Search

In this paper, we analyze a multi-server queueing system with a marked Markov arrival process of two types of customers and a phase-type distribution of service time depending on the type of customer. Customers of both types are assumed to be impatient and renege from the buffers after an exponentially distributed number of times. The strategy of flexible provisioning of priorities is analyzed. It assumes a randomized choice of the customers from the buffers, with probabilities dependent on the relation between the number of customers in a priority finite buffer and the fixed threshold value. To simplify the construction of the underlying Markov chain and the derivation of the explicit form of its generator, we use the so-called generalized phase-type distribution. It is shown that the created Markov chain fits the category of asymptotically quasi-Toeplitz Markov chains. Using this fact, we show that the considered Markov chain is ergodic for any value of the system parameters and compute its stationary distribution. Expressions for key performance measures are presented. Numerical results that show how the parameters of the control strategy affect the system’s performance measurements are given. It is shown that the results can be used for managerial purposes and that it is crucial to take correlation in the arrival process into account.

Exascale workloads are expected to incorporate data-intensive processing in close coordination with traditional physics simulations. These emerging scientific, data-analytics and machine learning applications need to access a wide variety of datastores in flat files and structured databases. Programmer productivity is greatly enhanced by mapping datastores into the application process's virtual memory space to provide a unified “in-memory” interface. Currently, memory mapping is provided by system software primarily designed for generality and reliability. However, scalability at high concurrency is a formidable challenge on exascale systems. Also, there is a need for extensibility to support new datastores potentially requiring HPC data transfer services. In this article, we present UMap , a scalable and extensible userspace service for memory-mapping datastores. Through decoupled queue management, concurrency aware adaptation, and dynamic load balancing, UMap enables application performance to scale even at high concurrency. We evaluate UMap in data-intensive applications, including sorting, graph traversal, database operations, and metagenomic analytics. Our results show that UMap as a userspace service outperforms an optimized kernel-based service across a wide range of intra-node concurrency by 1.22-1.9 ${\times}$ × . We performed two case studies to demonstrate UMap 's extensibility. First, a new datastore residing in remote memory is incorporated into UMap as an application-specific plugin. Second, we present a persistent memory allocator Metall built atop UMap for unified storage/memory.

The many-to-many assignment problem (M- MAP), and the CPU/FPGA scheduling problem are two correlated issues in the field of combinatorial optimization. The framework for Kernighan Lin-driven logarithmic barrier approach (KD-LBA) is made to solve the many-to-many assignment problem. KD-LBA is a deterministic technique, which initiates to achieve the globally optimal solutions for MMAP using logarithmic barrier function-based gradient descent technique. Then, the obtained solution is optimized further using the Kernighan Lin-based local search method. Successive Kernighan Lin-driven logarithmic barrier approach (Successive KD-LBA) is also proposed to sort the issue of scheduling in a CPU/FPGA heterogeneous system. It solves the CPU/ FPGA scheduling problem by transforming it into an MMAP. KD-LBA outperforms the state-of-art methods in terms of convergence speed for MMAPs with a group size greater than 40. Successive KD-LBA presents novel scheduling solutions for the CPU/FPGA scheduling problem as compared to the existing works, regarding average makespan and computation time.

Recently, significant improvements have been achieved for object detection algorithm by increasing the size of convolutional neural network (CNN) models, but the resulting increase of computational complexity poses an obstacle to practical applications. And some of the lightweight methods fail to consider the characteristics of object detection into and suffer a huge loss of accuracy. In this paper, we design a multi-scale feature lightweight network structure and specific convolution module for object detection based on depthwise separable convolution, which not only reduces the computational complexity but also improves the accuracy by using the specific position information in object detection. Furthermore, in order to improve the detection accuracy for small objects, we construct a multi-channel position-aware map and propose training based on knowledge distillation for object detection to train the lightweight model effectively. Last, we propose a training strategy based on a key-layer guiding structure to balance performance with training time. The experimental results show that on the COCO dataset that takes the state-of-the-art object detection algorithm, YOLOv3, as the baseline, our model size is compressed to 1/11 while accuracy drops by 7.4 mmAP, and the computational latency on the GPU and ARM platforms are reduced to 43.7% and 0.29%, respectively. Compared with the state-of-the-art lightweight object detection model, MNet V2 + SSDLite, the accuracy of our model increases by 3.5 mmAP while the inferencing time stays nearly the same. On the PASCAL VOC2007 dataset, the accuracy of our model increases by 5.2 mAP compared to the state-of-the-art lightweight algorithm based on knowledge distillation. Therefore, in terms of accuracy, parameter count, and real-time performance, our algorithm has better performance than lightweight algorithms based on knowledge distillation or depthwise separable convolution.

With the surge in online education, more universities have shifted classes online. The growing popularity of MOOC courses and the changing education landscape could mean more and more people switching to online education. A primary drawback is the difficulty in monitoring of students during an online examination which leads to a lot of malpractices used by candidates. This paper explores computer vision based techniques to propose a five-fold proctoring mechanism for online tests. The features incorporated are authentication, head movement, eye motion tracking, speech detection and object detection. The solution has an overall accuracy of 91% accuracy.

SSD (Single Shot Multibox Detector) is one of the most successful object detectors for its high accuracy and fast speed. However, the features from shallow layer (mainly Conv4_3) of SSD lack semantic information, resulting in poor performance in small objects. In this paper, we proposed DDSSD (Dilation and Deconvolution Single Shot Multibox Detector), an enhanced SSD with a novel feature fusion module which can improve the performance over SSD for small object detection. In the feature fusion module, dilation convolution module is utilized to enlarge the receptive field of features from shallow layer and deconvolution module is adopted to increase the size of feature maps from high layer. Our network achieves 79.7% mAP on PASCAL VOC2007 test and 28.3% mmAP on MS COCO test-dev at 41 FPS with only 300x300 input using a single Nvidia 1080 GPU. Especially, for small objects, DDSSD achieves 10.5% on MS COCO and 22.8% on FLIR thermal dataset, outperforming a lot of state-of-the-art object detection algorithms in both aspects of accuracy and speed.

In this paper, we consider two single server queueing systems to which customers of two distinct priorities (P1 and P2) arrive according to a Marked Markovian arrival process (MMAP). They are served according to two distinct phase type distributions. The probability of a P1 customer to feedback is θ on completion of his service. The feedback (P1) customers, as well as P2 customers, join the low priority queue. Low priority (P2) customers are taken for service from the head of the line whenever the P1 queue is found to be empty at the service completion epoch. We assume a finite waiting space for P1 customers and infinite waiting space for P2 customers. Two models are discussed in this paper. In model I, we assume that the service of P2 customers is according to a non-preemptive service discipline and in model II, the P2 customers service follow a preemptive policy. No feedback is permitted to customers in the P2 line. In the steady state these two models are compared through numerical experiments which reveal their respective performance characteristics.

Persistent Memory (PM) is a new device which provides faster access than conventional storage devices, such as SSDs. Among several methods prepared for accessing files on PM, a combination of filesystem direct access (DAX) and mmap() is used to take advantage of its native abilities. We can avoid buffer cache and access PM with byte granularity. To make sure that stored data is persisted, we need to synchronize the data on the CPU cache to PM. One option is an msync() system call and another is pmem_persist(). DAX enabled filesystems implement msync(), as it can synchronize data with CPU instructions, such as CLWB or CLFLUSHOPT. An alternative is pmem_persist(), which is included in the Persistent Memory Development Kit and issues those CPU instructions directly from the user space. msync() has the advantage of not having to modify legacy applications, while pmem_persist() is faster since it does not incur context switches. There are legacy applications which use mmap() and msync() to gain high performance, even though their original targets are traditional storage devices. They should be able to take advantage of PM's ability without any code modifications. Kyoto Cabinet, a library of routines for managing a database, is an example [1]. This paper reports on a case in which msync() resulted in tremendously slow performance on a DAX enabled ext4 filesystem. Throughput of a 4KiB random store followed by a 4KiB msync() was 1.1k IOPS, while 192k IOPS can be obtained with pmem_persist(). File size was 100MiB for both cases and the whole region was mapped. Moreover, a SET operation for Kyoto Cabinet database with OAUTOSYNC enabled takes 4.87ms when stored on a PM, while 0.77ms is taken when stored on an SSD. The time used for msync() is 3.00ms and 0.33ms respectively. PM was Intel Optane DC Persistent Memory and Linux kernel was 5.4.0. This performance loss comes from the side effect of the 2MiB hugepage.mmap() tries to allocate 2MiB pages instead of 4KiB pages when a region greater than 2MiB is requested. Since a dirty flag is associated to a whole 2MiB page, a kernel has to flush all of the 2MiB at once, even though msync() only requests a 4KiB flush. Though hugepage support has several advantages, such as fewer page faults and smaller page tables, this performance problem hinders legacy applications to use PM without any code modifications. The result below shows how this hugepage side effect is large. It is obtained with a kernel in which CONFIG_FS_DAX_PMD is disabled. Throughput of a 4KiB random store followed by a 4KiB msync() increases to 174k IOPS. In addition, the latency of a Kyoto Cabinet SET operation decreases to 0.30ms and the time spent for msync() decreases to 0.03ms. Though disabling hugepage results in better performance, it cannot be the best solution, as hugepage offers several advantages. So, I will conclude the abstract by mentioning a method which would avoid a performance loss, even with hugepages. Since the root cause is that a dirty flag manages the 2MiB page as a whole, preparing 512 dirty flags every 4KiB would work. It will reduce the size of wasted cache line flush to nearly zero when msync() requests the flushing of a small region. In addition, some optimizations may be required for a large flush request, since flushing 512 4KiB pages would take longer compared to flushing a 2MiB page.

Matrices representing genetic relatedness among individuals (i.e., Genomic Relationship Matrices, GRMs) play a central role in genetic analysis. The eigen-decomposition of GRMs (or its alternative that generates fewer top singular values using genotype matrices) is a necessary step for many analyses including estimation of SNP-heritability, Principal Component Analysis (PCA), and genomic prediction. However, the GRMs and genotype matrices provided by modern biobanks are too large to be stored in active memory. To accommodate the current and future “bigger-data”, we develop a disk-based tool, Out-of-Core Matrices Analyzer (OCMA), using state-of-the-art computational techniques that can nimbly perform eigen and Singular Value Decomposition (SVD) analyses. By integrating memory mapping (mmap) and the latest matrix factorization libraries, our tool is fast and memory-efficient. To demonstrate the impressive performance of OCMA, we test it on a personal computer. For full eigen-decomposition, it solves an ordinary GRM (N = 10,000) in 55 sec. For SVD, a commonly used faster alternative of full eigen-decomposition in genomic analyses, OCMA solves the top 200 singular values (SVs) in half an hour, top 2,000 SVs in 0.95 hr, and all 5,000 SVs in 1.77 hr based on a very large genotype matrix (N = 1,000,000, M = 5,000) on the same personal computer. OCMA also supports multi-threading when running in a desktop or HPC cluster. Our OCMA tool can thus alleviate the computing bottleneck of classical analyses on large genomic matrices, and make it possible to scale up current and emerging analytical methods to big genomics data using lightweight computing resources.

Memory-mapped I/O (mmio) is emerging as a viable alternative for accessing directly-attached fast storage devices compared to explicit I/O with system calls. Mmio removes the need for costly lookups in the DRAM I/O cache for cache hits, as they are handled in hardware via the virtual memory mechanism. In this work we present HugeMap, a custom mmio path in the Linux kernel that uses huge pages for file-backed mappings to accelerate applications with sequential I/O access patterns or large I/O operations. HugeMap uses huge pages to reduce CPU processing in the kernel I/O path compared to regular mmap. We explore the benefits and trade-offs of huge pages in HugeMap using microbenchmarks, IOR, and an in-house persistent key-value store designed for mmio. Our experiments show up to \(3.7\times \) higher throughput and up to \(4.76\times \) lower system time, compared to regular page configurations.

We consider a retrial queueing system with limited processor sharing which can be used for modeling the operation of a cell of fixed capacity in a wireless cellular network with two types of customers (handover and new customers). Customers of two types arrive at the system according to the Marked Markovian Arrival Process (MMAP). Arriving customers of each type follow a bandwidth sharing policy. In period when the number of customers of definite type in the system exceeds a threshold (different between new and handover customers) newly arriving customers of one type (handover customers) are considered to be lost while the customers of another type (new customers) go to orbit of infinite size. From the orbit, they try their attempts to reach a server in exponentially distributed time.

In this paper, we consider a single-server queueing inventory model with two types of customers, say, type-1 and type-2. The queue formed by the type-1 customer is infinite. The queue formed by the type-2 customer can accommodate a maximum of N customers. Even though the same server provides service to both types of customers, type-1 customers are served one by one, while type-2 customers are served in batches of varying sizes. The service is initiated only when inventory is available and the service time is assumed to be positive. If at least one item is available in the inventory and the server is idle, an arriving type-1 customer can directly enter into the service. The service of the type-2 customer is initiated either upon realization of a random clock that started ticking with the arrival of the first type-2 customer or by the accumulation of N type-2 customers, whichever occurs first. The arrivals follow a Marked Markovian process. Service time distribution of both type-1 and type-2 customers follow two different phase-type distribution. Replenishment of inventory follows the (s, S) policy having a positive lead time. It is assumed that N is less than s. Steady-state analysis, as well as evaluation of some performance measures, have been done. The model is analyzed numerically and graphically.

We consider a single-server queueing system with two parallel queues of which one is a finite buffer for priority customers and the other infinite for ordinary customers. Two types of customers arrive according to the Marked Markovian Arrival Process (MMAP). Service times are assumed to follow phase-type distributions. A customer gets priority either by paying a cost or by any other means. Priority customers receive service on the basis of a token system which works according to the following rule: \(K-1\) lower priority customers are served consecutively, and the \( K^{th}\) one is from the priority queue, if there is any. Priority customers have the right to take the strategic decision in choosing the queue on arrival, if such a customer joins an ordinary queue then he loses the special benefit that he would have got otherwise. We introduce the joining strategy for the priority customers and call it ‘K-policy’ along with probabilistic decision whether to give up their additional benefit (reward). Steady-state analysis of the model is done. Some system characteristics are evaluated, a social optimization problem is discussed and numerical illustrations are provided.

In many applications, existing image detection algorithms detect small targets in complex scenes with low recognition rate and biased classification. To address this issue, we develop a multi-target detection algorithm based on improved Retinanet. It consists of the following four parts: 1) extract rich texture features using the information interaction module; and 2) extract high-level abstract features through the improved FPN+ module; and 3) make full use of contextual information for detection leveraging SSH detection head; and 4) adopt the weighted loss for regression. For the accuracy of large and small categories and avoid bias, we employ the ensemble of loss function to optimize parameters. The experimental results confirm the feasibility of our detection algorithm, with the improved performance on four metrics. The mAP50 value and mmAP value increase to 96.75% and 75.08% respectively, and the ACD value decreases by 0.907.

The essential understanding of disease pathogenesis and enabling genetic findings to be used for developing new therapeutics, is missing in the identifications of genomic loci through whole genome association studies (GWAS). Here we describe a new computational method (mMap) that reduces this gap by characterizing the functional and regulatory impact of allelic variation. The method incorporates the precomputed annotations of 26 protein functional regions and eight regulatory regions and recover SNPs that fall/lie in these regions. After annotating SNPs to functional or regulatory data, method link them to biological functions and pathways, and predicts significantly disrupted biological regions, processes and pathways, by controlling false discovery through hypergeometric test. By doing so, the method limits data to human interpretation level by prioritizing SNPs that have the potential to mediate a biological phenotype. The method is applicable to procedures that rely on the understanding of the biological causal role of mouse SNPs and is available online. In two example mMap applications, including whole genomes SNPs data from 48 inbred mice strains, we identify biological mechanisms by which SNPs can regulate pathways to govern phenotypes by targeting different coding and regulatory regions, even in closely related strains.

[ENGLISH] Our world has been going through a digitization process for some decades now, with thetransformation curve being sharper in the last few years. The rise of this new digital societybrings with it a huge amount of information, and so it is paramount to create new architec-tures and hardware devices capable of storing and analyzing such information in a reason-able time. The advancements in some fields such as persistent storage is notable, while inother fields such as RAM some extra work must be done to counter the lack of physical im-provement. For example, NVMe discs offer high transfer speeds and low latency in data accesswhile offering high capacities and moderate price points. The RAM, on the other hand, hasadvanced less, being the transfer rates the only improved factor, letting both memory capac-ity and prices almost intact. This results in data intensive systems with insufficient RAM thatcontinuously have to access disk in order to execute their programs, bringing a significantdegradation of the overall performance. The objective of this project is to analyze means ofoperating with files other than the classic ones, analyzing the data access patterns of a givenapplication and developing a specific disk access policy for it to function with less hassle.Specifically, we will focus on converting one existing program to a new memory-interactingparadigm and developing new mapped-file page replacement policies at kernel level.

With the emerging of NVM (Non-Volatile Memories) technologies, NVMM-based (Non-Volatile Main Memories) file systems have attracted more and more attention. Compared to traditional file systems, most NVMM-based file systems bypass the page cache and the I/O software stack. With the new mmap interface known as the DAX-mmap interface (DAX: direct access), the CPU can access the NVMM much faster by loading from/storing to it directly. However, the existing file system benchmark tools are designed for traditional file systems and do not support the new features of NVMM-based file systems, so the returned results are very often not accurate. In this paper, a new benchmark tool called NVMFS-IOzone is proposed. The behavior of the tool is redesigned to reflect the new features of NVMM-based file systems. The NVM-lib from Intel is used instead of traditional msync() to keep data consistent when evaluating the performance of the DAX-mmap interface. Experimental results show that the new benchmark tool can reveal a hidden improvement of 1.4~2.1 times in NVMM-based file systems, which cannot be seen by the traditional evaluation tools. The data paths of direct load/store to NVMM and bypassing CPU cache are also provided to support the new ##features of NVMM-based file systems for multidimensional evaluation. Furthermore, embedded cleaning-ups has also been added to NVMFS-IOzone to support convenient evaluation consistency, which benefits both NVMM-based and non-NVMM-based file system benchmarking even for quick and casual tests. The whole experimental evaluation is based on real physical NVMs rather than simulated NVMs, and the experimental results confirm the effectiveness of our design.

The purpose of this Special Communication is to discuss the rationale and design of the Movement Matters Activity Program for Stroke (MMAP) and explore implementation successes and challenges in home health and outpatient therapy practices across the stroke belt state of North Carolina. MMAP is an interventional component of the Comprehensive Postacute Stroke Services Study, a randomized multicenter pragmatic trial of stroke transitional care. MMAP was designed to maximize survivor health, recovery, and functional independence in the community and to promote evidence-based rehabilitative care. MMAP provided training, tools, and resources to enable rehabilitation providers to (1) prescribe physical activity and exercise according to evidence-based guidelines and programs, (2) match service setting and parameters with survivor function and benefit coverage, and (3) align treatment with quality metric reporting to demonstrate value-based care. MMAP implementation strategies were aligned with the Expert Recommendations for Implementing Change project, and MMAP site champion and facilitator survey feedback were thematically organized into the Consolidated Framework for Implementation Research domains. MMAP implementation was challenging, required modification and was affected by provider- and system-level factors. Program and study participation were limited and affected by practice priorities, productivity standards, and stroke patient volume. Sites with successful implementation appeared to have empowered MMAP champions in vertically integrated systems that embraced innovation. Findings from this broad evaluation can serve as a road map for the design and implementation of other comprehensive, complex interventions that aim to bridge the currently disconnected realms of acute care, postacute care, and community resources.

We consider a single server retrial queueing system with parallel queues of ordinary and priority customers. Priority customers join the queues according to a probabilistic joining strategy until the ordinary queue is full. We also assume that there is a waiting room for the blocked ordinary customers and they make retrials from there, to enter into the ordinary queue. Priority customers on arrival, seeing both ordinary and priority queue as full loss the system forever. Customers receive service on the basis of a token system. Ordinary customers, upon arrival finding the ordinary queue as full, enter into the orbit. Ordinary customers make their retrials from the orbit in an exponential duration of time intervals to enter into the ordinary queue. Ordinary customer from the orbit make their retrials in an exponential duration of time intervals. Whenever the queue size of ordinary queue is less than N then the ordinary customer in the orbit successfully enter into the ordinary queue. Two types of customers arrive according to Marked Markovian Arrival Process (MMAP). Service times are assumed to follow phase type distributions. Steady-state analysis of the model is done. Some system characteristics are evaluated.

In this paper, we analyze a queueing system with two main unreliable servers and backup reliable server. The input flow is defined by the BMAP (Batch Markovian Arrival Process). Heterogeneous breakdowns arrive to the main servers according to a MMAP (Marked Markovian Arrival Process). Service times and repair times have PH (Phase type) distribution. The queueing system under consideration is an adequate model of operation of hybrid communication systems which combine the use of Free Space Optics and radio technologies. We derive a condition for the stable operation of the system, compute its stationary distribution and the key performance measures. Illustrative numerical examples give some insight into the behavior of the system.

In the present paper, the infinite-server queue model \(MMAP_{k}(t)|G_{k}|\infty \) in transient MMAP random environment with time varying marked MAP arrival of k types of customers subject to catastrophes is considered. The transient joint probability generating functions (PGF) of the number of different types of customers present in the model at moment t and the number of different types of customers departing from the system in the time interval (0, t] are found. The Laplace-Stieltjes transform (LST) of total volume of customers being in service at moment t is defined. The basic differential equations for joint probability generating functions of the number of busy servers and served customers for transient and stationary random environment are obtained.

The emergence of machine learning applications has brought the new demands for high performance computing in cloud environment. Besides accelerators, such as GPU or TPU, fast interconnection among and within computers becomes more and more important to achieve efficient training and learning. One of the high-bandwidth, low-latency interconnection architectures is InfiniBand. However, software based virtualization of InfiniBand in QEMU/KVM suffers large performance degradation owing to virtualization overhead and memory allocation problem. In this paper, two techniques, doorbell mapping and memlink, are proposed to optimize the performance the InfiniBand virtualization on QEMU/KVM. Doorbell mapping allows the applications in guest user-space to access the doorbell memory page directly so that the virtualization overhead is minimized. Memlink ensures memory contiguity after virtualization, which is a critical requirement for zero-copy between guest and host. Experiments show that the virtualized InfiniBand with mmap and memlink can achieve near native performance for large data transmissions. Comparing to the previous InfiniBand virtualization on QEMU/KVM, our implementation obtains over 3.5 times performance improvement in various benchmarks.

Leadership supercomputers feature a diversity of storage, from node-local persistent memory and NVMe SSDs to network-interconnected flash memory and HDD. Memory mapping files on different tiers of storage provides a uniform in- terface in applications. However, system-wide services like mmap are optimized for generality and lack flexibility for enabling application-specific optimizations. In this work, we present UMap to enable user-space page management that can be easily adapted to access patterns in applications and storage characteristics. UMap uses the userfaultfd mechanism to handle page faults in multi-threaded applications efficiently. By providing a data object abstraction layer, UMap is extensible to support various backing stores. The design of UMap supports dynamic load balancing and I/O decoupling for scalable performance. UMap also uses application hints to improve the selection of caching, prefetching, and eviction policies. We evaluate UMap in five benchmarks and real applications on two systems. Our results show that leveraging application knowledge for page management could substantially improve performance. On average, UMap achieved 1.25 to 2.5 times improvement using the adapted configurations compared to the system service.

We study a two-layer millimeter wave (mmWave) network with multi-connectivity. Each user equipment (UE) accesses to multiple mmWave access points (mmAPs) and requests services. It is a challenging task to control the load caused on backhauls considering both the service requirement of UEs and the capacity of backhauls in this system. As such, we develop a multi-leader multi-follower Stackelberg game to analyze the power control problem. In this game, each mmAP limits the service requirements of UEs by charging a price on unit transmit power, and each UE purchases a certain amount of transmit power at each time slot based on the prices. Given the prediction of UEs’ optimal strategy, we solve a Nash Equilibrium Problem (NEP) among the mmAPs with the sub-gradient method. By achieving the Stackelberg equilibrium (SE), UEs can obtain satisfying services without causing intolerable load on the backhauls. The Simulation shows our proposed strategy significantly increases the revenue and utility for both mmAPs and UEs.

Address translation fundamentally embodies a translation function that maps from virtual to physical addresses. In current systems, the translation function is encoded by the kernel in an in-memory radix tree structure (the page table hierarchy) which is then interpreted by the hardware (the pagewalker, pagewalk-caches, and TLBs). We consider implementing the translation function itself as reconfigurable hardware—does this make any sense? To study this question, we collected numerous in-situ Linux page tables for a wide range of workloads, including those from HPC, to serve as example translation functions. We then prototyped several potential mechanisms to implement the translation function, including inverted page tables with function-specific perfect hashing, translation functions directly implemented using Espresso-minimized PLAs, translation functions genetically-evolved in a language suitable for FPGA-like synthesis, and translation functions based on recovered/manufactured region (segment/mmap) lookup using multiplexor trees. Each mechanism was then evaluated using the Linux page tables, primarily for space and lookup speed. We report our findings and try to address the question.

Inspired by the ITER organization's CODAC system, the current experimental system architecture is being progressively redesigned using the experimental physics and industrial control system (EPICS) to meet the needs of future long pulse experiments. As part of the overall experimental system, the data processing and storage system (DPSS) is mainly used to provide services of data processing and data storage for the experimental system. However, with the data processing mechanism of the previous DPSS, the previous experimental system is unsatisfactory in terms of real-time experimental data feedback, and the data processing mechanism of the previous DPSS is not compatible with the new architecture. Given that the original DPSS no longer fits the current experimental characteristics in terms of both functionality and demand so that it is necessary to redesign and upgrade the original DPSS to adapt to the new architecture. In view of the soft real-time and distributed characteristics of EPICS, the upgraded DPSS will be redesigned based on the data processing mechanism of EPICS. And it uses an adaptive strategy that allows the experimental system to automatically match more appropriate data uploading modes according to the rate of data acquisition. For this purpose, a data transmission protocol based on the TCP/IP protocol has been proposed. With regard to software optimization, the upgraded DPSS uses technologies of double-buffer and MMAP to optimize data exchanging between memory spaces. In terms of data storage, the upgraded DPSS adopts the segment strategy to manage experimental data efficiently. In addition, to prevent problems caused by abnormal circumstances, a guarantee mechanism for data storage is provided in the upgraded DPSS. After being tested, the upgraded DPSS will later be integrated into the general control system of EAST NBI.

In this paper, we consider a queue where the inter-arrival times are correlated and, additionally, service times are also correlated with inter-arrival times. We show that the resulting model can be interpreted as an MMAP[K]/PH[K]/1 queue for which matrix geometric solution algorithms are available. The major result of this paper is the presentation of approaches to fit the parameters of the model, namely the MMAP, the PH distribution and the parameters introducing correlation between inter-arrival and service times, according to some trace of inter-arrival and corresponding service times. Two different algorithms are presented. The first algorithm is based on available methods to compute a MAP from the inter-arrival times and a PH distribution from the service times. Afterward, the correlation between inter-arrival and service times is integrated by solving a quadratic programming problem over some joint moments. The second algorithm is of the expectation maximization type and computes all parameters of the MAP and the PH distribution in an iterative way. It is shown that both algorithms yield sufficiently accurate results with an acceptable effort.

Main memory is a critical component for modern computer systems, and a requirement for a larger memory capacity has been increasing every year. Although DRAM has been commonly used as main memory, its capacity and cost are not expected to scale further. In contrast, those of nonvolatile memory used in block devices, such as NAND flash memory used in SSDs, still continue to scale. Therefore, it is a promising approach to configure high-capacity and low-cost hybrid memory systems by using block devices as an extension of DRAM. A memory-mapped file is a useful feature for this approach, because it enables applications to transparently access data stored on block devices. However, the write latency of applications becomes very high when they update data not stored in fast DRAM, because an operating system needs to synchronously read corresponding pages from slow block devices. In this paper, we propose AR-MMAP that is an asynchronous read method to reduce the write latency of applications applying memory-mapped files. It hides the read latency of block devices by reading the corresponding pages from block devices in background. As AR-MMAP requires the assistance of applications to guarantee data consistency, we modify an in-memory key-value store as a use case. We implement AR-MMAP in a Linux kernel and evaluate its performance on a server containing two types of SSDs. The evaluation results demonstrate that it reduces the execution time compared to a default kernel.

We consider a tandem queueing network with two service stations without buffers. An infinite capacity station named as the main station provides usual paid service. Another station is a finite capacity station named as the offer zone. The offer zone is an intermediate station strategically designed to attract the maximum number of customers to the main station. The offer zone works under various random environments. Sojourn times of each random environment follows Phase Type distribution. Two types of customers arrive to the system according to an MMAP. Service times of customers at both the stations are exponentially distributed. The stationary probability distribution of the states of the Markov chain representing the proposed model is computed. Some operational and probabilistic characteristics of the system are determined. A control problem is discussed. A cost function is proposed. The effect of the maximum capacity of the offer zone on various performance measures are considered. Numerical as well as graphical illustrations are given.

A complex multi-state system subject to different types of failures, repairable and/or nonrepairable, external shocks and preventive maintenance is modelled by considering a discrete Markovian arrival process with marked arrivals (D-MMAP). The internal performance of the system is composed of several degradation states partitioned into minor and major damage states according to the risk of failure. Random external events can produce failures throughout the system. If an external shock occurs, there may be an aggravation of the internal degradation, cumulative external damage or extreme external failure. The internal performance and the cumulative external damage are observed by random inspection. If major degradation is observed, the unit goes to the repair facility for preventive maintenance. If a repairable failure occurs then the system goes to corrective repair with different time distributions depending on the failure state. Time distributions for corrective repair and preventive maintenance depend on the failure state. Rewards and costs depending on the state at which the device failed or was inspected are introduced. The system is modelled and several measures of interest are built into transient and stationary regimes. A preventive maintenance policy is shown to determine the effectiveness of preventive maintenance and the optimum state of internal and cumulative external damage at which preventive maintenance should be taken into account. A numerical example is presented, revealing the efficacy of the model. Correlations between the numbers of different events over time and in non-overlapping intervals are calculated. The results are expressed in algorithmic-matrix form and are implemented computationally with Matlab., Junta de Andalucía, Spain, under the grant FQM307, Ministerio de Economía y Competitividad, España, MTM2017-88708-P, European Regional Development Fund (ERDF)

At EUROCRYPT 2013, Garg, Gentry and Halevi proposed a candidate construction (later referred as GGH13) of cryptographic multilinear map (MMap). Despite weaknesses uncovered by Hu and Jia (EUROCRYPT 2016), this candidate is still used for designing obfuscators.The naive version of the GGH13 scheme was deemed susceptible to averaging attacks, i.e., it could suffer from a statistical leak (yet no precise attack was described). A variant was therefore devised, but it remains heuristic. Recently, to obtain MMaps with low noise and modulus, two variants of this countermeasure were developed by Döttling et al. (EPRINT:2016/599).In this work, we propose a systematic study of this statistical leakage for all these GGH13 variants. In particular, we confirm the weakness of the naive version of GGH13. We also show that, among the two variants proposed by Döttling et al., the so-called conservative method is not so effective: it leaks the same value as the unprotected method. Luckily, the leakage is more noisy than in the unprotected method, making the straightforward attack unsuccessful. Additionally, we note that all the other methods also leak values correlated with secrets.As a conclusion, we propose yet another countermeasure, for which this leakage is made unrelated to all secrets. On our way, we also make explicit and tighten the hidden exponents in the size of the parameters, as an effort to assess and improve the efficiency of MMaps.

Network On Chip is determined as an On-chip packet-switched communication in the study of multi-processors. As the number of components is proliferating, there were many issues regarding the performance of a network. By taking into consideration of this problem, in this paper, we proposed an efficient core mapping algorithm named BMAP, used for customizing the NoC platforms and performing the SPLASH-2 Benchmark synthesis evaluation. This proposed mapping technique is applied to various NoC applications and SPLASH-2 Benchmark applications. The experimental outcomes exhibit optimized NoC platforms on outperforming the performance and cost metrics when compared with other mapping algorithms. In contrast, the SPLASH-2 Benchmarks reveal that the proposed algorithm has a significant improvement in performance metrics against NMAP, MMAP, and EMAP algorithms. The metrics such as Speed-up Execution Time has improved by 40%, 30%, and 20%, and Latency reduced by 42%, 34%, and 28%, Energy efficiency was improved by 36%, 30%, 26%, and Power consumption reduced by 32.6%, 28.2%, and 26.4% respectively.