is the most advanced currently known redundancy elimination technique. It subsumes many previous approaches, such as common subexpression elimination, global common subexpression elimination, and loop-invariant code motion. Previously known lospre algorithms have high time complexity; faster but less powerful approaches have been used and developed further instead. We present a simple linear-time algorithm for lospre for structured programs that can also handle some more general scenarios compared to previous approaches. We prove that our approach is optimal and that the runtime is linear in the number of nodes in the control-flow graph. The condition on programs of being structured is true for many programming languages and for others, such as C, is equivalent to a bound on the number of goto labels per function. An implementation in a mainstream C compiler demonstrates the practical feasibility of our approach, which is based on graph-structure theory and uses tree-decompositions. For structured programs, we also improve the runtime bound of deterministic implementations of the previously known MC-PRE and MC-SSAPRE algorithms from O(n3) to O(n2.5).","PeriodicalId":420586,"journal":{"name":"Proceedings of the 24th International Workshop on Software and Compilers for Embedded Systems","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125375095","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}