MAGMA MIC enables ease of use and adoption of Intel Xeon Phi architectures in applications such as linear algebra which is fundamental to scientific computing.
MAGMA MIC uses hybrid algorithms where the computation is split into tasks of varying granularity and their execution is scheduled over the hardware components. Scheduling can be static or dynamic. In either case, small non-parallelizable tasks, often on the critical path, are scheduled on the CPU, and larger more parallelizable ones, often Level 3 BLAS, are scheduled on the MICs.
MAGMA MIC uses hybrid algorithms, representing linear algebra algorithms as collections of tasks and data dependencies among them. They are properly scheduling tasks' execution over multicore CPUs and manycore coprocessors.
The goal is to derive new methods and algorithmic improvements, to develop linear algebra on small matrices, and to develop numerical algorithms that recognize and exploit the presence of mixed-precision mathematics.
Other aims are to develop a set of benchmarks for both performance and energy consumption and to show essential communication and computation patterns in various applications. The goal is to encourage the focus of both hardware and software developers on architecture features and application needs, and to incorporate them in performance analysis tools.