Using Relationships for Matching Textual Domain Models with Existing Code

We address the task of mapping a given textual domain model (e.g., an industry-standard reference model) for a given domain (e.g., ERP), with the source code of an independently developed application in the same domain. This has applications in improving the understandability of an existing application, migrating it to a more flexible architecture, or integrating it with other related applications. We use the vector-space model to abstractly represent domain model elements as well as source-code artifacts. The key novelty in our approach is to leverage the relationships between source-code artifacts in a principled way to improve the mapping process. We describe experiments wherein we apply our approach to the task of matching two real, open-source applications to corresponding industry-standard domain models. We demonstrate the overall usefulness of our approach, as well as the role of our propagation techniques in improving the precision and recall of the mapping task.

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.

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.

Space-Time Coded Spatial Modulated Physical Layer Network Coding for Two-Way Relaying

Using the spatial modulation approach, where only one transmit antenna is active at a time, we propose two transmission schemes for two-way relay channel using physical layer network coding with space time coding using coordinate interleaved orthogonal designs (CIODs). It is shown that using two uncorrelated transmit antennas at the nodes, but using only one RF transmit chain and space-time coding across these antennas can give a better performance without using any extra resources and without increasing the hardware implementation cost and complexity. In the first transmission scheme, two antennas are used only at the relay, adaptive network coding (ANC) is employed at the relay and the relay transmits a CIOD space time block code (STBC). This gives a better performance compared to an existing ANC scheme for two-way relay channel which uses one antenna each at all the three nodes. It is shown that for this scheme at high SNR the average end-to-end symbol error probability (SEP) is upper bounded by twice the SEP of a point-to-point fading channel. In the second transmission scheme, two transmit antennas are used at all the three nodes, CIOD STBCs are transmitted in multiple access and broadcast phases. This scheme provides a diversity order of two for the average end-to-end SEP with an increased decoding complexity of O(M-3) for an arbitrary signal set and O(M-2 root M) for square QAM signal set. Simulation results show that the proposed schemes performs better than the existing ANC schemes under perfect and imperfect channel state information.

LDPC Code Designs Based on root I Matrices

LDPC codes can be constructed by tiling permutation matrices that belong to the square root of identity type and similar algebraic structures. We investigate into the properties of such codes. We also present code structures that are amenable for efficient encoding.

An Optimizing Code Generator for a Class of Lattice-Boltzmann Computations

The Lattice-Boltzmann method (LBM), a promising new particle-based simulation technique for complex and multiscale fluid flows, has seen tremendous adoption in recent years in computational fluid dynamics. Even with a state-of-the-art LBM solver such as Palabos, a user has to still manually write the program using library-supplied primitives. We propose an automated code generator for a class of LBM computations with the objective to achieve high performance on modern architectures. Few studies have looked at time tiling for LBM codes. We exploit a key similarity between stencils and LBM to enable polyhedral optimizations and in turn time tiling for LBM. We also characterize the performance of LBM with the Roofline performance model. Experimental results for standard LBM simulations like Lid Driven Cavity, Flow Past Cylinder, and Poiseuille Flow show that our scheme consistently outperforms Palabos-on average by up to 3x while running on 16 cores of an Intel Xeon (Sandybridge). We also obtain an improvement of 2.47x on the SPEC LBM benchmark.

GRMHD formulation of highly super-Chandrasekhar rotating magnetized white dwarfs: stable configurations of non-spherical white dwarfs

Here we extend the exploration of significantly super-Chandrasekhar magnetized white dwarfs by numerically computing axisymmetric stationary equilibria of differentially rotating magnetized polytropic compact stars in general relativity (GR), within the ideal magnetohydrodynamic regime. We use a general relativistic magnetohydrodynamic (GRMHD) framework that describes rotating and magnetized axisymmetric white dwarfs, choosing appropriate rotation laws and magnetic field profiles (toroidal and poloidal). The numerical procedure for finding solutions in this framework uses the 3 + 1 formalism of numerical relativity, implemented in the open source XNS code. We construct equilibrium sequences by varying different physical quantities in turn, and highlight the plausible existence of super-Chandrasekhar white dwarfs, with masses in the range of 2-3 solar mass, with central (deep interior) magnetic fields of the order of 10(14) G and differential rotation with surface time periods of about 1-10 s. We note that such white dwarfs are candidates for the progenitors of peculiar, overluminous Type Ia supernovae, to which observational evidence ascribes mass in the range 2.1-2.8 solar mass. We also present some interesting results related to the structure of such white dwarfs, especially the existence of polar hollows in special cases.

MPI parallel programming of mixed integer optimization problems using CPLEX with COIN-OR

The aim of this technical report is to present some detailed explanations in order to help to understand and use the Message Passing Interface (MPI) parallel programming for solving several mixed integer optimization problems. We have developed a C++ experimental code that uses the IBM ILOG CPLEX optimizer within the COmputational INfrastructure for Operations Research (COIN-OR) and MPI parallel computing for solving the optimization models under UNIX-like systems. The computational experience illustrates how can we solve 44 optimization problems which are asymmetric with respect to the number of integer and continuous variables and the number of constraints. We also report a comparative with the speedup and efficiency of several strategies implemented for some available number of threads.

Lancaster and Morecambe College: Uses serious computer games as a teaching tool

John Latham, International Projects Co-ordinator at Lancaster and Morecambe College (LMC), got involved with the project Serious Computer Games as a Teaching Tool (SCOGATT) after using the game EnerCities with his students. The web based platform at www.scogatt.eu serves as a One Stop Toolkit for vocational teachers and trainers who want to use serious computer games (SCG) in their teaching environments but might need a helping hand. There they will be able to find a compendium of serious games, SCOGATT pilot results, teacher reports and the exemplar games, EnerCities.