Parallel Jobs Research Articles

OCTOPUS: operation control system for task optimization and job parallelization via a user-optimal scheduler

The material acceleration platform, empowered by robotics and artificial intelligence, is a transformative approach for expediting material discovery processes across diverse domains. However, the development of an operating system for material acceleration platform faces challenges in simultaneously managing diverse experiments from multiple users. Specifically, when it is utilized by multiple users, the overlapping challenges of experimental modules or devices can lead to inefficiencies in both resource utilization and safety hazards. To overcome these challenges, we present an operation control system for material acceleration platform, namely, OCTOPUS, which is an acronym for operation control system for task optimization and job parallelization via a user-optimal scheduler. OCTOPUS streamlines experiment scheduling and optimizes resource utilization through integrating its interface node, master node and module nodes. Leveraging process modularization and a network protocol, OCTOPUS ensures the homogeneity, scalability, safety and versatility of the platform. In addition, OCTOPUS embodies a user-optimal scheduler. Job parallelization and task optimization techniques mitigate delays and safety hazards within realistic operational environments, while the closed-packing schedule algorithm efficiently executes multiple jobs with minimal resource waste. Copilot of OCTOPUS is developed to promote the reusability of OCTOPUS for potential users with their own sets of lab resources, which substantially simplifies the process of code generation and customization through GPT recommendations and client feedback. This work offers a solution to the challenges encountered within the platform accessed by multiple users, and thereby will facilitate its widespread adoption in material development processes.

MallobSat: Scalable SAT Solving by Clause Sharing

SAT solving in large distributed environments has previously led to some famous results and to impressive speedups for selected inputs. However, in terms of general-purpose SAT solving, prior approaches still cannot make efficient use of a large number of processors. We aim to address this issue with a complete and systematic overhaul of the distributed solver HordeSat with a focus on its algorithmic building blocks. In particular, we present a communication-efficient approach to clause sharing, careful buffering and filtering of produced clauses, and effective orchestration of state-of-the-art solver backends. In extensive evaluations, our approach named MallobSat significantly outperforms an updated HordeSat, doubling its mean speedup. Our clause sharing results in effective parallelization even if all threads execute identical solver programs that only differ based on which clauses they import at which times. We thus argue that MallobSat is not a portfolio solver with the added bonus of clause sharing but rather a clause-sharing solver where adding some explicit diversification is useful but not essential. We also discuss the last four iterations of the International SAT Competition (2020–2023), where our system ranked very favorably, and identify several previously unsolved competition problems that MallobSat solved successfully. Last but not least, our approach is malleable, i.e., supports running on a fluctuating set of resources, which allows us to combine parallel job processing and parallel SAT solving in a flexible manner for best resource efficiency

Opposition‐based improved memetic algorithm for placement of concurrentInternet of Thingsapplications in fog computing

AbstractThe transpiration of a new computing standard, Fog computing exploits the computing resources immediacy to the Internet of Things (IoT) devices and hence together with the cloud servers bestow the required resources promptly. Even though the fog server provides timely services due to the fewer resources it cannot tackle the resource‐consuming applications developed by more IoT devices. Hence the placement of applications in the fog computing paradigm with more cloud and fog servers is a major challenging problem. To overcome this problem and to minimize the execution time and energy consumption of the IoT applications in a computing platform consisting of numerous IoT devices, and several fog and cloud servers, a weighted cost model is proposed. The number of parallel jobs for concurrent execution is increased due to the heterogeneity of IoT applications, and hence a pre‐scheduling technique is presented to accomplish this. In addition to that, a novel application placement method using the Opposition‐based Improved Memetic Algorithm (OBIMA) is adapted to formulate the placement of parallel IoT operations. The experimental results show that the proposed model outperforms well compared to other state‐of‐the‐art methods.

Logic-based Benders decomposition for bi-objective parallel machine selection and job scheduling with release dates and resource consumption

This work addresses a new bi-objective parallel machine selection and job scheduling problem with release dates and resource consumption. It consists in optimally selecting subcontractors (machines) from a set of geographically dispersed locations and scheduling the orders (jobs) to the selected subcontractors for processing while meeting the order release dates and resource consumption restrictions, so as to simultaneously minimize the maximum completion time, i.e., the makespan, and the total cost including machine usage cost and resource consumption cost. The problem is first formulated into a bi-objective mixed-integer linear program based on linear ordering (LO-MILP), and then valid inequalities are explored based on property analysis. To solve it, an ɛ-constraint method based on LO-MILP (ɛ-LO-MILP) is first proposed. To more efficiently solve it, we also develop a tailored logic-based Benders decomposition combined with ɛ-constraint method (ɛ-LBBD) where a novel method to obtain a tight lower bound of the identical parallel machine with machine-dependent release dates and some problem-specific cuts are proposed. Numerical experiments on an illustrative example are conducted to show the applicability of the model and algorithm and intuitively reveal the trade-off between production efficiency and cost. Experimental results on 200 instances with up to 100 orders demonstrate that ɛ-LBBD can reduce the computation time by about 28.92% compared to ɛ-LO-MILP and yields better Pareto solutions than ɛ-LO-MILP and the well-known non-dominated sorted genetic algorithm II do.

Организация мелкосерийного производства деталей на принципах модульной технологии

It is emphasized that the traditional job shop production is characterized by low efficiency due to the use of superuniversal machining facilities for their underemployment of performance capabilities at work and also low-productivity of process flowsheet when machining part surfaces as a sequential surfaced job, the organizational form of production in the form of sections by type of machine tools, i.e. a section of lathes, a section of milling machines, etc. It results in large travel length of workpieces in the machining process at work. The desire to increase the efficiency of production by organizing launch of batch workpieces that are formed in a constructive similarity, such as structures, parts of rotation body types, etc., does not have a significant effect, because the task is just minimizing the run time spent on changing fixtures and reinstalling workpieces, but not taking into account an optimal sequence of parts in the batch. For increasing the efficiency of small-scale production, the module technology is proposed. The main idea of it is to represent the part not as a network of separate surfaces, but as surface modules, where the surface module is understood as a combination of surfaces and the part performs its proper function with the help of this module. This enables a good specialization of workplaces that are organized not for processing methods, but for manufacturing the appropriate groups of surface modules. The latter make it possible to take advantage of a more productive process flowsheet for machining part surfaces. It is sequential parallel surface job that is used. When organizing production, it becomes possible to use the layout of workplaces in line according to the in-line method, which will significantly reduce the part travel length at workplaces.

Multi-resource scheduling of moldable workflows

Resource scheduling plays a vital role in High-Performance Computing (HPC) systems. Most scheduling research in HPC has focused on only a single type of resource (e.g., computing cores or I/O resource). With the advancement in hardware architectures and the increase in data-intensive HPC applications, there is a need to simultaneously consider a diverse set of resources (e.g., computing cores, cache, memory, I/O, and network resources) in the design of runtime schedulers for improving the overall application performance. In this paper, we study multi-resource scheduling to minimize the makespan of computational workflows comprised of moldable parallel jobs. Moldable jobs allow the scheduler to flexibly select a variable set of resources before execution, thus can adapt to the available system resources (as compared to rigid jobs) while staying easy to design and implement (as compared to malleable jobs). We propose a Multi-Resource Scheduling Algorithm (MRSA), which combines a novel resource allocation strategy and an extended list scheduling scheme to schedule the jobs. We prove that, on a system with d types of schedulable resources, MRSA achieves an approximation ratio of 1.619d+2.545d+1 for any d≥1, and a ratio of d+3d23+O(d3) when d is large (i.e., d≥22). We also present improved approximation results for workflows comprised of jobs with special precedence constraints (e.g., series-parallel graphs, trees, and independent jobs). Further, we prove a lower bound of d on the approximation ratio of any list-based scheduling algorithm with local priority considerations. Finally, we conduct a comprehensive set of simulations to evaluate the performance of the algorithm using synthetic workflows of different structures and moldable jobs following different speedup models. The results show that MRSA fares better than the theoretical bound predicts, and that it consistently outperforms two baseline heuristics under a variety of parameter settings, illustrating its robust practical performance.

Performance models of data parallel DAG workflows for large scale data analytics

Directed Acyclic Graph (DAG) workflows are widely used for large-scale data analytics in cluster-based distributed computing systems. The performance model for a DAG on data-parallel frameworks (e.g., MapReduce) is a research challenge because the allocation of preemptable system resources among parallel jobs may dynamically vary during execution. This resource allocation variation during execution makes it difficult to accurately estimate the execution time. In this paper, we tackle this challenge by proposing a new cost model, called Bottleneck Oriented Estimation (BOE), to estimate the allocation of preemptable resources by identifying the bottleneck to accurately predict task execution time. For a DAG workflow, we propose a state-based approach to iteratively use the resource allocation property among stages to estimate the overall execution plan. Furthermore, to handle the skewness of various jobs, we refine the model with the order statistics theory to improve estimation accuracy. Extensive experiments were performed to validate these cost models with HiBench and TPC-H workloads. The BOE model outperforms the state-of-the-art models by a factor of five for task execution time estimation. For the refined skew-aware model, the average prediction error is under 3% when estimating the execution time of 51 hybrid analytics (HiBench) and query (TPC-H) DAG workflows.

Preemptive scheduling of parallel jobs of two sizes with controllable processing times

AbstractIn parallel machine scheduling, a size of a job is defined as the number of machines that are simultaneously required for its processing. This paper considers a scheduling problem in which the set of jobs consists of jobs of two sizes: the conventional jobs (each job requires a single machine for its processing) and parallel jobs (each job to be simultaneously processed on more than one machine). The processing times are controllable, and they have to be chosen from given intervals in order to guarantee the existence of a preemptive schedule in which all jobs are completed by a common deadline. The objective is to minimize the total compression cost which reflects possible reductions in processing times. Unlike problems of classical scheduling with conventional jobs, the model under consideration cannot be directly handled by submodular optimization methods. We reduce the problem to maximizing the total weighted work of all jobs parametrized with respect to total work of the parallel jobs. The solution is delivered by separately finding the breakpoints for the maximum total weighted work of the parallel jobs and for the maximum total weighted work of the conventional jobs. This results in a polynomial-time algorithm that is no slower than needed to perform sorting of jobs’ parameters.

Integrative analysis of individual-level data and high-dimensional summary statistics.

Researchers usually conduct statistical analyses based on models built on raw data collected from individual participants (individual-level data). There is a growing interest in enhancing inference efficiency by incorporating aggregated summary information from other sources, such as summary statistics on genetic markers' marginal associations with a given trait generated from genome-wide association studies. However, combining high-dimensional summary data with individual-level data using existing integrative procedures can be challenging due to various numeric issues in optimizing an objective function over a large number of unknown parameters. We develop a procedure to improve the fitting of a targeted statistical model by leveraging external summary data for more efficient statistical inference (both effect estimation and hypothesis testing). To make this procedure scalable to high-dimensional summary data, we propose a divide-and-conquer strategy by breaking the task into easier parallel jobs, each fitting the targeted model by integrating the individual-level data with a small proportion of summary data. We obtain the final estimates of model parameters by pooling results from multiple fitted models through the minimum distance estimation procedure. We improve the procedure for a general class of additive models commonly encountered in genetic studies. We further expand these two approaches to integrate individual-level and high-dimensional summary data from different study populations. We demonstrate the advantage of the proposed methods through simulations and an application to the study of the effect on pancreatic cancer risk by the polygenic risk score defined by BMI-associated genetic markers. R package is available at https://github.com/fushengstat/MetaGIM.

Multi-Job Intelligent Scheduling With Cross-Device Federated Learning

Recent years have witnessed a large amount of decentralized data in various (edge) devices of end-users, while the decentralized data aggregation remains complicated for machine learning jobs because of regulations and laws. As a practical approach to handling decentralized data, Federated Learning (FL) enables collaborative global machine learning model training without sharing sensitive raw data. The servers schedule devices to jobs within the training process of FL. In contrast, device scheduling with multiple jobs in FL remains a critical and open problem. In this article, we propose a novel multi-job FL framework, which enables the training process of multiple jobs in parallel. The multi-job FL framework is composed of a system model and a scheduling method. The system model enables a parallel training process of multiple jobs, with a cost model based on the data fairness and the training time of diverse devices during the parallel training process. We propose a novel intelligent scheduling approach based on multiple scheduling methods, including an original reinforcement learning-based scheduling method and an original Bayesian optimization-based scheduling method, which corresponds to a small cost while scheduling devices to multiple jobs. We conduct extensive experimentation with diverse jobs and datasets. The experimental results reveal that our proposed approaches significantly outperform baseline approaches in terms of training time (up to 12.73 times faster) and accuracy (up to 46.4% higher).

A Quantum Approach to the Problem of Charging Electric Cars on a Motorway

In this paper, the problem of charging electric motor vehicles on a motorway is considered. Charging points are located alongside the motorway. It is assumed that there are a number of vehicles on a given section of a motorway. In the motorway, there are several nodes, and for each vehicle, the entering and the leaving nodes are known, as well as the time of entrance. For each vehicle, we know the total capacity of its battery, and the current amount of energy in the battery when entering the motorway. It is also assumed that for each vehicle, there is a finite set of speeds it can use when traveling the motorway. The speed is chosen when entering the motorway, and cannot be changed before reaching the charging station. For each speed, there is given a corresponding power usage; the higher the speed, the larger the power usage. Each vehicle can only use one charger, and when its battery is full, the amount of energy is sufficient for reaching the outgoing node. We look for a feasible solution to the problem, i.e., a solution in which no vehicle has to wait for a charger. The problem is formulated as a problem of scheduling independent, nonpreemptable jobs in parallel, unrelated machines under an additional doubly constrained resource, which is power. Quantum approaches to solve the defined problem are proposed. They use the quantum approximate optimization algorithm and the quantum annealing technique. A computational experiment is presented and discussed. Some conclusions and directions for future research are given.

Optimizing the Expected Maximum of Two Linear Functions Defined on a Multivariate Gaussian Distribution

We study stochastic optimization problems with objective function given by the expectation of the maximum of two linear functions defined on the component random variables of a multivariate Gaussian distribution. We consider random variables that are arbitrarily correlated, and we show that the problem is NP-hard even if the space of feasible solutions is unconstrained. We exploit a closed-form expression for the objective function from the literature to construct a cutting-plane algorithm for a highly nonlinear function, which includes the evaluation of the cumulative distribution function and probability density function of a standard normal random variable with decision variables as part of the arguments. To exhibit the model’s applicability, we consider two featured applications. The first is daily fantasy sports, where the algorithm identifies entries with positive returns during the 2018–2019 National Football League season. The second is a special case of makespan minimization for two parallel machines and jobs with uncertain processing times; for the special case where the jobs are uncorrelated, we prove the equivalence between its deterministic and stochastic versions and show that our algorithm can deliver a constant-factor approximation guarantee for the problem. The results of our computational evaluation involving synthetic and real-world data suggest that our discretization and upper bounding techniques lead to significant computational improvements and that the proposed algorithm outperforms suboptimal solutions approaches. History: Andrea Lodi, Area Editor for Design & Analysis of Algorithms–Discrete. Supplemental Material: The online appendix is available at https://doi.org/10.1287/ijoc.2022.1259 .

Online scheduling of parallelizable jobs in the directed acyclic graphs and speed-up curves models

This paper considers scheduling jobs online on m identical machines such that the jobs can be parallelized across the machines. Two models of parallelizability are considered, one is the speed-up curves model, and the other is the directed-acyclic-graph (DAG) model. For both models, the objectives considered are the average, maximum, and ℓk-norms of flow time for k≥1.We establish an Ω(m) lower bound on the competitive ratio of any algorithm for optimizing average flow time in both models without resource augmentation. With resource augmentation, we give a (1+ϵ)-speed O(1ϵ2k+1)-competitive algorithm in the DAG model for the ℓk-norms of flow time. This essentially matches the best-known result in the speed-up curve model for the ℓk-norms of flow time. Finally, we show an O(1)-competitive algorithm for minimizing the maximum flow time in the speed-up curves model.

Cooperative and non-cooperative algorithms for distributed parallel jobs scheduling

Cooperative and non-cooperative algorithms for distributed parallel jobs scheduling

A job scheduling algorithm based on parallel workload prediction on computational grid

Generally, the computational grid consists of a large number of computing nodes, some of them are idle due to the uneven geographical distribution of computing requirements. This may cause workload unbalancing problems, which affect the performance of large-scale computational grids. In order to balance the computing requirements and computing nodes, we propose a job scheduling algorithm based on the workload prediction of computing nodes. We first analyze the causes of workload imbalance and the feasibility of reallocating computing resources. Secondly, we design an application and workload-aware scheduling algorithm (AWAS) by combining the previously designed workload prediction model. To reduce the complexity of the AWAS algorithm, we propose a parallel job scheduling method based on computing node workload prediction. The experiments show that the AWAS algorithm can balance the workload among different computing nodes on the real-world dataset. In addition, we propose the parallelism of workload prediction model from the perspective of internal structure and data set to make AWAS apply to more computing nodes of the large-scale computing grids. Experimental results show that the combination of the two can achieve satisfactory acceleration efficiency.

On Quantum Algorithms for Random Walks in the Nonnegative Quarter Plane

It is well known that strong connections exist between random walks (RWs) in the nonnegative quarter plane and the mathematical performance modeling, analysis and optimization of computer systems and communication networks. Examples include adaptive threshold-based load balancing in concert with affinity scheduling (e.g., [16, 10]), scheduling parallel jobs on parallel computer systems (e.g., [15]), and asymptotic scalabilty in wireless and linear loss networks (e.g., [9]).

I S pan : Parallel Identification of Strongly Connected Components with Spanning Trees

Detecting strongly connected components (SCCs) in a directed graph is crucial for understanding the structure of graphs. Most real-world graphs have one large SCC that contains the majority of the vertices as well as many small SCCs whose sizes are reversely proportional to the frequency of their occurrences. For both types of SCCs, current approaches that rely on depth first search (DFS) or breadth first search (BFS) face the challenges of both strict synchronization requirements and high computation cost. In this article, we advocate a new paradigm of identifying SCCs with simple spanning trees since SCC detection requires only the knowledge of connectivity among the vertices. We have developed a prototype called i S pan , which consists of parallel, relaxed synchronization construction of spanning trees for detecting large and small SCCs combined with fast trims for small SCCs. We further scale i S pan to the distributed memory system by applying different distribution strategies to the data and task parallel jobs. Not limited, we also extend i S pan to the GPU architecture. The evaluations show that i S pan is able to significantly outperform current state-of-the-art DFS- and BFS-based methods by an average 18× and 4×, respectively.

Resilient Scheduling of Moldable Parallel Jobs to Cope With Silent Errors

We study the resilient scheduling of moldable parallel jobs on high-performance computing (HPC) platforms. Moldable jobs allow for choosing a processor allocation before execution, and their execution time obeys various speedup models. The objective is to minimize the overall completion time or the makespan, when jobs can fail due to silent errors and hence may need to be re-executed after each failure until successful completion. Our work generalizes the classical scheduling framework for failure-free jobs. To cope with silent errors, we introduce two resilient scheduling algorithms, <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Lpa-List</small> and <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Batch-List</small> , both of which use the <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">List</small> strategy to schedule the jobs. Without knowing a priori how many times each job will fail, <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Lpa-List</small> relies on a local strategy to allocate processors to the jobs, while <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Batch-List</small> schedules the jobs in batches and allows only a restricted number of failures per job in each batch. We prove approximation ratios for the two algorithms under several prominent speedup models (e.g., roofline, communication, Amdahl, power, monotonic, and a mix model). An extensive set of simulations is conducted to evaluate different variants of the two algorithms, and the results show that they consistently outperform some baseline heuristics. Overall, our best algorithm is within a factor of 1.6 of a lower bound on average over the entire set of experiments, and within a factor of 4.2 in the worst case.

THE HUMAN CAPITAL: VITAL, SOFT, HARD, META AND MENTAL SKILLS

The article is devoted to there search of skills that form the human capital. The need for research is due to the transformation of social spaсe in general, as well as with changes in economic, educational and labor relations. These transformations have affected human professional activities and showed that a person is weak being faced with changes in the external and internal environment. There search revealed that the division of all skills into hard skills and soft skills, which takes place in modern science, has no exact basis. However, this division is very important, because these skills are necessary for a person to carry out successful professional activities. The research also showed that the isolated vital skills are not studied and do not have an exact terminological definition. In this regard, the authors proposed to single out mental skills and meta skills separately. The authors gave definitions and the list of skills that make up both groups. As a mental skills were named: 1) fast decision-making, fast retrieval of information from memory, quickability to react, rapid adaptation to changing rules, quick selection of required parameters for solving problems, which are due to the speed of thinking; 2) ability of a person to change his behavior in a changed situation, changing roles, ability to assess effects, ability to run multiple jobs in parallel, effectively switching from one task to another, planning the best effective ways to solve the problem, ability to solve complex problems, neutralization of routine work, which are determined by flexibility of thinking. Meta skills included logical thinking, critical thinking, analogical thinking, creative thinking, tactical and strategic thinking. At the end of the study, it was concluded that the implementation of all types of skills forms human capital. Human capital, in turn, comes into clash with other human capitals and in interaction with them and forms a human resource.

Near-Zero Downtime Recovery From Transient-Error-Induced Crashes

Due to the system scaling, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">transient errors</i> caused by external noise, e.g., heat fluxes and particle strikes, have become a growing concern for the current and upcoming exa-scale high-performance-computing (HPC) systems. Applications running on these systems are expected to experience transient errors more frequently than ever before, which will either lead them to generate incorrect outputs or cause them to crash. However, since such errors are still quite rare as compared to no-fault cases, desirable solutions call for low/no-overhead systems that do not compromise the performance under no-fault conditions and also allow very fast fault recovery to minimize downtime. In this article, we present <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> , a light-weight compiler-assisted resilience technique to quickly and accurately recover processes from transient-error-induced crashes. During the compilation of applications, <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> constructs a set of recovery kernels for crash-prone instructions. These recovery kernels are executed to repair the corrupted process states on-the-fly upon occurrences of errors, enabling applications to continue their executions instead of being terminated. When constructing recovery kernels, <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> exploits side effects introduced by induction variable based code optimization techniques based on loop unrolling and strength reduction to improve its recovery capability. To this end, two new code transformation passes are introduced to expose the side effects for resilience purposes. We evaluated <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> with 4 scientific workloads as well as the NPB benchmarks suite. During their normal execution, <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> incurs almost <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">zero</b> runtime overhead and a small, fixed <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">27MB</b> memory overhead. Meanwhile, <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> can recover on an average 83.55 percent of crash-causing errors within dozens of milliseconds with negligible downtime. We also evaluated <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> with parallel jobs running on 3072 cores and showed that <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> can successfully mask the impact of crash-causing errors by providing almost uninterrupted execution. Finally, we present our preliminary evaluation result for BLAS, which shows that <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> is capable of recovering failures in libraries with a very high coverage rate of 83 percent and negligible overheads. With such an effective recovery mechanism, <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IterPro</b> could tremendously mitigate the overheads and resource requirements of the resilience subsystem in future exa-scale systems.