Juggle: addressing extrinsic load imbalances in SPMD applications on multicore computers

We investigate proactive dynamic load balancing on multicore systems, in which threads are continually migrated to reduce the impact of processor/thread mismatches. Our goal is to enhance the flexibility of the SPMD-style programming model and enable SPMD applications to run efficiently in multiprogrammed environments. We present Juggle, a practical decentralized, user-space implementation of a proactive load balancer that emphasizes portability and usability. In this paper we assume perfect intrinsic load balance and focus on extrinsic imbalances caused by OS noise, multiprogramming and mismatches of threads to hardware parallelism. Juggle shows performance improvements of up to 80 % over static load balancing for oversubscribed UPC, OpenMP, and pthreads benchmarks. We also show that Juggle is effective in unpredictable, multiprogrammed environments, with up to a 50 % performance improvement over the Linux load balancer and a 25 % reduction in performance variation. We analyze the impact of Juggle on parallel applications and derive lower bounds and approximations for thread completion times. We show that results from Juggle closely match theoretical predictions across a variety of architectures, including NUMA and hyper-threaded systems.     

neCODEC: nearline data compression for scientific applications

Advances on multicore technologies lead to processors with tens and soon hundreds of cores in a single socket, resulting in an ever growing gap between computing power and available memory and I/O bandwidths for data handling. It would be beneficial if some of the computing power can be transformed into gains of I/O efficiency, thereby reducing this speed disparity between computing and I/O. In this paper, we design and implement a NEarline data COmpression and DECompression (neCODEC) scheme for data-intensive parallel applications. Several salient techniques are introduced in neCODEC, including asynchronous compression threads, elastic file representation, distributed metadata handling, and balanced subfile distribution. Our performance evaluation indicates that neCODEC can improve the performance of a variety of data-intensive microbenchmarks and scientific applications. Particularly, neCODEC is capable of increasing the effective bandwidth of S3D, a combustion simulation code, by more than 5 times.     

Delta Execution: A preemptive Java thread migration mechanism

Delta Execution is a preemptive and transparent thread migration mechanism for supporting load distribution and balancing in a cluster of workstations. The design of Delta Execution allows the execution system to migrate threads of a Java application to different nodes of a cluster so as to achieve parallel execution. The approach is to break down and group the execution context of a migrating thread into sets of consecutive machine-dependent and machine-independent execution sub-contexts. Each set of machine-independent sub-contexts, also known as a delta set, is then migrated to a remote node in a regulated manner for continuing the execution. Since Delta Execution is implemented at the virtual machine level, all the migration-related activities are conducted transparently with respect to the applications. No new migration-related instructions need to be added to the programs and existing applications can immediately benefit from the parallel execution capability of Delta Execution without any code modification. Furthermore, because the Delta Execution approach identifies and migrates only the machine-independent part of a thread's execution context, the implementation is therefore reasonably manageable and the resulting software is portable. This revised version was published online in July 2006 with corrections to the Cover Date.     

The physicochemical essence of the purine·pyrimidine transition mismatches with Watson-Crick geometry in DNA: A·C* versa A*·C. A QM and QTAIM atomistic understanding

It was established for the first time by DFT and MP2 quantum-mechanical (QM) methods either in vacuum, so in the continuum with a low dielectric constant (ε = 4), typical for hydrophobic interfaces of specific protein-nucleic acid interactions, that the repertoire for the tautomerisation of the biologically important adenine·cytosine* (A·C*) mismatched DNA base pair, formed by the amino tautomer of the A and the imino mutagenic tautomer of the C, into the A*·C base mispair (?G = 2.72 kcal?mol^?1 obtained at the MP2 level of QM theory in the continuum with ε = 4), formed by the imino mutagenic tautomer of the A and the amino tautomer of the C, proceeds via the asynchronous concerted double proton transfer along two antiparallel H-bonds through the transition state (TS_A·C*?A*·C). The limiting stage of the A·C*→A*·C tautomerisation is the final proton transfer along the intermolecular N6H···N4 H-bond. It was found that the A·C*/A*·C DNA base mispairs with Watson–Crick geometry are associated by the N6H?N4/N4H?N6, N3H?N1/N1H?N3 and C2H?O2 H-bonds, respectively, while the TS_A·C*?A*·C is joined by the N6–H–N4 covalent bridge and the N1H?N3 and C2H?O2 H-bonds. It was revealed that the A·C*?A*·C tautomerisation is assisted by the true C2H?O2 H-bond, that in contrast to the two others conventional H-bonds exists along the entire intrinsic reaction coordinate (IRC) range herewith becoming stronger at the transition from vacuum to the continuum with ε = 4. To better understand the behavior of the intermolecular H-bonds and base mispairs along the IRC of the A·C*?A*·C tautomerisation, the profiles of their electron-topological, energetical, geometrical, polar and charge characteristics are reported in this study. It was established based on the profiles of the H-bond energies that all three H-bonds are cooperative, mutually strengthening each other. The nine key points, providing a detailed physicochemical picture of the A·C*?A*·C tautomerisation, were revealed and thoroughly examined along the IRC. It was shown that the A*·C base mispair with the population ~1 % obtained at the MP2 level of QM theory in the continuum with ε = 4 is thermodynamically and dynamically stable structure. Its lifetime was calculated to be 5.76·10^?10 s at the MP2 level of QM theory in the continuum with ε = 4. This lifetime, from the one side, enables all six low-frequency intermolecular vibrations to develop, but, from the other side, it is by order less than the time (several ns) required for the replication machinery to forcibly dissociate a base pair into the monomers during DNA replication. This means that the A*·C base mispair “slips away from the hands” of the replication machinery into the A·C* mismatched base pair. Consequently, the authors came to the conclusion that exactly the A·C* base mispair is an active player of the point mutational events and is effectively dissociated by the replication machinery into the A and C* monomers in contrast to the A*·C base mispair, playing the mediated role of a provider of the A·C* base mispair in DNA that is synthesised.     

An Adaptive Hybrid OLAP Architecture with optimized memory access patterns

OLAP (On-Line Analytical Processing) is an approach to efficiently evaluate multidimensional data for business intelligence applications. OLAP contributes to business decision-making by identifying, extracting, and analyzing multidimensional data. The fundamental structure of OLAP is a data cube that enables users to interactively explore the distinct data dimensions. Processing depends on the complexity of queries, dimensionality, and growing size of the data cube. As data volumes keep on increasing and the demands by business users also increase, higher processing speed than ever is needed, as faster processing means faster decisions and more profit to industry. In this paper, we are proposing an Adaptive Hybrid OLAP Architecture that takes advantage of heterogeneous systems with GPUs and CPUs and leverages their different memory subsystems characteristics to minimize response time. Thus, our approach (a) exploits both types of hardware rather than using the CPU only as a frontend for GPU; (b) uses two different data formats (multidimensional cube and relational cube) to match the GPU and CPU memory access patterns and diverts queries adaptively to the best resource for solving the problem at hand; (c) exploits data locality of multidimensional OLAP on NUMA multicore systems through intelligent thread placement; and (d) guides its adaptation and choices by an architectural model that captures the memory access patterns and the underlying data characteristics. Results show an increase in performance by roughly four folds over the best known related approach. There is also the important economical factor. The proposed hybrid system costs only 10 % more than same system without GPU. With this small extra cost, the added GPU increases query processing by almost 2 times.     

Short and long range order of the morphology of silk from Latrodectus hesperus (Black Widow) as characterized by atomic force microscopy.

The surfaces of both stretched and unstretched silk threads from the cobweb weaver, Latrodectus hesperus (Black Widow) have been examined by atomic force microscopy (AFM). AFM images of cobweb scaffolding threads show both unordered and highly ordered regions. Two types of fibers within the threads were observed: thicker (approximately 300 nm in diameter) fibers oriented parallel to the thread axis and thinner (10-100 nm) fibrils oriented across the thread axis. While regions which lacked parallel fibers or fibrils were observed on threads at all strain values, the probability of observing fibers and/or fibrils increased with strain. High-resolution AFM images show that with increasing strain, both mean fiber and fibril diameters decrease and that fibrils align themselves more closely with the thread axis. The observation of fibers and fibrils within the cobweb threads has implications for current models of the secondary and tertiary structure and organization of spider silk.     

Theoretical evaluation of inhibition performance of purine corrosion inhibitors

Inhibition mechanism of three purine compounds, adenine (A), 2-amino-6-thiol-9H-purine (B) and 2,6-dithiol-9H-purine (C), was investigated by quantum chemical calculation and molecular dynamic simulation. The molecular reactivity was studied by quantum chemical calculation, and the distribution of the highest occupied molecular orbital (HOMO), the lowest unoccupied molecular orbital (LUMO), the energy gap between HOMO and LUMO and the Fukui index were proposed to describe the active sites of molecules, and the inferred inhibition efficiency followed the order of A < B < C. Furthermore, the adsorption behaviour of these three purine molecules on a metal surface was investigated via molecular dynamics simulation. The analysis of adsorption configuration indicated that these three purine molecules adsorbed parallely onto the metal surface, and the inferred inhibition efficiency from interaction energy also followed the order of A < B < C. These inferred inhibition efficiency from theoretical calculation was in good accordance with experimental results. This accordance indicated that our proposed theoretical method might be a feasible approach to assess the inhibition performance of inhibitors. Moreover, our research was helpful to filter the aimed inhibitor and design of the new inhibitor.     

An efficient approach to collaborative simulation of variable structure systems on multi-core machines

Complex variable-structure systems (CVSSs) are a common type of complex systems that exhibit changes both at structural and behavior levels. Simulations of CVSSs challenge current collaborative execution methods with increasingly big and complex models. The emergence of multi-core paradigm presents an exciting opportunity to address such challenge, so an advanced parallel simulator under multi-core environments is proposed. The simulator: (1) provides thread simulation kernels and five kinds of management services to support dynamic model structure flexibly; (2) can explore both inherent and dynamic parallelism among models based on interaction relations, and employ the multi-thread paradigm to gain good speedup; (3) adopts an efficient dynamic load-balancing method, which can migrate models among cores with very low cost and support dynamic core allocation on demand, to address evident load-imbalance problems brought by variable-structure. The experiments show that structure changes can be supported while up to 23 % performance increase can be gained.     

Factors governing the stickiness of cribellar prey capture threads in the spider family Uloboridae

The surface of a cribellar prey capture thread is formed of thousands of fine, looped fibrils, each issuing from one of the spigots on an oval spinning plate termed the cribellum. This plesiomorphic capture thread is retained by members of the family Uloboridae, in which its stickiness differs among genera. An examination of five cribellar thread features in nine uloborid species shows that only the number of fibrils that form a thread explains these differences in thread stickiness. Neither the physical features of these fibrils, nor the manner in which they are combined to form threads differs among species. Threads produced by orb-weaving species contain fewer fibrils than those produced by species that build reduced webs. Relative to spider weight, the number of fibrils that form a cribellar thread is greatest in simple-web species of the genus Miagrammopes, less in triangle-web species of the genus Hyptiotes, and least in orb-weaving species representing five genera. A transformational analysis shows that change in the number of cribellum spigots is directly related to change in the stickiness of cribellar thread. This direct relationship between the material invested in a cribellar thread and its stickiness may have been a limiting factor that favored the switch from the dry cribellar threads of uloborids to the adhesive capture threads produced by other orb-weaving families. © 1994 Wiley-Liss, Inc.     

Early effects of aluminum chloride on beta-secretase mRNA expression in a neuronal model of ß-amyloid toxicity

Amyloid ß peptide (Aß), generated by proteolytic cleavage of the amyloid precursor protein (APP), plays a pivotal role in the pathogenesis of Alzheimer's disease (AD). The key step in the generation of Aß is cleavage of APP by ß-secretases (beta-site APP-cleaving enzyme 1 (BACE1) and BACE2). There has been suggestion of interaction between aluminum and several AD-associated pathways. However, the underlying mechanisms still remain unclear. Here, we report the effects of aluminum chloride (AlCl₃) in Aß-induced toxicity using differentiated neuronal SH-SY5Y cells. The metal significantly enhances Aß-induced cell death at concentrations ranging from 50 to 300 µM after 24 and 48 h. After 72 and 96 h treatment, cell death is increased already at 10 µM. Early coexposure of cells to 10 µM AlCl₃ and 2 µM Aß differentially affected ß-secretase mRNA levels as compared to single Aß treatment after 1 and 3 h. BACE1 levels were slightly reduced after 1 h and significantly increased after 3 h exposure, whereas BACE2 levels were increased at both times considered. Both genes’ mRNA levels were downregulated at longer times (6, 12, and 24 h). Although these results indicate that aluminum toxicity is correlated to changes in both BACE1 and BACE2 expression levels, the subsequent common downregulation observed suggests that aluminum involvement in the Aß cascade is subtle, and other underlying mechanisms might be involved.     

Sequence Variation Of the α-Lactalbumin Gene In Holstein and Nellore Cows

The α-lactalbumin is a subunit of lactose-synthase, an enzyme responsible for lactose production, a disaccharide that influences milk production. Sequence variations of bovine α -lactalbumin have been associated with differences in milk yield. This study aimed to analyze allelic frequency differences at position ? 1689 (g. A > G) and + 15 (g. A > G) of the α-lactalbumin gene in Holstein (Bos taurus) and Nellore (Bos indicus) cows. Blood samples were analyzed from 34 Holstein, 104 Nellore, and 99 Dairy Nellore cows using PCR-RFLP. The different RFLP patterns were sequenced and a novel sequence variation on nucleotide ? 46 was identified. An adenine at this position was designated as the A allele and a guanine was designated B allele. The frequencies of alleles A ? 1689, A ? 46, and A + 15 differed between Holstein and both Nellore breeds. The results show that differences in α-lactalbumin allelic variants in the 5′-flanking and the 5′-UTR region might be associated with differences in milk production between Holstein cows and cows from Nellore breeds. However, the lack of difference between Nellore and Dairy Nellore suggests that other sequence variantions that regulate milk production might be responsible for the selection of Dairy Nellore cows with superior milk production.     

The silk road of <Emphasis Type="Italic">Tetranychus urticae</Emphasis>: is it a single or a double lane?

Tetranychus urticae (Acari: Tetranychidae) is a phytophagous mite that forms huge colonies. All active members of a colony (immatures and matures, females and males) spin silken threads. These mites construct a common web that protects the colony from external aggression. The silk coverage is well-known to provide advantages to the colony but very little is known about the characteristics of the threads themselves. Here is the first quantification of the diameter of silken threads spun by two different stages (adult females and larvae) and its relationship with body size of the spinning individuals. Moreover, we observed how silk was deposited on the substrate through their two pedipalps. Threads were observed by means of transmission electron and fluorescence microscopy. Silken threads spun by larvae (0.055 ± 0.018 μm) were significantly thinner than threads spun by adult females (0.111 ± 0.038 μm). In the first step of the silk depositing behaviour, the mite attached the thread to the substrate by putting its pedipalps in contact with the surface (adhesion, double silken threads). When walking, silken threads became detached from the substrate and spitted up (silken threads were free). Finally, silken threads adhered to the surface. The presence of single and double threads makes thread diameter highly variable.     

Performance Monitoring and Analysis of Task-Based OpenMP

OpenMP, a typical shared memory programming paradigm, has been extensively applied in high performance computing community due to the popularity of multicore architectures in recent years. The most significant feature of the OpenMP 3.0 specification is the introduction of the task constructs to express parallelism at a much finer level of detail. This feature, however, has posed new challenges for performance monitoring and analysis. In particular, task creation is separated from its execution, causing the traditional monitoring methods to be ineffective. This paper presents a mechanism to monitor task-based OpenMP programs with interposition and proposes two demonstration graphs for performance analysis as well. The results of two experiments are discussed to evaluate the overhead of monitoring mechanism and to verify the effects of demonstration graphs using the BOTS benchmarks.     

A stochastic approach to estimating earliest start times of nodes for scheduling DAGs on heterogeneous distributed computing systems

Previously, DAG scheduling schemes used the mean (average) of computation or communication time in dealing with temporal heterogeneity. However, it is not optimal to consider only the means of computation and communication times in DAG scheduling on a temporally (and spatially) heterogeneous distributed computing system. In this paper, it is proposed that the second order moments of computation and communication times, such as the standard deviations, be taken into account in addition to their means, in scheduling “stochastic” DAGs. An effective scheduling approach which accurately estimates the earliest start time of each node and derives a schedule leading to a shorter average parallel execution time has been developed. Through an extensive computer simulation, it has been shown that a significant improvement (reduction) in the average parallel execution times of stochastic DAGs can be achieved by the proposed approach.     

Can the modeling for simplification of a dental implant surface affect the accuracy of 3D finite element analysis?

The aim of this study was to assess stress/strain of different implant modeling simplifications by 3D-FEA. Three variation of external hexagon implant (Ø3.75?×?10 mm) supporting one molar crown were simulated: A (no threads); B (slightly threads simplification); C (original design). 200 N (axial) and 100 N (oblique) were applied. Cortical bone was evaluated by maximum principal stress and microstrain qualitatively and quantitatively (ANOVA and Tukey post hoc (p < 0.05)). Higher stress levels (p < 0.05) were observed in model A. Models B and C presented similar stress transmission. It was possible to conclude that slightly simplification should be used for studies evaluating stress transferring for bone tissue.     

Growth rate,labeling index,and radiation survival of cells grown in the matrigel threadin vitro tumor model

Summary Six rodent cell lines (36B10 rat glioma cells, 9L rat gliosarcoma cells, V79 Chinese hamster lung fibroblasts, EMT6/UW and EMT6/Ro mouse mammary sarcoma cells, and RIF-1 mouse fibrosarcoma cells) were tested for growth in cylindrical threads of Matrigel. These cells grew in the threads with doubling times of 17–23 h, reaching maximum cell densities on the order of 10⁸ cells/ml. Histological sections of these threads showed a heterogeneous cell distribution: cells grew to confluence at the thread surface and at somewhat lower cell densities in the thread core. [H-3]thymidine labeling index and radiation sensitivity were measured for 9L and EMT6/UW cells in Matrigel threads. For both cell types, the labeling index in Matrigel was lower than observed in cell monolayers, with higher labeling indexes at the thread periphery than in the thread core. When these threads were grown in stirred medium, lower thread diameters, higher cell yields per thread, and higher labeling indices were obtained. EMT6 cell monolayers coated with Matrigel were less radiosensitive than cells in uncoated monolayers. This protective effect was eliminated by irradiating in the presence of 1 mg/ml misonidazole. EMT6 cells consume nearly three times as much oxygen (mole/cm³-sec) as do 9L cells, which are equally radiosensitive in monolayers with or without a Matrigel coating. The radiation sensitivity of EMT6/UW cells in Matrigel threads was similar to that for monolayers of plateau phase cells, whereas for 9L cells, the response in threads was more similar to exponentially growing cells. We conclude that Matrigel threads provide an alternativein vitro model for studying the radiation response of cells in a three-dimensional geometry.     

Evolution of adhesive mechanisms in cribellar spider prey capture thread: evidence for van der Waals and hygroscopic forces

Sticky prey capture threads are produced by many members of the spider infraorder Araneomorphae. Cribellar threads are plesiomorphic for this clade, and viscous threads are apomorphic. The outer surface of cribellar thread is formed of thousands of fine, looped fibrils. Basal araneomorphs produce non-noded cribellar fibrils, whereas more derived members produce noded fibrils. Cribellar fibrils snag and hold rough surfaces, but other forces are required to explain their adherence to smooth surfaces. Threads of Hypochilus pococki (Hypochilidae) formed of non-noded fibrils held to a smooth plastic surface with the same force under low and high humidities. In contrast, threads of Hyptiotes cavatus and Uloborus glomosus (Uloboridae) formed of noded fibrils held with greater force to the same surface at intermediate and high humidities. This supports the hypothesis that van der Waals forces allow non-noded cribellar fibrils to adhere to smooth surfaces, whereas noded fibrils, owing to the hydrophilic properties of their nodes, add hygroscopic forces at intermediate and high humidities. Thus, there appear to have been two major events in the evolution of adhesive mechanisms in spider prey capture thread: the addition of hydrophilic nodes to the fibrils of cribellar threads and the replacement of cribellar fibrils by viscous material and glycoprotein glue.  © 2002 The Linnean Society of London, Biological Journal of the Linnean Society , 2002, 77 , 1–8.     

Keratin-like components of gland thread cells modulate the properties of mucus from hagfish (Eptatretus stouti)

Summary The hagfishes (cyclostomes) are known to secrete copious amounts of mucus mainly by the holocrine mode from the slime glands. Stressed animals release two types of cells (gland thread cells, GTCs; gland mucous cells, GMCs) which rupture on contact with water and rapidly form a mass of viscous mucus. Herein we report some key sequential events of this process and document a novel role for cytoskeletal polymers. After electrostimulation of Pacific hagfish (Eptatretus stouti), the exudate was collected in a stabilization buffer and GTCs segregated from GMC vesicles. Water was added progressively to mixtures of known quantities of these entities. The changing mucous composition and properties were monitored by light- and electron microscopy, viscometry and immunogold assay. Sequentially, the threads uncoil from GTCs, aggregate with the vesicles, the vesicles rupture and release mucin-like substances, at least some of which adhere to the thread. It was found that the intermediate filament (IF)-rich threads markedly facilitate hydration and modulate the viscoelastic and cohesive properties of the resultant mucus. It was speculated that the thread abets localization of mucus in an aqueous environment and promotes adhesion of mucus to surfaces such as the fish integument. As judged by immunostaining in situ, GTCs, as well as several cell-types in the epidermis, contain keratin-like components. The role of biopolymers on the properties of teleost and mammalian mucus is discussed.