Contemporary GPUs have significantly higher arithmetic throughput than a memory
throughput. Hence, many GPU kernels are memory bound and cannot exploit arithmetic …

Memory bandwidth limits the performance of important kernels in many scientific
applications. Such applications often use sequences of Basic Linear Algebra Subprograms …

(57) ABSTRACT A method for optimization of machine learning (ML) work loads on a
graphics processor unit (GPU). The method includes identifying a computation having a …

Programming Abstractions for Data Locality Page 1 Programming Abstractions for Data Locality
Item Type Technical Report Authors Tate, Adrian;Kamil, Amir;Dubey, Anshu;Groblinger …

The performance bottleneck for many scientific applications is the cost of memory access
inside linear algebra kernels. Tuning such kernels for memory efficiency is a complex task …

GraphBLASis a recent standard that allows the expression of graph algorithms in the
language of linear algebra and enables automatic code parallelization and optimization …

While recognition of the advantages of heterogeneous computing is steadily growing, the
issues of programmability and portability hinder its exploitation. The introduction of the …

The design and implementation of high level tools for parallel programming is a major
challenge as the complexity of modern architectures increases. Domain Specific Languages …

Object oriented languages can be easily extended with new data types, which facilitate
prototy** new language extensions. A very challenging problem is the development of …

R is one of the most popular programming languages for statistics and machine learning, but
it is slow and unable to scale to large datasets. The general approach for having an efficient …