Automatic Compilation of MATLAB Programs for Synergistic Execution on Heterogeneous Processors

MATLAB is an array language, initially popular for rapid prototyping, but is now being increasingly used to develop production code for numerical and scientific applications. Typical MATLAB programs have abundant data parallelism. These programs also have control flow dominated scalar regions that have an impact on the program's execution time. Today's computer systems have tremendous computing power in the form of traditional CPU cores and throughput oriented accelerators such as graphics processing units(GPUs). Thus, an approach that maps the control flow dominated regions to the CPU and the data parallel regions to the GPU can significantly improve program performance. In this paper, we present the design and implementation of MEGHA, a compiler that automatically compiles MATLAB programs to enable synergistic execution on heterogeneous processors. Our solution is fully automated and does not require programmer input for identifying data parallel regions. We propose a set of compiler optimizations tailored for MATLAB. Our compiler identifies data parallel regions of the program and composes them into kernels. The problem of combining statements into kernels is formulated as a constrained graph clustering problem. Heuristics are presented to map identified kernels to either the CPU or GPU so that kernel execution on the CPU and the GPU happens synergistically and the amount of data transfer needed is minimized. In order to ensure required data movement for dependencies across basic blocks, we propose a data flow analysis and edge splitting strategy. Thus our compiler automatically handles composition of kernels, mapping of kernels to CPU and GPU, scheduling and insertion of required data transfer. The proposed compiler was implemented and experimental evaluation using a set of MATLAB benchmarks shows that our approach achieves a geometric mean speedup of 19.8X for data parallel benchmarks over native execution of MATLAB.

Automatic compilation of MATLAB programs for synergistic execution on heterogeneous processors

MATLAB is an array language, initially popular for rapid prototyping, but is now being increasingly used to develop production code for numerical and scientific applications. Typical MATLAB programs have abundant data parallelism. These programs also have control flow dominated scalar regions that have an impact on the program's execution time. Today's computer systems have tremendous computing power in the form of traditional CPU cores and throughput oriented accelerators such as graphics processing units(GPUs). Thus, an approach that maps the control flow dominated regions to the CPU and the data parallel regions to the GPU can significantly improve program performance. In this paper, we present the design and implementation of MEGHA, a compiler that automatically compiles MATLAB programs to enable synergistic execution on heterogeneous processors. Our solution is fully automated and does not require programmer input for identifying data parallel regions. We propose a set of compiler optimizations tailored for MATLAB. Our compiler identifies data parallel regions of the program and composes them into kernels. The problem of combining statements into kernels is formulated as a constrained graph clustering problem. Heuristics are presented to map identified kernels to either the CPU or GPU so that kernel execution on the CPU and the GPU happens synergistically and the amount of data transfer needed is minimized. In order to ensure required data movement for dependencies across basic blocks, we propose a data flow analysis and edge splitting strategy. Thus our compiler automatically handles composition of kernels, mapping of kernels to CPU and GPU, scheduling and insertion of required data transfer. The proposed compiler was implemented and experimental evaluation using a set of MATLAB benchmarks shows that our approach achieves a geometric mean speedup of 19.8X for data parallel benchmarks over native execution of MATLAB.

Toward a Scalable Working Set Size Estimation Method and Its Application for Chip Multiprocessors

It is essential to accurately estimate the working set size (WSS) of an application for various optimizations such as to partition cache among virtual machines or reduce leakage power dissipated in an over-allocated cache by switching it OFF. However, the state-of-the-art heuristics such as average memory access latency (AMAL) or cache miss ratio (CMR) are poorly correlated to the WSS of an application due to 1) over-sized caches and 2) their dispersed nature. Past studies focus on estimating WSS of an application executing on a uniprocessor platform. Estimating the same for a chip multiprocessor (CMP) with a large dispersed cache is challenging due to the presence of concurrently executing threads/processes. Hence, we propose a scalable, highly accurate method to estimate WSS of an application. We call this method ``tagged WSS (TWSS)'' estimation method. We demonstrate the use of TWSS to switch-OFF the over-allocated cache ways in Static and Dynamic NonUniform Cache Architectures (SNUCA, DNUCA) on a tiled CMP. In our implementation of adaptable way SNUCA and DNUCA caches, decision of altering associativity is taken by each L2 controller. Hence, this approach scales better with the number of cores present on a CMP. It gives overall (geometric mean) 26% and 19% higher energy-delay product savings compared to AMAL and CMR heuristics on SNUCA, respectively.

Notch sensitivity of circular rings subjected to impact loading

'Notch-sensitive regions' have been observed during a series of experimental investigations into the dynamic plastic behaviour and failure of thin-walled metallic radially notched circular rings with are-shaped supports subjected to concentrated impact loads. The experimental results show that the exterior notches at some regions have no effect on the deformation of the rings, but do have effect at the remaining regions. The notch-sensitive region is theoretically determined by using the equivalent structures technique; fairly good agreement has been reached between the simple theory and the experimental results. Both dimensional and theoretical analyses prove that whether a plastic hinge formed or not at the notched section does not depend on the mean radius of the ring and the input kinetic energy. It depends on the weak coefficient of the notched section and the angle of the support. Generally speaking, there are mainly three failure modes for a notched circular ring with are-shaped support under impact loading: Mode I, large inelastic deformation when the notch is outside the sensitive region, in this case the ring deforms as a normal one; Mode II, large inelastic deformation only at some part of the ring and tearing occurred at the notched sections; Mode III, large inelastic deformation and total rupture occurred at the notched sections. It is believed that the present study could assist the understanding of the dynamic behaviour and failure of other kinds of nonstraight components with macroscopic imperfections under impulsive loading.

The age and growth of the tongue sole Cynoglossus canariensis (Steind 1882)

The geometric mean regression equation for the weight; length relationship of Cynoglossus canariensis was W = 0.0025 L super(3.1770). The Von Bertalanffy constants Woo, Loo, K, and to were 507.5852 g, 47.3683 cm, 0.3333 and 0.1397 for males and 839.0753 g, 54.4720 cm, 0.3062 and 0.1737 for females. Total mortality coefficient Z ranged from 0.6482 and 0.8021

Transceiver designs and analysis for LTI, LTV and broadcast channels - new matrix decompositions and majorization theory

Signal processing techniques play important roles in the design of digital communication systems. These include information manipulation, transmitter signal processing, channel estimation, channel equalization and receiver signal processing. By interacting with communication theory and system implementing technologies, signal processing specialists develop efficient schemes for various communication problems by wisely exploiting various mathematical tools such as analysis, probability theory, matrix theory, optimization theory, and many others. In recent years, researchers realized that multiple-input multiple-output (MIMO) channel models are applicable to a wide range of different physical communications channels. Using the elegant matrix-vector notations, many MIMO transceiver (including the precoder and equalizer) design problems can be solved by matrix and optimization theory. Furthermore, the researchers showed that the majorization theory and matrix decompositions, such as singular value decomposition (SVD), geometric mean decomposition (GMD) and generalized triangular decomposition (GTD), provide unified frameworks for solving many of the point-to-point MIMO transceiver design problems.

In this thesis, we consider the transceiver design problems for linear time invariant (LTI) flat MIMO channels, linear time-varying narrowband MIMO channels, flat MIMO broadcast channels, and doubly selective scalar channels. Additionally, the channel estimation problem is also considered. The main contributions of this dissertation are the development of new matrix decompositions, and the uses of the matrix decompositions and majorization theory toward the practical transmit-receive scheme designs for transceiver optimization problems. Elegant solutions are obtained, novel transceiver structures are developed, ingenious algorithms are proposed, and performance analyses are derived.

The first part of the thesis focuses on transceiver design with LTI flat MIMO channels. We propose a novel matrix decomposition which decomposes a complex matrix as a product of several sets of semi-unitary matrices and upper triangular matrices in an iterative manner. The complexity of the new decomposition, generalized geometric mean decomposition (GGMD), is always less than or equal to that of geometric mean decomposition (GMD). The optimal GGMD parameters which yield the minimal complexity are derived. Based on the channel state information (CSI) at both the transmitter (CSIT) and receiver (CSIR), GGMD is used to design a butterfly structured decision feedback equalizer (DFE) MIMO transceiver which achieves the minimum average mean square error (MSE) under the total transmit power constraint. A novel iterative receiving detection algorithm for the specific receiver is also proposed. For the application to cyclic prefix (CP) systems in which the SVD of the equivalent channel matrix can be easily computed, the proposed GGMD transceiver has K/log_2(K) times complexity advantage over the GMD transceiver, where K is the number of data symbols per data block and is a power of 2. The performance analysis shows that the GGMD DFE transceiver can convert a MIMO channel into a set of parallel subchannels with the same bias and signal to interference plus noise ratios (SINRs). Hence, the average bit rate error (BER) is automatically minimized without the need for bit allocation. Moreover, the proposed transceiver can achieve the channel capacity simply by applying independent scalar Gaussian codes of the same rate at subchannels.

In the second part of the thesis, we focus on MIMO transceiver design for slowly time-varying MIMO channels with zero-forcing or MMSE criterion. Even though the GGMD/GMD DFE transceivers work for slowly time-varying MIMO channels by exploiting the instantaneous CSI at both ends, their performance is by no means optimal since the temporal diversity of the time-varying channels is not exploited. Based on the GTD, we develop space-time GTD (ST-GTD) for the decomposition of linear time-varying flat MIMO channels. Under the assumption that CSIT, CSIR and channel prediction are available, by using the proposed ST-GTD, we develop space-time geometric mean decomposition (ST-GMD) DFE transceivers under the zero-forcing or MMSE criterion. Under perfect channel prediction, the new system minimizes both the average MSE at the detector in each space-time (ST) block (which consists of several coherence blocks), and the average per ST-block BER in the moderate high SNR region. Moreover, the ST-GMD DFE transceiver designed under an MMSE criterion maximizes Gaussian mutual information over the equivalent channel seen by each ST-block. In general, the newly proposed transceivers perform better than the GGMD-based systems since the super-imposed temporal precoder is able to exploit the temporal diversity of time-varying channels. For practical applications, a novel ST-GTD based system which does not require channel prediction but shares the same asymptotic BER performance with the ST-GMD DFE transceiver is also proposed.

The third part of the thesis considers two quality of service (QoS) transceiver design problems for flat MIMO broadcast channels. The first one is the power minimization problem (min-power) with a total bitrate constraint and per-stream BER constraints. The second problem is the rate maximization problem (max-rate) with a total transmit power constraint and per-stream BER constraints. Exploiting a particular class of joint triangularization (JT), we are able to jointly optimize the bit allocation and the broadcast DFE transceiver for the min-power and max-rate problems. The resulting optimal designs are called the minimum power JT broadcast DFE transceiver (MPJT) and maximum rate JT broadcast DFE transceiver (MRJT), respectively. In addition to the optimal designs, two suboptimal designs based on QR decomposition are proposed. They are realizable for arbitrary number of users.

Finally, we investigate the design of a discrete Fourier transform (DFT) modulated filterbank transceiver (DFT-FBT) with LTV scalar channels. For both cases with known LTV channels and unknown wide sense stationary uncorrelated scattering (WSSUS) statistical channels, we show how to optimize the transmitting and receiving prototypes of a DFT-FBT such that the SINR at the receiver is maximized. Also, a novel pilot-aided subspace channel estimation algorithm is proposed for the orthogonal frequency division multiplexing (OFDM) systems with quasi-stationary multi-path Rayleigh fading channels. Using the concept of a difference co-array, the new technique can construct M^2 co-pilots from M physical pilot tones with alternating pilot placement. Subspace methods, such as MUSIC and ESPRIT, can be used to estimate the multipath delays and the number of identifiable paths is up to O(M^2), theoretically. With the delay information, a MMSE estimator for frequency response is derived. It is shown through simulations that the proposed method outperforms the conventional subspace channel estimator when the number of multipaths is greater than or equal to the number of physical pilots minus one.

Episodic variations in stream water chemistry associated with acid rainfall and run-off and the effect on aquatic ecosystem, with particular reference to fish populations in N.W. England

This is the episodic variations in stream water chemistry associated with acid rainfall and run-off and the effect on aquatic ecosystems, with particular reference to fish populations in North West England produced by the North West Water Authority in 1985. This report looks at the biological, physical and chemical information collected over a five year period from over 100 sites on upland streams in the North West Region of which drained rocks of low buffering capacity. In both Lake District and South Pennine sites striking differences were found between the composition of invertebrate communities inhabiting acid-stressed and less acid-stressed streams. Electric fishing surveys showed that acidic streams (geometric mean pH <5.5) generally had abnormally low densities of salmonids ( < 0 .2m2) and that 0+ fish were very few or absent. The latter indicates recruitment failure. Salmon were more sensitive than trout to low pH.

Friction of sliding surfaces carrying adsorbed lubricant layers

In concentrated contacts the behaviour of lubricants is much modified by the high local pressures: changes can arise both from molecular ordering within the very thin film lubricant layers present at the interface as well as from the deposition on the component surfaces of more solid-like polymeric boundary layers. These 'third bodies' separating the solid surfaces may have rheological or mechanical properties very different from those observed in the bulk. Classical elasto-hydrodynamic theory considers the entrapped lubricant to exhibit a piezo-viscous behaviour while the conventional picture of more solid boundary lubricant layers views their shear strength r as being linearly dependent on local pressure p, so that T = TO + ap where TO and a are constants. If TO is relatively small, then the coefficient of friction \i = T Ip ~ a and so Amonton's laws are recovered. However, the properties of adsorbed or deposited surface films, or indeed other third bodies such as debris layers, may be more complex than this. A preliminary study has looked quantitatively at the influence of the pressure dependence of the shear strength of any surface layer on the overall friction coefficient of a contact which is made up of an array of asperities whose height varies in a Gaussian manner. Individual contact points may be elastic or plastic. The analysis results in plots of coefficient of friction versus the service or load parameter PIH&NRa where P is the nominal pressure on the contact, HS the hardness of the deforming surface, N the asperity density, R the mean radius of curvature of the asperities, and a is the standard deviation of their height distribution. In principle, any variation oft withp can be incorporated into the model; however, in this initial study we have used data on colloidal suspensions from the group at the Ecole Centrale de Lyon as well as examining the effect of functional relationships of somewhat greater complexity than a simple linear form. Results of the analysis indicate that variations in fj. are possible as the load is varied which depend on the statistical spread of behaviour at individual asperity contacts. The value of this analysis is that it attempts to combine the behaviour of films on the molecular scale with the topography of real engineering surfaces and so give an indication of the effects at the full-size or macro-scale that can be achieved by chemical or molecular surface engineering.

On further improvement of the techinque of evaluating summer institutes

Data from an earlier study are reanalyzed to improve upon the evaluation measures of summer institutes. The Summer Institute Efficiency Index is improved by using weighted geometric mean using different ranks as weights for the several dimensions. The coverage utility index is improved by taking the ratings given by all the participants.

Updated correlation between aircraft smoke number and black carbon concentration

Aircraft emissions of black carbon (BC) contribute to anthropogenic climate forcing and degrade air quality. The smoke number (SN) is the current regulatory measure of aircraft particulate matter emissions and quantifies exhaust plume visibility. Several correlations between SN and the exhaust mass concentration of BC (CBC) have been developed, based on measurements relevant to older aircraft engines. These form the basis of the current standard method used to estimate aircraft BC emissions (First Order Approximation version 3 [FOA3]) for the purposes of environmental impact analyses. In this study, BC with a geometric mean diameter (GMD) of 20, 30, and 60 nm and filter diameters of 19 and 35 mm are used to investigate the effect of particle size and sampling variability on SN measurements. For BC with 20 and 30 nm GMD, corresponding to BC emitted by modern aircraft engines, a smaller SN results from a given CBC than is the case for BC with 60 nm GMD, which is more typical of older engines. An updated correlation between CBC and SNthat accounts for typical size of BC emitted by modern aircraft is proposed. An uncertainty of ±25% accounts for variation in GMD in the range 20-30 nm and for the range of filter diameters. The SN-CBC correlation currently used in FOA3 underestimates by a factor of 2.5-3 for SN <15, implying that current estimates of aircraft BC emissions derived from SN are underestimated by the same factor. Copyright © American Association for Aerosol Research.

Defect Cluster-Induced X-Ray Diffuse Scattering in GaN Films Grown by MOCVD

High-resolution X-ray diffraction has been employed to investigate the diffuse scattering in a (0001) oriented GaN epitaxial film grown on sapphire substrate. The analysis reveals that defect clusters are present in GaN films and their concentration increases as the density of threading dislocations increases. Meanwhile, the mean radius of these defect clusters shows a reverse tendency. This result is explained by the effect of clusters preferentially forming around dislocations, which act as effective sinks for the segregation of point defects. The electric mobility is found to decrease as the cluster concentration increases.

Tunable ordered droplets induced by convection in phase-separating P2VP/PS blend film

Ordered hexagonal droplets patterns in phase-separating polymeric blend films of polystyrene and poly(2-vinylpyridine) (PS/PVP) formed due to the convection effect by solvent evaporation. The influences of PS molecular weight, solvent evaporation rate, and the weight ratio of PS to PVP on the PVP-rich domains pattern formation and distributions were investigated by atomic force microscope (AFM). Only in an appropriate range of molecular weight of PS, can the ordered pattern form. Too low or too high molecular weight of PS led no ordered pattern due to the viscosity effects. The increase of solvent evaporation rate decreased the mean radius of the PVP-rich domains and the intervals between the centers of the domains due to the enhancement of the viscosity on the top layer of the fluid film. The increase of the weight ratio of PS to PVP decreased mean radius of the PVP-rich domains whereas the intervals between the centers of droplets remained constant. Therefore, the size and the distributions of ordered patterns can be tuned by the polymer molecular weight, the weight ratio of the two components and the solvent evaporation rate.

Urinary naphthalene and phenanthrene as biomarkers of occupational exposure to polycyclic aromatic hydrocarbons.

OBJECTIVES: The study investigated the utility of unmetabolised naphthalene (Nap) and phenanthrene (Phe) in urine as surrogates for exposures to mixtures of polycyclic aromatic hydrocarbons (PAHs). METHODS: The report included workers exposed to diesel exhausts (low PAH exposure level, n = 39) as well as those exposed to emissions from asphalt (medium PAH exposure level, n = 26) and coke ovens (high PAH exposure level, n = 28). Levels of Nap and Phe were measured in urine from each subject using head space-solid phase microextraction and gas chromatography-mass spectrometry. Published levels of airborne Nap, Phe and other PAHs in the coke-producing and aluminium industries were also investigated. RESULTS: In post-shift urine, the highest estimated geometric mean concentrations of Nap and Phe were observed in coke-oven workers (Nap: 2490 ng/l; Phe: 975 ng/l), followed by asphalt workers (Nap: 71.5 ng/l; Phe: 54.3 ng/l), and by diesel-exposed workers (Nap: 17.7 ng/l; Phe: 3.60 ng/l). After subtracting logged background levels of Nap and Phe from the logged post-shift levels of these PAHs in urine, the resulting values (referred to as ln(adjNap) and ln(adjPhe), respectively) were significantly correlated in each group of workers (0.71 < or = Pearson r < or = 0.89), suggesting a common exposure source in each case. Surprisingly, multiple linear regression analysis of ln(adjNap) on ln(adjPhe) showed no significant effect of the source of exposure (coke ovens, asphalt and diesel exhaust) and further suggested that the ratio of urinary Nap/Phe (in natural scale) decreased with increasing exposure levels. These results were corroborated with published data for airborne Nap and Phe in the coke-producing and aluminium industries. The published air measurements also indicated that Nap and Phe levels were proportional to the levels of all combined PAHs in those industries. CONCLUSION: Levels of Nap and Phe in urine reflect airborne exposures to these compounds and are promising surrogates for occupational exposures to PAH mixtures.

Equal IgG antibody response to pneumococcal vaccination in all stages of human immunodeficiency virus disease.

To evaluate the immunogenicity and safety of a 23-valent pneumococcal vaccine in human immunodeficiency virus (HIV)-seropositive patients, 80 men and 18 women received 1 dose of the vaccine (Pneumo 23; Pasteur Mérieux MSD, Brussels). The total IgG antibody response against all 23 Streptococcus pneumoniae capsular antigens was measured. Antibody levels were expressed in arbitrary units per microliter, referring to a standard curve. Geometric mean titers of the total IgG capsular antibodies on the day of vaccination and 30-45 days later were compared. The ratios of titers after and before vaccination in patients with > 500, 200-500, and < 200 CD4 lymphocytes/microL were 10, 10, and 12.6, respectively. Nonresponse (ratio < 4) occurred in 17% of patients and was unrelated to CD4 cell count. The vaccine was well tolerated; no serious side effects occurred. In 83% of the patients with HIV infection, the total antipneumococcal IgG level was higher after vaccination.

Daily intake of antioxidants in relation to survival among adult patients diagnosed with malignant glioma.

BACKGROUND: Malignant glioma is a rare cancer with poor survival. The influence of diet and antioxidant intake on glioma survival is not well understood. The current study examines the association between antioxidant intake and survival after glioma diagnosis. METHODS: Adult patients diagnosed with malignant glioma during 1991-1994 and 1997-2001 were enrolled in a population-based study. Diagnosis was confirmed by review of pathology specimens. A modified food-frequency questionnaire interview was completed by each glioma patient or a designated proxy. Intake of each food item was converted to grams consumed/day. From this nutrient database, 16 antioxidants, calcium, a total antioxidant index and 3 macronutrients were available for survival analysis. Cox regression estimated mortality hazard ratios associated with each nutrient and the antioxidant index adjusting for potential confounders. Nutrient values were categorized into tertiles. Models were stratified by histology (Grades II, III, and IV) and conducted for all (including proxy) subjects and for a subset of self-reported subjects. RESULTS: Geometric mean values for 11 fat-soluble and 6 water-soluble individual antioxidants, antioxidant index and 3 macronutrients were virtually the same when comparing all cases (n=748) to self-reported cases only (n=450). For patients diagnosed with Grade II and Grade III histology, moderate (915.8-2118.3 mcg) intake of fat-soluble lycopene was associated with poorer survival when compared to low intake (0.0-914.8 mcg), for self-reported cases only. High intake of vitamin E and moderate/high intake of secoisolariciresinol among Grade III patients indicated greater survival for all cases. In Grade IV patients, moderate/high intake of cryptoxanthin and high intake of secoisolariciresinol were associated with poorer survival among all cases. Among Grade II patients, moderate intake of water-soluble folate was associated with greater survival for all cases; high intake of vitamin C and genistein and the highest level of the antioxidant index were associated with poorer survival for all cases. CONCLUSIONS: The associations observed in our study suggest that the influence of some antioxidants on survival following a diagnosis of malignant glioma are inconsistent and vary by histology group. Further research in a large sample of glioma patients is needed to confirm/refute our results.