Energy aware software for many-core systems

Many-core systems provide a great potential in application performance with the massively parallel structure. Such systems are currently being integrated into most parts of daily life from high-end server farms to desktop systems, laptops and mobile devices. Yet, these systems are facing increasing challenges such as high temperature causing physical damage, high electrical bills both for servers and individual users, unpleasant noise levels due to active cooling and unrealistic battery drainage in mobile devices; factors caused directly by poor energy efficiency. Power management has traditionally been an area of research providing hardware solutions or runtime power management in the operating system in form of frequency governors. Energy awareness in application software is currently non-existent. This means that applications are not involved in the power management decisions, nor does any interface between the applications and the runtime system to provide such facilities exist. Power management in the operating system is therefore performed purely based on indirect implications of software execution, usually referred to as the workload. It often results in over-allocation of resources, hence power waste. This thesis discusses power management strategies in many-core systems in the form of increasing application software awareness of energy efficiency. The presented approach allows meta-data descriptions in the applications and is manifested in two design recommendations: 1) Energy-aware mapping 2) Energy-aware execution which allow the applications to directly influence the power management decisions. The recommendations eliminate over-allocation of resources and increase the energy efficiency of the computing system. Both recommendations are fully supported in a provided interface in combination with a novel power management runtime system called Bricktop. The work presented in this thesis allows both new- and legacy software to execute with the most energy efficient mapping on a many-core CPU and with the most energy efficient performance level. A set of case study examples demonstrate realworld energy savings in a wide range of applications without performance degradation.

The effect of social condition and diet on the frequency of male-male agonistic displays in the field cricket, gryllus bimaculatus /

The frequency and type of agonistic displays involved in male-male encounters should be significantly influenced by the presence of females. Discrete agonistic displays vary in energy expenditure and risk, and therefore should be dependent on available resources. The influence of live females and the scent of females, on the frequency of male agonistic displays was observed in a laboratory terrarium using the field cricket Gryllus bimaculatus. The effect of energy constraints on display frequency was also determined. Half the males were fed a diet high in protein and fet; the other males were fed a lower quality diet, for a 7-11 day period. The frequency of five individual displays and mating frequency were recorded using an Event Recorder and notebook. Each group of males was presented with three experimental conditions, over three days, involving the presence or absence of live females and female scent. The presence of females elicited an increase in all displays except antennation; female scent increased the frequency of antennations, mandible flares and grapples, but to a lesser extent than did live females. The frequency of grapples significantly increased for males fed the high quality diet; however diet did not influence the other displays. The combined influence of diet and condition was significant for mandible flare only. Mating frequency was not influenced by diet. However, the frequency ofthe displays were positively correlated with mating frequency for high quality fed males. Escalated displays involving high costs, such as grapple and mandible flare, increased in frequency when the benefits of winning contests were high in G.bimaculatus. Escalation to grapple behaviour was less evident for males fed the lower quality diet as this imposed energy constraints on high cost displays.

Far infrared reflectance and optical properties of organic superconductor (TMTSF)2ClO4 /

Polarized reflectance measurements of the quasi I-D charge-transfer salt (TMTSFh CI04 were carried out using a Martin-Puplett-type polarizing interferometer and a 3He refrigerator cryostat, at several temperatures between 0.45 K and 26 K, in the far infrared, in the 10 to 70 cm- 1 frequency range. Bis-tetramethyl-tetraselena-fulvalene perchlorate crystals, grown electrochemically and supplied by K. Behnia, of dimensions 2 to 4 by 0.4 by 0.2 mm, were assembled on a flat surface to form a mosaic of 1.5 by 3 mm. The needle shaped crystals were positioned parallel to each other along their long axis, which is the stacking direction of the planar TMTSF cations, exposing the ab plane face (parallel to which the sheets of CI04 anions are positioned). Reflectance measurements were performed with radiation polarized along the stacking direction in the sample. Measurements were carried out following either a fast (15-20 K per minute) or slow (0.1 K per minute) cooling of the sample. Slow cooling permits the anions to order near 24 K, and the sample is expected to be superconducting below 1.2 K, while fast cooling yields an insulating state at low temperatures. Upon the slow cooling the reflectance shows dependence with temperature and exhibits the 28 cm- 1 feature reported previously [1]. Thermoreflectance for both the 'slow' and 'fast' cooling of the sample calculated relative to the 26 K reflectance data indicates that the reflectance is temperature dependent, for the slow cooling case only. A low frequency edge in the absolute reflectance is assigned an electronic origin given its strong temperature dependence in the relaxed state. We attribute the peak in the absolute reflectance near 30 cm-1 to a phonon coupled to the electronic background. Both the low frequency edge and the 30 cm-1 feature are noted te shift towards higher frequcncy, upon cntering the superconducting state, by an amount of the order of the expected superconducting energy gap. Kramers-Kronig analysis was carried out to determine the optical conductivity for the slowly cooled sample from the measured reflectance. In order to do so the low frequency data was extrapolated to zero frequency using a Hagen-Ru bens behaviour, and the high frequency data was extended with the data of Cao et al. [2], and Kikuchi et al. [3]. The real part of the optical conductivity exhibits an asymmetric peak at 35 cm-1, and its background at lower frequencies seems to be losing spectral weight with lowering of the temperature, leading us to presume that a narrow peak is forming at even lower frequencies.

Optical properties of organic superconductor K-(BETS)2FeBr4 /

K-(BETS)2FeBr4 is a quasi-2D charge transfer organic metal with interesting electronic and magnetic properties. It undergoes a transition to an antiferromagnetic (AF) state at ambient pressure at the Neel temperature (T^^) = 2.5 K, as well as to a superconducting (SC) state at 1.1 K [1]. The temperature dependence of the electrical resistivity shows a small decrease at T;v indicating the resistivity drops as a result of the onset of the ordering of Fe'*''" spins. A sharp drop in the resistivity at 1.1 K is due to its superconducting transition. The temperature dependence of the susceptibility indicates an antiferromagnetic spin structure with the easy axis parallel to the a-axis. The specific heat at zero-field shows a large peak at about 2.4 K, which corresponds to the antiferromagnetic transition temperature (Tat) and no anomaly is observed around the superconducting transition temperature (1.1 K) demonstrating that the magnetically ordered state is not destroyed by the appearance of another phase transition (the superconducting transition) in the 7r-electron layers [1], [2]. This work presents an investigation of how the low frequency electromagnetic response is affected by the antiferromagnetic and superconducting states, as well as the onset of strong correlation. The location of the easy axis of three samples was determined and polarized thermal reflectance measurements of these «-(BETS)2FeBr4 samples oriented with their vertical axis along the a- and c axes were then carried out using a *He refrigerator cryostat and a Martin-Puplett type polarizing interferometer at various temperatures (T = 0.5 K, 1.4 K. 1.9 K, 2.8 K) above and below the superconducting state and/or antiferromagnetic state. Comparison of the SC state to the normal state along the o- and c-axes indicates a rising thermal reflectance at low frequencies (below 10 cm"' ) which may be a manifestation of the superconducting energy gap. A dip-Hke feature is detected at low frequencies (below 15 cm"') in the thermal reflectance plots which probe the antiferromagnetic state along the two axes, and may be due to the opening of a gap in the excitation spectrum as a result of the antiferromagnetism. In another set of experiments, thermal reflectance measurements carried out along the a- and c-axes at higher temperatures (10 K-80 K) show that the reflectivity decreases with increasing temperature to 60 K (the coherence temperature) above which it increases again. Comparison of the thermal reflectance plots along the a- and c-axes at higher temperatures reveals an anisotropy between these two axes. The Hagen-Rubens thermal reflectance plots corresponding to an average over the ac-plane were calculated using experimental hterature resistivity values. Comparison of the Hagen-Rubens plots with the experimental thermal reflectance along the a- and c-axes indicates that both exhibit the general trend of a decrease in thermal reflectance with increasing frequency, however the calculated Hagen-Rubens thermal reflectance at different temperatures is much lower than the experimental curves.

Low frequency Raman scattering in amorphous materials: fused quartz, "pyrex" boro-silicate glass and soda-lime silicate glass

Raman scattering in the region 20 to 100 cm -1 for fused quartz, "pyrex" boro-silicate glass, and soft soda-lime silicate glass was investigated. The Raman spectra for the fused quartz and the pyrex glass were obtained at room temperature using the 488 nm exciting line of a Coherent Radiation argon-ion laser at powers up to 550 mW. For the soft soda-lime glass the 514.5 nm exciting line at powers up to 660 mW was used because of a weak fluorescence which masked the Stokes Raman spectrum. In addition it is demonstrated that the low-frequency Raman coupling constant can be described by a model proposed by Martin and Brenig (MB). By fitting the predicted spectra based on the model with a Gaussian, Poisson, and Lorentzian forms of the correlation function, the structural correlation radius (SCR) was determined for each glass. It was found that to achieve the best possible fit· from each of the three correlation functions a value of the SCR between 0.80 and 0.90 nm was required for both quartz and pyrex glass but for the soft soda-lime silicate glass the required value of the SCR. was between 0.50 and 0.60 nm .. Our results support the claim of Malinovsky and Sokolov (1986) that the MB model based on a Poisson correlation function provides a universal fit to the experimental VH (vertical and horizontal polarizations) spectrum for any glass regardless of its chemical composition. The only deficiency of the MB model is its failure to fit the experimental depolarization spectra.

Sediment genotoxicity and its relationship to the frequency of chironomid labial plate deformities

Sediment samples were taken from seven locations in the WeIland River in December 1986 and April 1987. The DMSO extracts of these sediment samples showed a significant (pfrequency of chironomid labial plate deformities (over 1000 individual specimens were observed) . The samples from station D-l showed the highest frequency of chironomid labial plate deformities (10 . 9% ± 3.2%), while samples from the upstream control (station A) displayed t,he lowest frequency of deformities (3.8% ± 1.3%). All samples were coded to avoid unconscious biases . The results of the genotoxicity study indicated that station D-l in the WeIland River was contaminated with genotoxic materials. The genotoxic materials may have induced the observed increased frequency in chironomid labial plate deformities . Samples from stations C and D-l, located in a downstream portion of the river bounded by an industrialized area were slightly toxic according to -the alkaline phosphatase inhib_ii~ion component of the 50S chromotest analyses. The toxicity of these samples was only evident once they had been activated by the 59 (liver extract) mixture.

Frequency-dependent selection at the amylase locus in Drosophila melanogaster

A. strain of Drosophila melanog-aster deficient in null amylase activity (Amylase ) was isolated from a wild null population of flies. The survivorship of Amylase homozygous flies is very low when the principal dietary carbohydrate source is starch. However, the survivorship of the null Amylase genotype is comparable to the wild type when the dietary starch is replaced by glucose. In addition, the null viability of the amylase-producing and Amylase strains is comparable v and very lm<] f on a medium with no carbohydrates . Furthermore, amylase-producing genotypes were shovm to excrete enzymatically active amylase protein into the food medium. The excreted amylase causes the external breakdown of dietary starch to sugar. These results led to the following null prediction: the viability of the A.mvlase genotype (fed on a starch rich diet) might increase in the presence of individuals which were amylase-producing. It was shown experimentally that such an increase in viability did in fact occur and that this increase v\Tas proportional to the number of mnylase..::producing fli.es present. These results provide a unique example of a non-"competi ti ve inter-genotype interaction, and one where the underlying physio~ logical and biochemical mechanism has been fully understood.

Improving Short DNA Sequence Alignment with Parallel Computing

Variations in different types of genomes have been found to be responsible for a large degree of physical diversity such as appearance and susceptibility to disease. Identification of genomic variations is difficult and can be facilitated through computational analysis of DNA sequences. Newly available technologies are able to sequence billions of DNA base pairs relatively quickly. These sequences can be used to identify variations within their specific genome but must be mapped to a reference sequence first. In order to align these sequences to a reference sequence, we require mapping algorithms that make use of approximate string matching and string indexing methods. To date, few mapping algorithms have been tailored to handle the massive amounts of output generated by newly available sequencing technologies. In otrder to handle this large amount of data, we modified the popular mapping software BWA to run in parallel using OpenMPI. Parallel BWA matches the efficiency of multithreaded BWA functions while providing efficient parallelism for BWA functions that do not currently support multithreading. Parallel BWA shows significant wall time speedup in comparison to multithreaded BWA on high-performance computing clusters, and will thus facilitate the analysis of genome sequencing data.

Volume-Synchronized Probability of Informed Trading (VPIN), Market Volatility, and High-Frequency Liquidity

We assess the predictive ability of three VPIN metrics on the basis of two highly volatile market events of China, and examine the association between VPIN and toxic-induced volatility through conditional probability analysis and multiple regression. We examine the dynamic relationship on VPIN and high-frequency liquidity using Vector Auto-Regression models, Granger Causality tests, and impulse response analysis. Our results suggest that Bulk Volume VPIN has the best risk-warning effect among major VPIN metrics. VPIN has a positive association with market volatility induced by toxic information flow. Most importantly, we document a positive feedback effect between VPIN and high-frequency liquidity, where a negative liquidity shock boosts up VPIN, which, in turn, leads to further liquidity drain. Our study provides empirical evidence that reflects an intrinsic game between informed traders and market makers when facing toxic information in the high-frequency trading world.

Force potentiation as a modulator of contractile performance: Implications for control of skeletal muscle force and energetics of work

This thesis investigated the modulation of dynamic contractile function and energetics of work by posttetanic potentiation (PTP). Mechanical experiments were conducted in vitro using software-controlled protocols to stimulate/determine contractile function during ramp shortening, and muscles were frozen during parallel incubations for biochemical analysis. The central feature of this research was the comparison of fast hindlimb muscles from wildtype and skeletal myosin light chain kinase knockout (skMLCK-/-) mice that does not express the primary mechanism for PTP: myosin regulatory light chain (RLC) phosphorylation. In contrast to smooth/cardiac muscles where RLC phosphorylation is indispensable, its precise physiological role in skeletal muscle is unclear. It was initially determined that tetanic potentiation was shortening speed dependent, and this sensitivity of the PTP mechanism to muscle shortening extended the stimulation frequency domain over which PTP was manifest. Thus, the physiological utility of RLC phosphorylation to augment contractile function in vivo may be more extensive than previously considered. Subsequent experiments studied the contraction-type dependence for PTP and demonstrated that the enhancement of contractile function was dependent on force level. Surprisingly, in the absence of RLC phosphorylation, skMLCK-/- muscles exhibited significant concentric PTP; consequently, up to ~50% of the dynamic PTP response in wildtype muscle may be attributed to an alternate mechanism. When the interaction of PTP and the catchlike property (CLP) was examined, we determined that unlike the acute augmentation of peak force by the CLP, RLC phosphorylation produced a longer-lasting enhancement of force and work in the potentiated state. Nevertheless, despite the apparent interference between these mechanisms, both offer physiological utility and may be complementary in achieving optimal contractile function in vivo. Finally, when the energetic implications of PTP were explored, we determined that during a brief period of repetitive concentric activation, total work performed was ~60% greater in wildtype vs. skMLCK-/- muscles but there was no genotype difference in High-Energy Phosphate Consumption or Economy (i.e. HEPC: work). In summary, this thesis provides novel insight into the modulatory effects of PTP and RLC phosphorylation, and through the observation of alternative mechanisms for PTP we further develop our understanding of the history-dependence of fast skeletal muscle function.

Correcting the Errors : A Note on Volatility Forecast Evaluation Based on High-Frequency Data and Realized Volatilities

This note develops general model-free adjustment procedures for the calculation of unbiased volatility loss functions based on practically feasible realized volatility benchmarks. The procedures, which exploit the recent asymptotic distributional results in Barndorff-Nielsen and Shephard (2002a), are both easy to implement and highly accurate in empirically realistic situations. On properly accounting for the measurement errors in the volatility forecast evaluations reported in Andersen, Bollerslev, Diebold and Labys (2003), the adjustments result in markedly higher estimates for the true degree of return-volatility predictability.

Bundling under the Threat of Parallel Trade

This paper examines the use of bundling by a firm that sells in two national markets and faces entry by parallel traders. The firm can bundle its main product, - a tradable good- with a non-traded service. It chooses between the strategies of pure bundling, mixed bundling and no bundling. The paper shows that in the low-price country the threat of grey trade elicits a move from mixed bundling, or no bundling, towards pure bundling. It encourages a move from pure bundling towards mixes bundling or no bundling in the high-price country. The set of parameter values for which the profit maximizing strategy is not to supply the low price country is smaller than in the absence of bundling. The welfare effects of deterrence of grey trade are not those found in conventional models of price arbitrage. Some consumers in the low-price country may gain from the threat of entry by parallel traders although they pay a higher price. This is due to the fact that the firm responds to the threat of arbitrageurs by increasing the amount of services it puts in the bundle targeted at consumers in that country. Similarly, the threat of parallel trade may affect some consumers in the hight-price country adversely.

Standardization of cytokine flow cytometry assays

Affiliation: Faculté de médecine, Université de Montréal & CANVAC

An Autoregressive Spectral Density Estimator at Frequency Zero for Nonstationarity Tests.

Many unit root and cointegration tests require an estimate of the spectral density function at frequency zero at some process. Kernel estimators based on weighted sums of autocovariances constructed using estimated residuals from an AR(1) regression are commonly used. However, it is known that with substantially correlated errors, the OLS estimate of the AR(1) parameter is severely biased. in this paper, we first show that this least squares bias induces a significant increase in the bias and mean-squared error of kernel-based estimators.