Automatic Data Allocation and Buffer Management for Multi-GPU Machines

Multi-GPU machines are being increasingly used in high-performance computing. Each GPU in such a machine has its own memory and does not share the address space either with the host CPU or other GPUs. Hence, applications utilizing multiple GPUs have to manually allocate and manage data on each GPU. Existing works that propose to automate data allocations for GPUs have limitations and inefficiencies in terms of allocation sizes, exploiting reuse, transfer costs, and scalability. We propose a scalable and fully automatic data allocation and buffer management scheme for affine loop nests on multi-GPU machines. We call it the Bounding-Box-based Memory Manager (BBMM). BBMM can perform at runtime, during standard set operations like union, intersection, and difference, finding subset and superset relations on hyperrectangular regions of array data (bounding boxes). It uses these operations along with some compiler assistance to identify, allocate, and manage data required by applications in terms of disjoint bounding boxes. This allows it to (1) allocate exactly or nearly as much data as is required by computations running on each GPU, (2) efficiently track buffer allocations and hence maximize data reuse across tiles and minimize data transfer overhead, and (3) and as a result, maximize utilization of the combined memory on multi-GPU machines. BBMM can work with any choice of parallelizing transformations, computation placement, and scheduling schemes, whether static or dynamic. Experiments run on a four-GPU machine with various scientific programs showed that BBMM reduces data allocations on each GPU by up to 75% compared to current allocation schemes, yields performance of at least 88% of manually written code, and allows excellent weak scaling.

Pressure-Induced Bond Rearrangement and Reversible Phase Transformation in a Metal-Organic Framework

Pressure-induced phase transformations (PIPTs) occur in a wide range of materials. In general, the bonding characteristics, before and after the PIPT, remain invariant in most materials, and the bond rearrangement is usually irreversible due to the strain induced under pressure. A reversible PIPT associated with a substantial bond rearrangement has been found in a metal-organic framework material, namely tmenH(2)]Er(HCOO)(4)](2) (tmenH(2)(2+) = N,N,N',N'-tetramethylethylenediammonium). The transition is first-order and is accompanied by a unit cell volume change of about 10%. High-pressure single-crystal X-ray diffraction studies reveal the complex bond rearrangement through the transition. The reversible nature of the transition is confirmed by means of independent nanoindentation measurements on single crystals.

Generating Efficient Data Movement Code for Heterogeneous Architectures with Distributed-Memory

Programming for parallel architectures that do not have a shared address space is extremely difficult due to the need for explicit communication between memories of different compute devices. A heterogeneous system with CPUs and multiple GPUs, or a distributed-memory cluster are examples of such systems. Past works that try to automate data movement for distributed-memory architectures can lead to excessive redundant communication. In this paper, we propose an automatic data movement scheme that minimizes the volume of communication between compute devices in heterogeneous and distributed-memory systems. We show that by partitioning data dependences in a particular non-trivial way, one can generate data movement code that results in the minimum volume for a vast majority of cases. The techniques are applicable to any sequence of affine loop nests and works on top of any choice of loop transformations, parallelization, and computation placement. The data movement code generated minimizes the volume of communication for a particular configuration of these. We use a combination of powerful static analyses relying on the polyhedral compiler framework and lightweight runtime routines they generate, to build a source-to-source transformation tool that automatically generates communication code. We demonstrate that the tool is scalable and leads to substantial gains in efficiency. On a heterogeneous system, the communication volume is reduced by a factor of 11X to 83X over state-of-the-art, translating into a mean execution time speedup of 1.53X. On a distributed-memory cluster, our scheme reduces the communication volume by a factor of 1.4X to 63.5X over state-of-the-art, resulting in a mean speedup of 1.55X. In addition, our scheme yields a mean speedup of 2.19X over hand-optimized UPC codes.

Impacts of modern transitions on thermal comfort in vernacular dwellings in warm-humid climate of Sugganahalli (India)

Sugganahalli, a rural vernacular community in a warm-humid region in South India, is under transition towards adopting modern construction practices. Vernacular local building elements like rubble walls and mud roofs are given way to burnt brick walls and reinforced cement concrete (RCC)/tin roofs. Over 60% of Indian population is rural, and implications of such transitions on thermal comfort and energy in buildings are crucial to understand. Vernacular architecture evolves adopting local resources in response to the local climate adopting passive solar designs. This paper investigates the effectiveness of passive solar elements on the indoor thermal comfort by adopting modern climate-responsive design strategies. Dynamic simulation models validated by measured data have also been adopted to determine the impact of the transition from vernacular to modern material-configurations. Age-old traditional design considerations were found to concur with modern understanding into bio-climatic response and climate-responsiveness. Modern transitions were found to increase the average indoor temperatures in excess of 7 degrees C. Such transformations tend to shift the indoor conditions to a psychrometric zone that is likely to require active air-conditioning. Also, the surveyed thermal sensation votes were found to lie outside the extended thermal comfort boundary for hot developing countries provided by Givoni in the bio-climatic chart.

Spinning strings and minimal surfaces in AdS(3) with mixed 3-form fluxes

Motivated by the recent proposal for the S-matrix in AdS(3) x S-3 with mixed three form fluxes, we study classical folded string spinning in AdS(3) with both Ramond and Neveu-Schwarz three form fluxes. We solve the equations of motion of these strings and obtain their dispersion relation to the leading order in the Neveu-Schwarz flux b. We show that dispersion relation for the spinning strings with large spin S acquires a term given by -root lambda/2 pi b(2) log(2) S in addition to the usual root lambda/pi log S term where root lambda is proportional to the square of the radius of AdS(3). Using SO(2, 2) transformations and re-parmetrizations we show that these spinning strings can be related to light like Wilson loops in AdS(3) with Neveu-Schwarz flux b. We observe that the logarithmic divergence in the area of the light like Wilson loop is also deformed by precisely the same coefficient of the b(2) log(2) S term in the dispersion relation of the spinning string. This result indicates that the coefficient of b(2) log(2) S has a property similar to the coefficient of the log S term, known as cusp-anomalous dimension, and can possibly be determined to all orders in the coupling lambda using the recent proposal for the S-matrix.

Exploiting oriented attachment in stabilizing La3+-doped gallium oxide nano-spindles

Retaining the morphology of gallium oxide nanostructures during structural transformations or after doping with lanthanide ions is not facile. Here we report on the sonochemical synthesis of nearly monodisperse similar to 550 nm long nano-spindles of undoped and La-doped alpha-GaOOH. The transformation of as-prepared undoped and La-doped alpha-GaOOH powders into the corresponding undoped and La-doped Ga2O3 phases (alpha and beta) was achieved by carrying out controlled annealing at elevated temperatures under optimized conditions. The formation of gallium oxide nano-spindles is explained by invoking the phenomenon of oriented attachment, as amply supported by electron microscopy. Interestingly, the morphology of the gallium oxide nano-spindles remained conserved even after doping them with more than 1.4 at% of La3+ ions. Such robust structural stability could be attributed to the oriented attachment-type growth observed in the nano-spindles. The as-prepared samples and the corresponding annealed ones were thoroughly characterized by powder X-ray diffraction (PXRD), electron microscopy (SEM, TEM, and STEM-EDS) and X-ray photoelectron spectroscopy (XPS). Finally, photoluminescence from the single-crystalline undoped and La-doped beta-Ga2O3 was explored.

Use of Exocentric and Egocentric Representations in the Concurrent Planning of Sequential Saccades

The concurrent planning of sequential saccades offers a simple model to study the nature of visuomotor transformations since the second saccade vector needs to be remapped to foveate the second target following the first saccade. Remapping is thought to occur through egocentric mechanisms involving an efference copy of the first saccade that is available around the time of its onset. In contrast, an exocentric representation of the second target relative to the first target, if available, can be used to directly code the second saccade vector. While human volunteers performed a modified double-step task, we examined the role of exocentric encoding in concurrent saccade planning by shifting the first target location well before the efference copy could be used by the oculomotor system. The impact of the first target shift on concurrent processing was tested by examining the end-points of second saccades following a shift of the second target during the first saccade. The frequency of second saccades to the old versus new location of the second target, as well as the propagation of first saccade localization errors, both indices of concurrent processing, were found to be significantly reduced in trials with the first target shift compared to those without it. A similar decrease in concurrent processing was obtained when we shifted the first target but kept constant the second saccade vector. Overall, these results suggest that the brain can use relatively stable visual landmarks, independent of efference copy-based egocentric mechanisms, for concurrent planning of sequential saccades.

Analogous stress and electric field driven structural transformation and decrease in polarization coherence on poling around the morphotropic phase boundary in BiScO3-PbTiO3

The nature of the stress and electric field driven structural and microstructural transformations in the morphotropic phase boundary (MPB) compositions of the high Curie point piezoelectric system BiScO3-PbTiO3 has been examined by ex situ based techniques. Using a powder poling technique, which is based on the concept of exploiting the irreversible structural change that occurs after the application of a strong electric field and stress independently, it was possible to ascertain that both moderate stress and electric field induce identical structural transformation-a fraction of the monoclinic phase transforms irreversibly to the tetragonal phase. Moreover, analysis of the dielectric response before and after poling revealed a counterintuitive phenomenon of poling induced decrease in the spatial coherence of polarization for compositions around the MPB and not so for compositions far away from the MPB range. Exploiting the greater sensitivity of this technique, we demonstrate that the criticality associated with the interferroelectric transition spans a wider composition range than what is conventionally reported in the literature based on bulk x-ray/neutron powder diffraction techniques.

Phytoplankton size structure in the southern Bay of Bengal modified by the Summer Monsoon Current and associated eddies: Implications on the vertical biogenic flux

The present study combines field and satellite observations to investigate how hydrographical transformations influence phytoplankton size structure in the southern Bay of Bengal during the peak Southwest Monsoon/Summer Monsoon (July-August). The intrusion of the Summer Monsoon Current (SMC) into the Bay of Bengal and associated changes in sea surface chemistry, traceable eastward up to 90 degrees E along 8 degrees N, seems to influence biology of the region significantly. Both in situ and satellite (MODIS) data revealed low surface chlorophyll except in the area influenced by the SMC During the study period, two well-developed cydonic eddies (north) and an anti-cyclonic eddy (south), closely linked to the main eastward flow of the SMC, were sampled. Considering the capping effect of the low-saline surface water that is characteristic of the Bay of Bengal, the impact of the cyclonic eddy, estimated in terms of enhanced nutrients and chlorophyll, was mostly restricted to the subsurface waters (below 20 m depth). Conversely, the anti-cyclonic eddy aided by the SMC was characterized by considerably higher nutrient concentration and chlorophyll in the upper water column (upper 60 m), which was contrary to the general characteristic of such eddies. Albeit smaller phytoplankton predominated the southern Bay of Bengal (60-95% of the total chlorophyll), the contribution of large phytoplankton was double in the regions influenced by the SMC and associated eddies. Multivariate analysis revealed the extent to which SMC-associated eddies spatially influence phytoplankton community structure. The study presents the first direct quantification of the size structure of phytoplankton from the southern Bay of Bengal and demonstrates that the SMC-associated hydrographical ramifications significantly increase the phytoplankton biomass contributed by larger phytoplankton and thereby influence the vertical opal and organic carbon flux in the region. (C) 2014 Elsevier B.V. All rights reserved.

Metastable monoclinic and orthorhombic phases and electric field induced irreversible phase transformation at room temperature in the lead-free classical ferroelectric BaTiO3

For decades it has been a well-known fact that among the few ferroelectric compounds in the perovskite family, namely, BaTiO3, KNbO3, PbTiO3, and Na1/2Bi1/2TiO3, the dielectric and piezoelectric properties of BaTiO3 are considerably higher than the others in polycrystalline form at room temperature. Further, similar to ferroelectric alloys exhibiting morphotropic phase boundary, single crystals of BaTiO3 exhibit anomalously large piezoelectric response when poled away from the direction of spontaneous polarization at room temperature. These anomalous features in BaTiO3 remained unexplained so far from the structural standpoint. In this work, we have used high-resolution synchrotron x-ray powder diffraction, atomic resolution aberration-corrected transmission electron microscopy, in conjunction with a powder poling technique, to reveal that at 300 K (i) the equilibrium state of BaTiO3 is characterized by coexistence of metastable monoclinic Pm and orthorhombic (Amm2) phases along with the tetragonal phase, and (ii) strong electric field switches the polarization direction from the 001] direction towards the 101] direction. These results suggest that BaTiO3 at room temperature is within an instability regime, and that this instability is the fundamental factor responsible for the anomalous dielectric and piezoelectric properties of BaTiO3 as compared to the other homologous ferroelectric perovskite compounds at room temperature. Pure BaTiO3 at room temperature is therefore more akin to lead-based ferroelectric alloys close to the morphotropic phase boundary where polarization rotation and field induced ferroelectric-ferroelectric phase transformations play a fundamental role in influencing the dielectric and piezoelectric behavior.

Intermetallic eutectic alloys in the Ni-Al-Zr system with attractive high temperature properties

We describe a group of alloys with ultrahigh strength of about 2 GPa at 700 degrees C and exceptional oxidation resistance to 1100 degrees C. These alloys exploit intermetallic phases with stable oxide forming elements that combine to form fine nanometric scale structures through eutectic transformations in ternary systems. The alloys offer engineering tensile plasticity of about 4% at room temperature though both conventional dislocation mechanisms and twinning in the more complex intermetallic constituent, along with slip lengths that are restricted by the interphase boundaries in the eutectics.

Detection of a quantum particle on a lattice under repeated projective measurements

We consider a quantum particle, moving on a lattice with a tight-binding Hamiltonian, which is subjected to measurements to detect its arrival at a particular chosen set of sites. The projective measurements are made at regular time intervals tau, and we consider the evolution of the wave function until the time a detection occurs. We study the probabilities of its first detection at some time and, conversely, the probability of it not being detected (i.e., surviving) up to that time. We propose a general perturbative approach for understanding the dynamics which maps the evolution operator, which consists of unitary transformations followed by projections, to one described by a non-Hermitian Hamiltonian. For some examples of a particle moving on one-and two-dimensional lattices with one or more detection sites, we use this approach to find exact expressions for the survival probability and find excellent agreement with direct numerical results. A mean-field model with hopping between all pairs of sites and detection at one site is solved exactly. For the one-and two-dimensional systems, the survival probability is shown to have a power-law decay with time, where the power depends on the initial position of the particle. Finally, we show an interesting and nontrivial connection between the dynamics of the particle in our model and the evolution of a particle under a non-Hermitian Hamiltonian with a large absorbing potential at some sites.

Contrasting Effects of Graphene Oxide and Poly(ethylenimine) on the Polymorphism in Poly(vinylidene fluoride)

The nature of interaction between a heteronucleating agent (graphene oxide, GO) and a strongly polar macromolecule (poly(ethylenimine), PEI) with poly(vinylidene fluoride) (PVDF) influencing the crystalline structure and morphology has been systematically investigated in this work. PEI interacts with PVDF via ion-dipole interaction, which helps in lowering the free energy barrier for nucleation thereby promoting faster crystallization. In contrast, besides interacting with PVDF, GO also promotes heteronucleation in PVDF. We observed that both GO and PEI have very different effects on the overall crystalline morphology of PVDF. For instance, the neat PVDF showed a mixture of both alpha and beta phases when cooled from the melt. However, incorporation of 0.1 wt % GO resulted in phase transformation from the stable alpha-phase to polar beta-polymorph in PVDF. In contrast, PEI, which also resulted in faster crystallization in PVDF predominantly, resulted in the stable alpha- phase. Various techniques like Fourier transform infrared spectroscopy, X-ray diffraction, and differential scanning calorimetry were employed to confirm the phase transformations in PVDF. PEI was further grafted onto GO nanosheets to understand the combined effects of both GO and PEI on the polymorphism in PVDF. The PVDF/PEI-GO composite showed a mixture of phases, predominantly rich in a. These phenomenal effects were further analyzed and corroborated with the specific interaction between GO and PEI with PVDF using X-ray photon scattering (XPS) and NMR. In addition, the dielectric permittivity increased significantly in the presence of GO and PEI in the composites. For instance, PVDF/PEI-GO showed the highest permittivity of 39 at 100 Hz.

Fermionic edge states and new physics

We investigate the properties of the Dirac operator on manifolds with boundaries in the presence of the Atiyah-Patodi-Singer boundary condition. An exact counting of the number of edge states for boundaries with isometry of a sphere is given. We show that the problem with the above boundary condition can be mapped to one where the manifold is extended beyond the boundary and the boundary condition is replaced by a delta function potential of suitable strength. We also briefly highlight how the problem of the self-adjointness of the operators in the presence of moving boundaries can be simplified by suitable transformations which render the boundary fixed and modify the Hamiltonian and the boundary condition to reflect the effect of moving boundary.

Feshbach modulation spectroscopy of the Fermi-Hubbard model

In the vicinity of a Feshbach resonance, a system of ultracold atoms in an optical lattice undergoes rich physical transformations which involve molecule formation and hopping of molecules on the lattice and thus goes beyond a single-band Hubbard model description. We explore theoretically the response of this system to a harmonic modulation of the magnetic field, and thus of the scattering length, across the Feshbach resonance. In the regime in which the single-band Hubbard model is still valid, we provide results for the doublon production as a function of the various parameters, such as frequency, amplitude, etc., that characterize the field modulation, as well as the lattice depth. The method may uncover a route towards the efficient creation of ultracold molecules and also provide an alternative to conventional lattice-depth-modulation spectroscopy.