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.

Structural basis for the function of anti-idiotypic antibody in immune memory

We had earlier proposed a hypothesis to explain the mechanism of perpetuation of immunological memory based on the operation of idiotypic network in the complete absence of antigen. Experimental evidences were provided for memory maintenance through anti-idiotypic antibody (Ab2) carrying the internal image of the antigen. In the present work, we describe a structural basis for such memory perpetuation by molecular modeling and structural analysis studies. A three-dimensional model of Ab2 was generated and the structure of the antigenic site on the hemagglutinin protein H of Rinderpest virus was modeled using the structural template of hemagglutinin protein of Measles virus. Our results show that a large portion of heavy chain containing the CDR regions of Ab2 resembles the domain of the hemagglutinin housing the epitope regions. The similarity demonstrates that an internal image of the H antigen is formed in Ab2, which provides a structural basis for functional mimicry demonstrated earlier. This work brings out the importance of the structural similarity between a domain of hemagglutinin protein to that of its corresponding Ab2. It provides evidence that Ab2 is indeed capable of functioning as surrogate antigen and provides support to earlier proposed relay hypothesis which has provided a mechanism for the maintenance of immunological memory.

Structural basis for the function of anti-idiotypic antibody in immune memory

We had earlier proposed a hypothesis to explain the mechanism of perpetuation of immunological memory based on the operation of idiotypic network in the complete absence of antigen. Experimental evidences were provided for memory maintenance through anti-idiotypic antibody (Ab(2)) carrying the internal image of the antigen. In the present work, we describe a structural basis for such memory perpetuation by molecular modeling and structural analysis studies. A three-dimensional model of Ab(2) was generated and the structure of the antigenic site on the hemagglutinin protein H of Rinderpest virus was modeled using the structural template of hemagglutinin protein of Measles virus. Our results show that a large portion of heavy chain containing the CDR regions of Ab(2) resembles the domain of the hemagglutinin housing the epitope regions. The similarity demonstrates that an internal image of the H antigen is formed in Ab(2), which provides a structural basis for functional mimicry demonstrated earlier. This work brings out the importance of the structural similarity between a domain of hemagglutinin protein to that of its corresponding Ab(2). It provides evidence that Ab(2) is indeed capable of functioning as surrogate antigen and provides support to earlier proposed relay hypothesis which has provided a mechanism for the maintenance of immunological memory.

Focused Prefetching: Performance Oriented Prefetching Based On Commit Stalls

Loads that miss in L1 or L2 caches and waiting for their data at the head of the ROB cause significant slow down in the form of commit stalls. We identify that most of these commit stalls are caused by a small set of loads, referred to as LIMCOS (Loads Incurring Majority of COmmit Stalls). We propose simple history-based classifiers that track commit stalls suffered by loads to help us identify this small set of loads. We study an application of these classifiers to prefetching. The classifiers are used to train the prefetcher to focus on the misses suffered by LIMCOS. This, referred to as focused prefetching, results in a 9.8% gain in IPC over naive GHB based delta correlation prefetcher along with a 20.3% reduction in memory traffic for a set of 17 memory-intensive SPEC2000 benchmarks. Another important impact of focused prefetching is a 61% improvement in the accuracy of prefetches. We demonstrate that the proposed classification criterion performs better than other existing criteria like criticality and delinquent loads. Also we show that the criterion of focusing on commit stalls is robust enough across cache levels and can be applied to any prefetcher without any modifications to the prefetcher.

Comment on path integration of a quadratic action with a generalized memory

Using the promeasure technique, we give an alternative evaluation of a path integral corresponding to a quadratic action with a generalized memory.

Constitutive modeling of shape memory alloy wire with non-local rate kinetics

A constitutive modeling approach for shape memory alloy (SMA) wire by taking into account the microstructural phase inhomogeneity and the associated solid-solid phase transformation kinetics is reported in this paper. The approach is applicable to general thermomechanical loading. Characterization of various scales in the non-local rate sensitive kinetics is the main focus of this paper. Design of SMA materials and actuators not only involve an optimal exploitation of the hysteresis loops during loading-unloading, but also accounts for fatigue and training cycle identifications. For a successful design of SMA integrated actuator systems, it is essential to include the microstructural inhomogeneity effects and the loading rate dependence of the martensitic evolution, since these factors play predominant role in fatigue. In the proposed formulation, the evolution of new phase is assumed according to Weibull distribution. Fourier transformation and finite difference methods are applied to arrive at the analytical form of two important scaling parameters. The ratio of these scaling parameters is of the order of 10(6) for stress-free temperature-induced transformation and 10(4) for stress-induced transformation. These scaling parameters are used in order to study the effect of microstructural variation on the thermo-mechanical force and interface driving force. It is observed that the interface driving force is significant during the evolution. Increase in the slopes of the transformation start and end regions in the stress-strain hysteresis loop is observed for mechanical loading with higher rates.

Reflexive Incidence Matrx (RIM) Representation of Petr Nets

Although incidence matrix representation has been used to analyze the Petri net based models of a system, it has the limitation that it does not preserve reflexive properties (i.e., the presence of selfloops) of Petri nets. But in many practical applications self-loops play very important roles. This paper proposes a new representation scheme for general Petri nets. This scheme defines a matrix called "reflexive incidence matrix (RIM) c which is a combination of two matrices, a "base matrix Cb,,, and a "power matrix CP." This scheme preserves the reflexive and other properties of the Petri nets. Through a detailed analysis it is shown that the proposed scheme requires less memory space and less processing time for answering commonly encountered net queries compared to other schemes. Algorithms to generate the RIM from the given net description and to decompose RIM into input and output function matrices are also given. The proposed Petri net representation scheme is very useful to model and analyze the systems having shared resources, chemical processes, network protocols, etc., and to evaluate the performance of asynchronous concurrent systems.

Rotating Beams and Nonrotating Beams With Shared Eigenpair

In this paper, we look for rotating beams whose eigenpair (frequency and mode-shape) is the same as that of uniform nonrotating beams for a particular mode. It is found that, for any given mode, there exist flexural stiffness functions (FSFs) for which the jth mode eigenpair of a rotating beam, with uniform mass distribution, is identical to that of a corresponding nonrotating uniform beam with the same length and mass distribution. By putting the derived FSF in the finite element analysis of a rotating cantilever beam, the frequencies and mode-shapes of a nonrotating cantilever beam are obtained. For the first mode, a physically feasible equivalent rotating beam exists, but for higher modes, the flexural stiffness has internal singularities. Strategies for addressing the singularities in the FSF for finite element analysis are provided. The proposed functions can be used as test-functions for rotating beam codes and for targeted destiffening of rotating beams.

Comment on “path integration of a quadratic action with a generalized memory

Using the promeasure technique, we give an alternative evaluation of a path integral corresponding to a quadratic action with a generalized memory.

Parallel polygon scan conversion algorithms: Performance evaluation on a shared bus architecture

In this paper, three parallel polygon scan conversion algorithms have been proposed, and their performance when executed on a shared bus architecture has been compared. It has been shown that the parallel algorithm that does not use edge coherence performs better than those that use edge coherence. Further, a multiprocessing architecture has been proposed to execute the parallel polygon scan conversion algorithms more efficiently than a single shared bus architecture.

Sensing Shape Recovery Using Conductivity Noise in Thin Films of NiTi Shape Memory Alloys

Low frequency fluctuations in the electrical resistivity, or noise, have been used as a sensitive tool to probe into the temperature driven martensite transition in dc magnetron sputtered thin films of nickel titanium shape-memory alloys. Even in the equilibrium or static case, the noise magnitude was more than nine orders of magnitude larger than conventional metallic thin films and had a characteristic dependence on temperature. We observe that the noise while the temperature is being ramped is far larger as compared to the equilibrium noise indicating the sensitivity of electrical resistivity to the nucleation and propagation of domains during the shape recovery. Further, the higher order statistics suggests the existence of long range correlations during the transition. This new characterization is based on the kinetics of disorder in the system and separate from existing techniques and can be integrated to many device applications of shape memory alloys for in-situ shape recovery sensing.

Low Complexity LID using Pruned Pattern Tables of LZW

We present two discriminative language modelling techniques for Lempel-Ziv-Welch (LZW) based LID system. The previous approach to LID using LZW algorithm was to directly use the LZW pattern tables forlanguage modelling. But, since the patterns in a language pattern table are shared by other language pattern tables, confusability prevailed in the LID task. For overcoming this, we present two pruning techniques (i) Language Specific (LS-LZW)-in which patterns common to more than one pattern table are removed. (ii) Length-Frequency product based (LF-LZW)-in which patterns having their length-frequency product below a threshold are removed. These approaches reduce the classification score (Compression Ratio [LZW-CR] or the weighted discriminant score [LZW-WDS]) for non native languages and increases the LID performance considerably. Also the memory and computational requirements of these techniques are much less compared to basic LZW techniques.

Effect of mechanical cycling on the stress-strain response of a martensitic Nitinol shape memory alloy

An experimental investigation into the ambient temperature, load-controlled tension�tension fatigue behavior of a martensitic Nitinol shape memory alloy (SMA) was conducted. Fatigue life for several stress levels spanning the critical stress for detwinning was determined and compared with that obtained on an alloy similar in composition but in the austenitic state at room temperature. Results show that the fatigue life of the pseudo-plastic alloy is superior to superelastic shape memory alloy. The stress�strain hysteretic response, monitored throughout the fatigue loading, reveals progressive strain accumulation with the cyclic loading. In addition, the area of hysteresis and recoverable and frictional energies were found to decrease with increasing number of fatigue cycles. Post-mortem characterization of the fatigued specimens through calorimetry and fractography was conducted in order to get further insight into the fatigue micromechanisms. These results are discussed in terms of reversible and irreversible microstructural changes that take place during cyclic loading. Aspects associated with self-heating of martensitic alloy undergoing high frequency stress cycling are discussed.

Direct physical evidence for the back-transformation of stress-induced martensite in the vicinity of cracks in pseudoelastic NiTi shape memory alloys

Crack loading and crack extension in pseudoelastic binary NiTi shape memory alloy (SMA) miniature compact tension (CT) specimens with 50.7 at.% Ni (austenitic, pseudoelastic) was investigated using infrared (IR) thermography during in situ loading and unloading. IR thermographic measurements allow for the observation of heat effects associated with the stress-induced transformation of martensite from B2 to BIT during loading and the reverse transformation during unloading. The results are compared with optical images and discussed in terms of the crack growth mechanisms in pseudoelastic NiTi SMAs. Direct experimental evidence is presented which shows that crack growth occurs into a stress-induced martensitic microstructure, which immediately retransforms to austenite in the wake of the crack.

A fluctuation-based characterization of athermal phase transitions:Application to shape memory alloys

We employ a fluctuation-based technique to investigate the athermal component associated with martensite phase transition, which is a prototype of temperature-driven structural transformation. Statistically, when the phase transition is purely athermal, we find that the temporal sequence of avalanches under constant drive is insensitive to the drive rate. We have used fluctuations in electrical resistivity or noise in nickel titanium shape memory alloys in three different forms: a thin film exhibiting well-defined transition temperatures,a highly disordered film, and a bulk wire of rectangular cross-section. Noise is studied in the realm of dynamic transition,viz.while the temperature is being ramped, which probes into the kinetics of the transformation at real time scales,and could probably stand out as a promising tool for material testing in various other systems, including nanoscale devices.