Félix Goudreault, Dominik Scheuble, Mario Bijelic et al. · 2023
LiDAR has become a cornerstone sensing modality for 3D vision. LiDAR systems emit pulses of light into the scene, take measurements of the returned signal, and rely on hardware digital signal processing (DSP) pipelines to construct 3D point clouds from these measurements. The resulting point clouds output by these DSPs are input to downstream 3D vision models - both, in the form of training datasets or as input at inference time. Existing LiDAR DSPs are composed of cascades of parameterized operations; modifying configuration parameters results in significant changes in the point clouds and consequently the output of downstream methods. Existing methods treat LiDAR systems as fixed black boxes and construct downstream task networks more robust with respect to measurement fluctuations. Departing from this approach, the proposed method directly optimizes LiDAR sensing and DSP parameters for downstream tasks. To investigate the optimization of LiDAR system parameters, we devise a realistic LiDAR simulation method that generates raw waveforms as input to a LiDAR DSP pipeline. We optimize LiDAR parameters for both 3D object detection IoU losses and depth error metrics by solving a nonlinear multi-objective optimization problem with a 0th-order stochastic algorithm. For automotive 3D object detection models, the proposed method outperforms manual expert tuning by 39.5% mean Average Precision (mAP).
Cited by 3
2026
Cited by 0
Youcef Bouchebaba, Valérie Gagné, Gabriela Nicolescu et al. · 2006
In today's embedded systems, the memory hierarchy is rapidly becoming a major factor in terms of power, performance and area. This is especially true for embedded multimedia applications using temporary multi-dimensional arrays that are typically used to store intermediate results during multimedia processing. In this paper, we introduce a new buffer allocation method to replace these temporary arrays and we combine it with loop fusion and tiling. The simple and effective method we present simultaneously applies tiling with fusion to a set of loop nests. Then, it replaces temporary arrays with smaller buffers containing the useful data. These new techniques allow to optimize memory space and reduce the number of cache misses. Our buffer allocation method is implemented in the PIPS compiler and the experiments are made on the StepNP simulator. They show that our technique yields a significant reduction in the number of data cache misses (on average, the data cache miss ratio is decreased by 14.3%)
Cited by 0
Ondřej Kubánek · Digital Repository (National Repository of Grey Literature) · 2023
Production compilers use numerous techniques to generate performant code. One such technique is Profile-guided optimization (PGO). The princi- ple of this technique is to insert instrumentation during compilation, gather information about program behaviour with training runs and use this infor- mation during recompilation to improve optimization. The thesis aims to improve the precision of Loop optimizations in GNU Compiler Collection (GCC) with PGO. Currently in GCC, only the average iteration count of a loop is known with PGO. This leads to inefficiencies in both the performance and size of the binary. We implement infrastructure for measuring more information about loop iterations and add new counters namely the histogram of iterations and his- togram of iterations modulo its size. With the histogram of iterations, we improve loop peeling and implement a new case of loop versioning optimiza- tion. This significantly improves the performance of the generated code with reasonable overhead.
Cited by 0
Uday Bondhugula, Albert Cohen, J. Ramanujam et al. · 2011
High-level loop transformations are a key instrument in mapping computational kernels to effectively exploit resources in modern processor architectures. However, determining appropriate compo-sitions of loop transformations to achieve this remains a signif-icantly challenging task; current compilers may achieve signifi-cantly lower performance than hand-optimized programs. To ad-dress this fundamental challenge, we first present a convex char-acterization of all distinct, semantics-preserving, multidimensional affine transformations. We then bring together algebraic, algorith-mic, and performance analysis results to design a tractable opti-mization algorithm over this highly expressive space. The frame-work has been implemented and validated experimentally on a representative set of benchmarks run on state-of-the-art multi-core platforms. Categories and Subject Descriptors D 3.4 [Programming lan-
Cited by 0
Louis-noël Pouchet, Uday Bondhugula, Cédric Bastoul et al. · NOT FOUND REPOSITORY (Indian Institute of Science Bangalore) · 2011
High-level loop transformations are a key instrument in mapping computational kernels to effectively exploit resources in modern processor architectures. 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 programs. To address this fundamental challenge, we first present a convex characterization of all distinct, semantics-preserving, multidimensional affine transformations. We then bring together algebraic, algorithmic, and performance analysis results to design a tractable optimization algorithm over this highly expressive space. The framework has been implemented and validated experimentally on a representative set of benchmarks run on state-of-the-art multi-core platforms.
Cited by 4Open access
Wu, Jianghua. · DR-NTU (Nanyang Technological University) · 2012
This report presents the process of implementation and optimization of two image resize algorithms namely, Bilinear and Bicubic Interpolation. The purpose of the optimization seeks to improve execution time and is primarily done with the use of Nvidia’s Compute Unified Device Architecture (CUDA). Both Algorithms are implemented in C++ before subsequent CUDA codes are added. The challenge in the project is to pick up CUDA programming and also the requirement of understanding the math involved before converting into algorithms. It was evident how the integration of CUDA, by substituting the use of loops in computations with threads running in parallel demonstrated a significant speed up in execution time. There is still room for code refactoring, better CUDA implementation and use of more powerful of Graphics Processing Unit (GPU) that will see improvements to both design and greater optimization of the developed application. In conclusion, the project has shown that under certain conditions, the leveraging the power of by the use of CUDA is a viable optimization tool in Graphics Processing Algorithms such as the ones mentioned above.
Cited by 0
Laboratoire de l'informatique du parallélisme, Fraboulet, Antoine, Huard, Guillaume et al. · LARA - Libre Acces aux Rapports scientifiques et techniques · 2006
(eng) Portable or embedded systems allow more and more complex applications like multimedia today. These applications and submicronic technologies have made the power consumption criterium crucial. We propose new techniques thanks to which we can optimize the behavioral description of an integrated system before the hardware/software partitioning (Codesign). These transformations are performed on ``for'' loops that constitute the main parts of the multimedia code which handle the arrays. We present in this paper two new (polynomial) techniques for minimizing memory accesses in loop nests by data temporal locality optimization.
Cited by 0
Antoine Fraboulet, Karen Kodary, Anne Mignotte · 2001
Portable or embedded systems as well as submicronic technologies have made the power consumption criterium crucial. Memory is known to be extremely power consuming. Moreover multimedia applications are memory intensive applications. Therefore, we propose new techniques to optimize a behavioral description of multimedia applications before the hardware/software partitioning (Codesign). These transformations are performed on "for" loops that constitute the main parts which handle the arrays of the multimedia code. This paper presents an optimal algorithm to reduce the use of temporary arrays by loop fusion. Although the algorithm is not polynomial, experiments have shown that it is very efficient.
Cited by 3
Liu, D., Giloi, W.K. · Publikationsdatenbank der Fraunhofer-Gesellschaft (Fraunhofer-Gesellschaft) · 1989
S.186-190
Cited by 0
Srinivasan Govindan, Srikrishnan Venkataraman, Beomseok Choi et al. · 2024
Fully Integrated Voltage Regulators (FIVR) are integral part of the power delivery network of high-performance server microprocessor chips. FIVR is a switched-inductor, multiphase on-chip buck voltage regulator. FIVR is also a closed loop control system that regulates the output voltage. The FIVR control loop consists of a Type-III op-amp compensator (CPS) that has programmable passive RC components. The CPS RC values need to be chosen from a wide range to optimize the performance of the FIVR control loop. In this paper Evolutionary Multi-objective Optimization (EMO) Algorithms are proposed for the optimization of FIVR control loop. The multi-objective functions and the constraints for optimizing the FIVR control loop are developed. Examples demonstrate that the EMO methods perform better than the existing Single-objective Optimization (SO) methods.
Cited by 0
Ross, Isaac Michael, Karpenko, Mark, Proulx, Ronald J. · Calhoun: The Naval Postgraduate School Institutional Archive (Naval Postgraduate School) · 2018
The disclosure provides a closed-loop controller for a controlled system comprising a comparison element generating an error e, a compensator generating a control uɴ value based on the error e, and a control allocator determining a manipulated parameter uᴍ value based on the control uɴ. The control allocator typically utilizes a control effectiveness function and determines uᴍ value by selecting one or more specific system xₒ signals from the system state xᵢ or system input yⱼ values or system parameters values Pk reported, defining a plurality of distributed XD around each specific system xₒ signal, and minimizing an error function E(zᵢ), where the error function E(zᵢ) is based on errors which arise from use of the plurality of distributed XD in the control effectiveness function rather than one or more specific system xₒ signals.
Cited by 0Open access
Jinyang Yao, Lili Liu, Xuanyu Fu et al. · Research Square · 2024
<title>Abstract</title> SIMD extensions are playing an increasingly important role in high-performance computing and artificial intelligence fields. To fully utilize these components, various manufacturers and institutions have implemented many optimizations for SIMD extensions, with dual SIMD extension pipeline optimization being one of them. This method generates instructions suitable for parallel execution of the integrated dual SIMD extension on processors by unrolling vectorizable loops in programs. It is integrated into a mainstream compiler GCC as an optimization pass and can be enabled with just one compilation option. Experiments were conducted on an SW421 processor, testing standard benchmark suites such as SPEC CPU 2006 and NPB. The experiments showed that after using this optimization pass, programs generated by the compiler can fully utilize the dual SIMD extension during execution. Compared with turning on the autovectorization option, the method has an acceleration effect on multiple applications in the test set, and the execution efficiency is improved by an average of 5.6% and a maximum of 14%.
Cited by 0Open access
Halilovic, Smajil, Hamacher, Thomas 1964-, Technische Universität München · mediaTUM (Technical University of Munich) · 2023
This thesis presents a set of novel approaches for efficient and effective optimization of open-loop shallow geothermal systems. The approaches enable the optimization of individual systems and multiple neighboring systems, addressing key aspects such as optimal well placement and sizing. The thesis provides valuable tools to a wide range of stakeholders, especially researchers and practitioners engaged in thermal groundwater management.
Cited by 0Open access
Chunhong Wang, Gaoming Li, Albert C. Reynolds · Proceedings of SPE Annual Technical Conference and Exhibition · 2007
Cited by 20
Tomoshi Kono · JAPAN TAPPI JOURNAL · 2015
近年，プラント操業の現場において，主にDCS制御の機能向上により少人数でのプラント操業が可能となり，合理化や省力化が進んできている。また，ベテランのリタイアでプラントの操業ノウハウの継承も難しくなってきており，操業オペレータや保守担当者の負担は増すばかりである。そのような状況の中，プラントの設備改善・最適化に目を向けるのは困難であろう。またプラントの最適化はデータを収集，データの検証，改善のアクションと大変工数のかかる作業である。そこで制御ループを連続で監視し，プラントの最適化に必要な制御ループの問題点を自動で解析する機能を持った「Metso PlantTriageプラントトリアージ」の機能と漂白プラントで薬品消費量の削減に成功した事例を紹介する。漂白プラントでは特にパルプ流量の変動は後段のタワーpH，タワー温度，等ブリーチ制御に影響し，薬品添加量に大きく関わる。従来のプラント最適化の方法ではトレンドデータを収集，データ検証，改善アクションと大変工数のかかる作業である。また相互干渉を伴った制御ループの改善は原因を特定するのが大変困難である。プラントトリアージのループ価値による改善対象の優先順付けの機能，制御ループの相互干渉マップ機能等パワフルなツールを活用し，短時間で効率的に漂白プラントの薬品削減に成功している。＊「メッツォプラントトリアージ」は，社名等変更に伴い2015年1月1日からメッツォジャパン(株)（Metso Japan Co., Ltd.）で取り扱います。
Cited by 0
Pallav Sarma, Louis J. Durlofsky, Khalid Aziz · 2005
Efficient Closed-loop Production Optimization Under Uncertainty Pallav Sarma; Pallav Sarma Stanford University Search for other works by this author on: This Site Google Scholar Louis J. Durlofsky; Louis J. Durlofsky Stanford University Search for other works by this author on: This Site Google Scholar Khalid Aziz Khalid Aziz Stanford University Search for other works by this author on: This Site Google Scholar Paper presented at the SPE Europec/EAGE Annual Conference, Madrid, Spain, June 2005. Paper Number: SPE-94241-MS https://doi.org/10.2118/94241-MS Published: June 13 2005 Connected Content Related to: Efficient Closed-Loop Production Optimization Under Uncertainty Cite View This Citation Add to Citation Manager Share Icon Share Twitter LinkedIn Get Permissions Search Site Citation Sarma, Pallav, Durlofsky, Louis J., and Khalid Aziz. "Efficient Closed-loop Production Optimization Under Uncertainty." Paper presented at the SPE Europec/EAGE Annual Conference, Madrid, Spain, June 2005. doi: https://doi.org/10.2118/94241-MS Download citation file: Ris (Zotero) Reference Manager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex Search Dropdown Menu nav search search input Search input auto suggest search filter All ContentAll ProceedingsSociety of Petroleum Engineers (SPE)SPE Europec featured at EAGE Conference and Exhibition Search Advanced Search AbstractThis work discusses a closed-loop approach for efficient real-time production optimization that consists of three key elements - adjoint models for efficient parameter and control gradient calculation, polynomial chaos expansions for efficient uncertainty propagation, and Karhunen-Loeve (K-L) expansions and Bayesian inversion theory for efficient real-time model updating (history-matching). The control gradients provided by the adjoint solution are used by a gradient-based optimization algorithm to determine optimal control settings, while the parameter gradients are used for model updating. Polynomial chaos expansions provide optimal encapsulation of information contained in the input random fields and output random variables. This approach allows the forward model to be used as a black box but is much faster than standard Monte Carlo techniques. The K-L representation allows for the direct application of adjoint techniques for history matching while assuring that the two-point geostatistics of the reservoir description are maintained. The benefits and efficiency of the overall closed-loop approach are demonstrated through real-time optimizations of net present value (NPV) for synthetic reservoirs under waterflood subject to production constraints and uncertain reservoir description. The closed-loop procedure is shown to provide a substantial improvement in NPV over the base case, and the results are seen to be very close to those obtained when the reservoir description is known apriori.IntroductionOne of the primary goals of the reservoir modeling and management process is to enable decisions that maximize the production potential of the reservoir. Among the various existing approaches to accomplish this, real-time model-based reservoir management, also known as the "closed-loop" approach, has recently generated significant interest. This methodology entails model-based optimization of reservoir performance under geological uncertainty, while also incorporating dynamic information in real-time, which acts to reduce model uncertainty. For such schemes to be practically applicable, a number of algorithmic advances are required. In this paper, we describe and apply algorithms for efficient closed-loop reservoir management based on the application of optimal control theory, Bayesian inversion theory, Karhunen-Loeve expansions and polynomial chaos expansions.Real-time model-based reservoir management can be explained with reference to Fig. 1. In the figure, the "System" box represents the real system over which some cost function (e.g., net present value or cumulative oil produced), designated J(u), is to be optimized. The system consists of the reservoir, wells and surface facilities. Here is a set of controls (well rates, bottom hole pressures), which can be controlled in order to maximize or minimize J(u). The "Low-order model" box represents the approximate model of the system, which in our case is the simulation model of the reservoir and facilities. Since our knowledge of the reservoir is generally uncertain, the simulation model and its output are also uncertain. The basic idea behind the closed-loop approach is to perform an optimization step on the approximate model, apply the optimal controls thus obtained on the real reservoir, gather new data as it becomes available, and assimilate this data to reduce model uncertainty. This loop can thus be repeated over the life of the reservoir [1,2]. Keywords: algorithm, reservoir characterization, upstream oil & gas, permeability field, expansion, artificial intelligence, coefficient, constraint, random variable, control step Subjects: Reservoir Characterization, Improved and Enhanced Recovery, Reservoir Simulation, Geologic modeling, Waterflooding This content is only available via PDF. 2005. Society of Petroleum Engineers You can access this article if you purchase or spend a download.
Cited by 20
Kavya Govindarajan, Seep Goel, Praveen Jayachandran et al. · 2023
The rising popularity of 5G is expected to drive transformation in a large number of industries, thereby generating a diverse set of use cases with distinct and specific requirements on throughput, latency, availability etc. 5G Slicing, one of the key enablers of 5G, will allow different applications with custom QoS requirements to be supported on the same physical infrastructure. A slice, during its lifecycle from planning to decommissioning, requires a scalable Management and Orchestration (MANO) framework that automates the slice deployment and its management. The dynamic nature of slices and the strict QoS requirements of 5G applications, requires a more intelligent MANO framework with analytics capabilities for slice SLO assurance and optimization capabilities for resource efficiency. In this demo, we present an end-to-end 5G slicing lifecycle framework with additional capabilities of optimal cost-efficient placement of slices and 3GPP compliant Data and Analytics Function (DAF) module for intelligent monitoring for assurance.
Cited by 3
Kavya Govindarajan, Seep Goel, Praveen Jayachandran et al. · 2023
With 5G, the telecommunication industry is promising higher connectivity and tailor made quality of service. The underpinning technology behind these promises is Network Slicing. Slicing allows applications having widely disparate requirements such as low latency, high bandwidth, high reliability etc. to co-exist on the same network infrastructure. 5G Slicing is facilitated by a Management and Orchestration (MANO) framework that handles its lifecycle through Day 0 operations of planning, Day 1 operations of deployment and Day 2 operations of assurance. In our demo, we present an end-to-end 5G Slicing lifecycle MANO framework, which in addition to the typical MANO functionalities, performs optimal deployment of slices and intelligent slice monitoring and assurance. We evaluate our MANO framework on IBM Cloud Pak for Network Automation (CP4NA) for a closed-loop fault detection and remediation scenario and show mean time to detect of 14.7s and mean remediation time of 15s.
Cited by 0
Uday Bondhugula, Albert Cohen, J. Ramanujam et al. · 2011
High-level loop transformations are a key instrument in mapping computational kernels to effectively exploit resources in modern processor architectures. However, determining appropriate compo-sitions of loop transformations to achieve this remains a signif-icantly challenging task; current compilers may achieve signifi-cantly lower performance than hand-optimized programs. To ad-dress this fundamental challenge, we first present a convex char-acterization of all distinct, semantics-preserving, multidimensional affine transformations. We then bring together algebraic, algorith-mic, and performance analysis results to design a tractable opti-mization algorithm over this highly expressive space. The frame-work has been implemented and validated experimentally on a representative set of benchmarks run on state-of-the-art multi-core platforms. Categories and Subject Descriptors D 3.4 [Programming lan-
Cited by 0