Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 12 of 12 for “"Communication avoiding"”.

  1. Communication avoiding parallel algorithms for amorphous problems

    … problems are specially costly since all-to-all communications are required and delivering an efficient parallel algorithm becomes more challenging. Another difficulty with these problems is that the amount of parallelism in them is limited which naturally makes them hard to parallelize. This is …

    uiuc Repository record for Communication avoiding parallel algorithms for amorphous problems (opens in a new tab)

  2. Analysis and Design of Communication Avoiding Algorithms for Out of Memory(OOM) SVD

    … starts cannot be used directly. Moving data (communication) between levels of memory hierarchy and the disk exposes extra challenges to design SVD for such big matrices because of the exponential growth in the gap between floating-point arithmetic rate and bandwidth for many different storage …

    utk-theses Repository record for Analysis and Design of Communication Avoiding Algorithms for Out of Memory(OOM) SVD (opens in a new tab)

  3. Asynchronous versions of Jacobi, multigrid, and Chebyshev solvers

    … alternate between phases of computation and communication. Standard software packages use synchronous implementations where there are one or more synchronization points per iteration. These synchronization points occur during communication phases where each process sends data to other …

    gatech Repository record for Asynchronous versions of Jacobi, multigrid, and Chebyshev solvers (opens in a new tab)

  4. Reducing Communication in the Solution of Linear Systems

    … growing performance gap between computation and communication on modern computers, making it crucial to develop algorithms with lower latency and bandwidth requirements. Because systems of linear equations are important for numerous scientific and engineering applications, I have studied several …

    utk-theses Repository record for Reducing Communication in the Solution of Linear Systems (opens in a new tab)

  5. Polynomial preconditioning with the minimum residual polynomial.

    Krylov subspace methods are often used to solve large, sparse systems of linear equations Ax=b. Preconditioning can help accelerate the Krylov iteration and reduce costs for solving the problem. We study a polynomial preconditioner p(A) based upon the minimum residual polynomial from GMRES. The …

    baylor Repository record for Polynomial preconditioning with the minimum residual polynomial. (opens in a new tab)

  6. Implementation and analysis of a parallel vertex-centered finite element segmental refinement multigrid solver

    … can be used to avoid all inter-process communication on the fine grids. While domain decomposition methods generally require coupled subdomain processing for the numerical solution to a nonlinear elliptic boundary value problem, segmental refinement exploits that subdomains are almost …

    gatech Repository record for Implementation and analysis of a parallel vertex-centered finite element segmental refinement multigrid solver (opens in a new tab)

  7. Scalable and distributed constrained low rank approximations

    … computation is inevitable and the network communication becomes a major performance bottleneck. Towards this end, we propose a novel distributed Communication Avoiding NMF (CANMF) algorithm that communicates only the right low rank factor to its neighboring machine. Finally, a general …

    gatech Repository record for Scalable and distributed constrained low rank approximations (opens in a new tab)

  8. Code Generation for Extreme Scale Parallel Systems

    … However, the inter- and intra-processor communication capabilities on such systems are not increasing at the same rate as the on-node computing capability. Consequently, achieving high performance requires careful orchestration of both single- and multiprocessor parallelism. This thesis …

    rice Repository record for Code Generation for Extreme Scale Parallel Systems (opens in a new tab)

  9. New Parallel Algorithms for Large-Scale Matrix Computations

    … In this dissertation, we first propose the communication-avoiding 3D matrix multiplication algorithm (CA3DMM). CA3DMM is an approach to parallelize general dense matrix multiplication that minimizes communication sizes by optimizing matrix partitioning and data transfer patterns. CA3DMM …

    gatech Repository record for New Parallel Algorithms for Large-Scale Matrix Computations (opens in a new tab)

  10. Evaluation of Distributed Programming Models and Extensions to Task-based Runtime Systems

    … of combining both the algorithmic approach of Communication-Avoiding (CA) and the communication-computation overlapping benefits provided by runtime systems using 2D five-point stencil as the test case. This broad programming models evaluation and extension work highlighted the abilities of …

    utk-theses Repository record for Evaluation of Distributed Programming Models and Extensions to Task-based Runtime Systems (opens in a new tab)

  11. Scalable and resilient sparse linear solvers

    … parallel computer systems are the high-cost of communication and the expected decrease in reliability of the hardware components. This dissertation contributes new techniques to address these issues. Regarding communication, we make two advances to reduce both on-node and inter-node …

    gatech Repository record for Scalable and resilient sparse linear solvers (opens in a new tab)