Programming Overhead Research Articles

Toward a Dynamic Allocation Strategy for Deadline‐Oriented Resource and Job Management in HPC Systems

ABSTRACTAs high‐performance computing (HPC) becomes a tool used in many different workflows, quality of service (QoS) becomes increasingly important. In many cases, this includes the reliable execution of an HPC job and the generation of the results by a certain deadline. The resource and job management system (RJMS) or simply RMS is responsible for receiving the job requests and executing the jobs with a deadline‐oriented policy to support the workflows. In this article, we evaluate how well static resource management policies cope with deadline‐constrained HPC jobs and explore two variations of a dynamic policy in this context. As the Hilbert curve‐based approach used by the SLURM workload manager represents the state‐of‐the‐art in production environments, it was selected as one of the static allocation strategies. The Manhattan median approach as a second allocation strategy was introduced as a research work that aims to minimize the communication overhead of the parallel programs by providing compact partitions more than the Hilbert curve approach. In contrast to the static partitions provided by the Hilbert curve approach and the Manhattan median approach, the leak approach focuses on supporting dynamic runtime behavior of the jobs and assigning nodes of the HPC system on demand at runtime. Since the contiguous leak version also relies on a compact set of nodes, the noncontiguous leak can provide additional nodes at a greater distance from the nodes already used by the job. Our preliminary results clearly show that a dynamic policy is needed to meet the requirements of a modern deadline‐oriented RMS scenario.

Static Scheduling of Weight Programming for DNN Acceleration with Resource Constrained PIM

Most existing architectural studies on ReRAM-based processing-in-memory (PIM) DNN accelerators assume that all weights of the DNN can be mapped to the crossbar at once. However, these studies are over-idealized. ReRAM crossbar resources for calculation are limited because of technological limitations, so multiple weight mapping procedures are required during the inference process. In this article, we propose a static scheduling framework which generates the mapping between DNN weights and ReRAM cells with minimum runtime weight programming cost. We first build a ReRAM crossbar programming latency model by simultaneously considering the DNN weight patterns, ReRAM programming operations, and PIM architecture characteristics. Then, the model is used in the searching process to obtain an optimized weight-to-OU mapping table with minimum online programming latency. Finally, an OU scheduler is used to coordinate the activation sequences of OUs in the crossbars to perform the inference computation correctly. Evaluation results show the proposed framework significantly reduces the weight programming overhead and the overall inference latency for various DNN models with different input datasets.

Compensation architecture design utilizing residual resource to mitigate impacts of nonidealities in RRAM-based computing-in-memory chips

Resistive random access memory (RRAM) is a promising technology for energy-efficient in-memory computing. However, due to technology limits, RRAM device faces a series of reliability issues. Deep neural network (DNN) computing based on RRAM suffers from accuracy degradation. On the one hand, offline DNN training solutions are difficult to fully consider and simulate all nonidealities. Worse still, new error or nonideality may come up with the usage of RRAM, which further deteriorates the effectiveness of offline training. On the other hand, online training poses great challenges on programming overhead and device lifetime. The iterative write-verify technique to program multi-bit RRAM cells prolongs write latency more than 10× longer than read latency. To overcome these issues, we propose a compensation architecture and training designs to mitigate the realistic accuracy loss in RRAM chips with negligible hardware resource for RRAM-based DNN computing. Firstly, we add trainable compensation channels in crossbars utilizing the residual resource after original weight mapping. Secondly, an offline training procedure with computing output from hardware is triggered to settle down appropriate weight value in compensation channels. Experimental results demonstrate that the proposed design can guarantee ≤ 0.8% loss of accuracy in DNN even when nonidealities reduce the original accuracy down to ≤73%.

Producing Fast and Convenient Machine Learning Benchmarks in R with the stressor Package

The programming overhead required to implement machine learning workflows creates a barrier for many discipline-specific researchers with limited programming experience. The stressor package provides an R interface to Python’s PyCaret package, which automatically tunes and trains 14-18 machine learning (ML) models for use in accuracy comparisons. In addition to providing an R interface to PyCaret, stressor also contains functions that facilitate synthetic data generation and variants of cross-validation that allow for easy benchmarking of the ability of machine-learning models to extrapolate or compete with simpler models on simpler data forms. We show the utility of stressor on two agricultural datasets, one using classification models to predict crop suitability and another using regression models to predict crop yields. Full ML benchmarking workflows can be completed in only a few lines of code with relatively small computational cost. The results, and more importantly the workflow, provide a template for how applied researchers can quickly generate accuracy comparisons of many machine learning models with very little programming.

Optimizing Inter-Core Communications Under the LET Paradigm using DMA Engines

Modern automotive applications are increasingly characterized by the need to transfer massive amounts of data in a predictable and deterministic way, possibly leveraging the Logical Execution Time (LET) paradigm. However, current proposals for LET communications are limited to core-commanded data transfers, which may result in large delays for data-intensive systems. To address this issue, we explore the use of Direct Memory Access (DMA) to handle LET communication with improved parallelism. Each DMA transfer operates on a contiguous memory area, thus calling for an optimized memory mapping to maximize performance. Modern DMA engines offer also advanced configurations, such as linked-lists of data transfers, which may provide more flexibility at the expenses of an increased (initial) programming overhead. Leveraging all such features of DMA engines, we propose a set of designs and protocols for LET communications with trade-offs between latency and space requirements. For each option we present the formulation to compute the optimal scheduling and memory allocation solution as a mixed-integer linear programming problem. Experimental results show the feasibility of the approach and a comparison of the solutions obtained using the proposed methods, showing a considerable improvement in terms of data acquisition latency when compared to LET communication without DMA.

Retracted: Lifetime maximization energy‐aware routing protocol for route optimization to improve quality of service in wireless sensor networks

AbstractWireless sensor networks (WSNs) are part of the short‐term networks, including sensitivity, computation, and Wi‐Fi connectivity capability. Many routing, range management, and log transfer protocols are specifically designed for WSN. The previous method showed less efficiency in routing management. The proposed lifetime maximization energy‐aware routing protocol (LTMEARP) protocol is known for its high routing efficiency, giving a higher lifetime and throughput performance. The requirement for a routing system approach is advanced with Internet service provider's protocols by recalculating the routing table after the link stage's substitution worldwide, leading to responses and connection failures by sharing important data after traffic. LTMEARP routing protocol has assured high availability routing performances during traffic conditions. LTMEARP support sends homogeneous and expanded nodes. Analyze a new approach to routing‐based selection algorithms for homogeneous node WSNs. The number of connections is limited and must be adjusted using separate paths and header packets to meet the user's network access location. The results show that the LTMEARP has achieved quality without introducing excessive network access program overhead, which deals with the study of communication and the benefits and problems with the performance of each routing technology.

Implementation of Dual Number Automatic Differentiation with John Newman’s BAND Algorithm

This paper asserts that the development of continuum-scale mathematical models utilizing John Newman’s BAND subroutine can be simplified through the use of dual number automatic differentiation. This paper covers the salient features of the BAND algorithm as well as dual numbers and how they can be leveraged to algorithmically linearize systems of partial differential equations; these two concepts can be combined to produce accurate and computationally efficient models while significantly reducing the amount of personnel time necessary by eliminating the time-consuming process of equation linearization. As a result, this methodology facilitates more rapid model prototyping and establishes a more intuitive relationship between the numerical model and the differential equations. By utilizing an existing and validated programming module, dnadmod, these advantages are achieved without burdening the general user with significant additional programming overhead.

An Extensible Compiler for Implementing Software Design Patterns as Concise Language Constructs

Design patterns are generic solutions to common programming problems. Design patterns represent a typical example of design reuse. However, implementing design patterns can lead to several problems, such as programming overhead and traceability. Existing research introduced several approaches to alleviate the implementation issues of design patterns. Nevertheless, existing approaches pose different implementation restrictions and require programmers to be aware of how design patterns should be implemented. Such approaches make the source code more prone to faults and defects. In addition, existing design pattern implementation approaches limit programmers to apply specific scenarios of design patterns (e.g. class-level), while other approaches require scattering implementation code snippets throughout the program. Such restrictions negatively impact understanding, tracing, or reusing design patterns. In this paper, we propose a novel approach to support the implementation of software design patterns as an extensible Java compiler. Our approach allows developers to use concise, easy-to-use language constructs to apply design patterns in their code. In addition, our approach allows the application of design patterns in different scenarios. We illustrate our approach using three commonly used design patterns, namely Singleton, Observer and Decorator. We show, through illustrative examples, how our design pattern constructs can significantly simplify implementing design patterns in a flexible, reusable and traceable manner. Moreover, our design pattern constructs allow class-level and instance-level implementations of design patterns.

BonZeb: open-source, modular software tools for high-resolution zebrafish tracking and analysis

We present BonZeb—a suite of modular Bonsai packages which allow high-resolution zebrafish tracking with dynamic visual feedback. Bonsai is an increasingly popular software platform that is accelerating the standardization of experimental protocols within the neurosciences due to its speed, flexibility, and minimal programming overhead. BonZeb can be implemented into novel and existing Bonsai workflows for online behavioral tracking and offline tracking with batch processing. We demonstrate that BonZeb can run a variety of experimental configurations used for gaining insights into the neural mechanisms of zebrafish behavior. BonZeb supports head-fixed closed-loop and free-swimming virtual open-loop assays as well as multi-animal tracking, optogenetic stimulation, and calcium imaging during behavior. The combined performance, ease of use and versatility of BonZeb opens new experimental avenues for researchers seeking high-resolution behavioral tracking of larval zebrafish.

Compiler-Based Efficient CNN Model Construction for 5G Edge Devices

With the increasing demand to deploy convolutional neural networks (CNNs) on 5G mobile platforms, architecture designs with efficient sparse kernels (SKs) were proposed, which can save more parameters than the standard convolution while maintaining the high accuracy. Despite the great potential, neural network designs with SKs still require a lot of expert knowledge and take ample time. In this paper, we first propose a search scheme that effectively reduces the SK design space based on three aspects: composition, performance, and efficiency. Meanwhile, we completely eliminate the model training from our search scheme. Instead, an easily measurable quantity, the information field, is identified and used to predict the model accuracy in the searching process. Additionally, we provide a detailed efficiency analysis on the final designs found by our scheme. Second, based on the analysis we propose a model transformation scheme to better utilize the SK designs on existing models to either reduce the number of parameters or increase the accuracy. Last, considering the extra programming overhead and the expert knowledge required by the model transformation scheme, we develop a compiler prototype to automate the entire process, given the source code of an existing model. Experimental results show that models composed of the sparse kernel designs searched by our search scheme can beat state-of-the-art networks such as ResNets in terms of the accuracy and the efficiency. Also by using our model transformation scheme we can easily improve the accuracy (the same number of parameters) or the efficiency (the same accuracy) upon existing state-of-the-art models.

Generalised cost-effectiveness analysis of 159 health interventions for the Ethiopian essential health service package

BackgroundCost effectiveness was a criterion used to revise Ethiopia’s essential health service package (EHSP) in 2019. However, there are few cost-effectiveness studies from Ethiopia or directly transferable evidence from other low-income countries to inform a comprehensive revision of the Ethiopian EHSP. Therefore, this paper reports average cost-effectiveness ratios (ACERs) of 159 health interventions used in the revision of Ethiopia’s EHSP.MethodsIn this study, we estimate ACERs for 77 interventions on reproductive maternal neonatal and child health (RMNCH), infectious diseases and water sanitation and hygiene as well as for 82 interventions on non-communicable diseases. We used the standardised World Health Organization (WHO) CHOosing Interventions that are cost effective methodology (CHOICE) for generalised cost-effectiveness analysis. The health benefits of interventions were determined using a population state-transition model, which simulates the Ethiopian population, accounting for births, deaths and disease epidemiology. Healthy life years (HLYs) gained was employed as a measure of health benefits. We estimated the economic costs of interventions from the health system perspective, including programme overhead and training costs. We used the Spectrum generalised cost-effectiveness analysis tool for data analysis. We did not explicitly apply cost-effectiveness thresholds, but we used US$100 and $1000 as references to summarise and present the ACER results.ResultsWe found ACERs ranging from less than US$1 per HLY gained (for family planning) to about US$48,000 per HLY gained (for treatment of stage 4 colorectal cancer). In general, 75% of the interventions evaluated had ACERs of less than US$1000 per HLY gained. The vast majority (95%) of RMNCH and infectious disease interventions had an ACER of less than US$1000 per HLY while almost half (44%) of non-communicable disease interventions had an ACER greater than US$1000 per HLY.ConclusionThe present study shows that several potential cost-effective interventions are available that could substantially reduce Ethiopia’s disease burden if scaled up. The use of the World Health Organization’s generalised cost-effectiveness analysis tool allowed us to rapidly calculate country-specific cost-effectiveness analysis values for 159 health interventions under consideration for Ethiopia’s EHSP.

Deep learning for genomics using Janggu

In recent years, numerous applications have demonstrated the potential of deep learning for an improved understanding of biological processes. However, most deep learning tools developed so far are designed to address a specific question on a fixed dataset and/or by a fixed model architecture. Here we present Janggu, a python library facilitates deep learning for genomics applications, aiming to ease data acquisition and model evaluation. Among its key features are special dataset objects, which form a unified and flexible data acquisition and pre-processing framework for genomics data that enables streamlining of future research applications through reusable components. Through a numpy-like interface, these dataset objects are directly compatible with popular deep learning libraries, including keras or pytorch. Janggu offers the possibility to visualize predictions as genomic tracks or by exporting them to the bigWig format as well as utilities for keras-based models. We illustrate the functionality of Janggu on several deep learning genomics applications. First, we evaluate different model topologies for the task of predicting binding sites for the transcription factor JunD. Second, we demonstrate the framework on published models for predicting chromatin effects. Third, we show that promoter usage measured by CAGE can be predicted using DNase hypersensitivity, histone modifications and DNA sequence features. We improve the performance of these models due to a novel feature in Janggu that allows us to include high-order sequence features. We believe that Janggu will help to significantly reduce repetitive programming overhead for deep learning applications in genomics, and will enable computational biologists to rapidly assess biological hypotheses.

Does transparency come at the cost of charitable services? Evidence from investigating British charities

Recent high-profile scandals related to misuse of funding and donations have raised the demand for scrutiny over financial transparency and operational activities of non-profit organizations in developed countries. Our analysis challenges the common practice in the sector of using programme ratios and overhead costs as indicators for non-profit accountability. Using Benford's Law to measure irregularities in financial data for a large sample of public charities we estimate that 25% of the sample potentially misreport their financial information. We show theoretically and empirically that charities with a higher programme ratio (their level of spending on charitable activities), will be less likely to misreport their financial information only when their overhead costs (spending on governing activities) are also sufficiently high. Tighter monitoring becomes ineffective in increasing the sectoral transparency and accountability unless accompanied by a sufficiently high level of charitable spending.

CF-PROV: A Content-Rich and Fine-Grained Scientific Workflow Provenance Model

Due to the complexity of scientific experiments, existing methods for capturing the provenance information from the scientific workflows (SWFs) face high programming overhead and code errors. In addition, the coarse-grained nature of the SWF provenance implies the loss of internal details of the workflow processes, which can lead to incomplete or inaccurate data dependencies and dependency differentiation problems. The diversity in scientific fields also reveals the limited versatility of the SWF provenance models. In this context, we propose a content-rich and fine-grained SWF provenance model (CF-PROV). This model provides normative transformations and documentation declarations for the multi-field SWFs, reducing the programming overhead and increasing the versatility. Our method of dividing the workflow provenance into data provenance and process provenance and our formal description of data deduction at the field level enrich the provenance information and transform the coarse-grained workflow provenance into a fine-grained provenance. Finally, the experiments on the model compression ratio and model generation time in multiple scientific fields demonstrate the versatility and rationality of the CF-PROV.

基于DPDK的用户态IPSec网关的研究和实现

本论文针对在使用IPSec网关时由于用户量不断增加、网络流量需求大的情况下所造成的时延、阻塞、吞吐量小等问题，分析了内核态IPSec解决方案处理流程中存在的不足，提出了基于DPDK在用户态实现IPSec网关的方法。针对传统方案处理开销大的问题提出利用思科的VPP作框架，DPDK作收包工具搭建平台实现数据包在用户态的处理。测试结果表明，基于DPDK的用户态IPSec解决方案与传统方案相比，性能得到了有效的提升。 This paper aims at the problems of delay, congestion, small throughput caused due to increasing users and more network traffic demand by using IPSec gateway, analyzes the shortcomings during the IPSec solution process, and puts forward the method of implementation of DPDK gateway based on IPSec in user space. Aiming at the problem of large overhead of traditional programs, we propose to build Cisco VPP as a platform and DPDK as a receiving packets tool to build data processing platform. The test results show that the performance of the IPSec solution of user space based on DPDK is improved effectively compared with the traditional solution.

Al3c: high-performance software for parameter inference using Approximate Bayesian Computation.

The development of Approximate Bayesian Computation (ABC) algorithms for parameter inference which are both computationally efficient and scalable in parallel computing environments is an important area of research. Monte Carlo rejection sampling, a fundamental component of ABC algorithms, is trivial to distribute over multiple processors but is inherently inefficient. While development of algorithms such as ABC Sequential Monte Carlo (ABC-SMC) help address the inherent inefficiencies of rejection sampling, such approaches are not as easily scaled on multiple processors. As a result, current Bayesian inference software offerings that use ABC-SMC lack the ability to scale in parallel computing environments. We present al3c, a C++ framework for implementing ABC-SMC in parallel. By requiring only that users define essential functions such as the simulation model and prior distribution function, al3c abstracts the user from both the complexities of parallel programming and the details of the ABC-SMC algorithm. By using the al3c framework, the user is able to scale the ABC-SMC algorithm in parallel computing environments for his or her specific application, with minimal programming overhead. al3c is offered as a static binary for Linux and OS-X computing environments. The user completes an XML configuration file and C++ plug-in template for the specific application, which are used by al3c to obtain the desired results. Users can download the static binaries, source code, reference documentation and examples (including those in this article) by visiting https://github.com/ahstram/al3c. astram@usc.edu Supplementary data are available at Bioinformatics online.

DELAY AND POWER REDUCTION IN NEW ROUTING FABRICS

In this study we created a new routing fabric for r educing power and delay. The power consumed in a FPGA core consists of both static and dynamic components. Static power contributes only 10% of the total power consumed in a FPGA. On the other hand, dynamic power contributes over 90% of the total power consumed and it is the main source for their power inefficiency. By reducing net length and/or programming overhead the power consumption reduced. Routed net length reduced by using short intersects segments in the routing channels. By dec reasing the switch box and/or connection box flexibilities programming overhead reduced. In this study ,we concentrated on achieving 1.80 times lower consumption of dynamic power and 1.50 times less significant average net delays by rearchitecting the programmable routing fabrics such that both routed net lengths and programming overhead reduced without adversely affecting delay.

IMPLEMENTING LOW-COST FAULT TOLERANCE VIA HYBRID SYNCHRONOUS/ASYNCHRONOUS CHECKS

As semiconductor technologies scale down to deep sub-micron dimensions, transient faults will soon become a critical reliability concern. Due to their prohibitive costs, traditional high-end solutions are unacceptable for the mainstream commodity market. This paper presents FTPIPE, a hybrid software/hardware solution, which provides sufficient fault coverage with affordable overhead for single-threaded programs running on commodity systems. Leveraging existing exception mechanisms with minor modifications to handle exception-causing faults, FTPIPE focuses on tolerating silent data corruptions by using compile-time analysis and performing selective instruction replication in a modern superscalar pipeline extended with minimal hardware overhead. Unlike existing instruction replication-based solutions, which detect faults by synchronous checks, the FTPIPE platform has exploited a novel hybrid synchronous/asynchronous check method for the replicated instructions. In this manner, better performance can be obtained without degradation of fault coverage. By synchronous checks, the validation of the result of a replicated instruction must be finished before it is committed, whereas such a guarantee is not required by an asynchronous check. Evaluation using a set of nine programs from the Mibench benchmark suite demonstrates that FTPIPE can tolerate 89.8% of transient faults under a modest performance overhead of 20.1%.

SIMPSON: A general simulation program for solid-state NMR spectroscopy

A computer program for fast and accurate numerical simulation of solid-state NMR experiments is described. The program is designed to emulate a NMR spectrometer by letting the user specify high-level NMR concepts such as spin systems, nuclear spin interactions, RF irradiation, free precession, phase cycling, coherence-order filtering, and implicit/explicit acquisition. These elements are implemented using the Tel scripting language to ensure a minimum of programming overhead and direct interpretation without the need for compilation, while maintaining the flexibility of a full-featured programming language. Basicly, there are no intrinsic limitations to the number of spins, types of interactions, sample conditions (static or spinning, powders, uniaxially oriented molecules, single crystals, or solutions), and the complexity or number of spectral dimensions for the pulse sequence. The applicability ranges from simple ID experiments to advanced multiple-pulse and multiple-dimensional experiments, series of simulations, parameter scans, complex data manipulation/visualization, and iterative fitting of simulated to experimental spectra. A major effort has been devoted to optimizing the computation speed using state-of-the-art algorithms for the time-consuming parts of the calculations implemented in the core of the program using the C programming language. Modification and maintenance of the program are facilitated by releasing the program as open source software (General Public License) currently at http://nmr.imsb.au.dk. The general features of the program are demonstrated by numerical simulations of various aspects for REDOR, rotational resonance, DRAMA, DRAWS, HORROR, C7, TEDOR, POST-C7, CW decoupling, TPPM, F-SLG, SLF, SEMA-CP, PISEMA, RFDR, QCPMG-MAS, and MQ-MAS experiments.

A Tensor Approach to the Mesh Resistance Matrix

This paper presents a tensor approach to obtain the mesh resistance matrix of a power system. The traditional approach to the mesh matrices naturally relates to graph theory, where the fundamental loops of a representative graph are found from a spanning tree. While valid, mesh identification is time consuming, involves unnecessary programming overhead, yields dense mesh matrices, and requires developing good search heuristics. The proposed approach uses a connection tensor to form a sparse mesh resistance matrix without resorting to graph theory. It allows for the interconnection of only those meshes circulating around the terminals of each power apparatus, which is more effective than searching for the meshes of an entire power system. Detailed steps to form the tensor and supporting examples illustrate the procedure presented in several scenarios. It is also shown that the mesh resistance matrix is sparse and that the computational effort to form the tensor is negligible.