MAGMA
MAGMA is a collection of next generation linear algebra (LA) GPU accelerated libraries?designed and implemented by the team that developed LAPACK and ScaLAPACK.
MAGMA is for heterogeneous GPU-based architectures, it supports
interfaces to current LA packages and standards, e.g., LAPACK and BLAS,
to allow computational scientists to effortlessly port any LA-relying
software components. The main benefits of using MAGMA are that it can
enable applications to fully exploit the power of current heterogeneous
systems of multi/manycore CPUs and multi-GPUs, and deliver the fastest
possible time to an accurate solution within given energy constraints.
Key Features
| ? |
|
By combining the strengths of multicore and GPU architectures, MAGMA? significantly outperforms corresponding packages for any of these homogeneous components taken separately. MAGMA's one-sided factorizations (and linear solvers) on a single Fermi GPU (and a basic CPU host) can outperform state-of-the-art CPU libraries on high-end multi-socket, multicore nodes (e.g., using up to 48 modern cores). The benefits for the two-sided factorizations (bases for eigen- and singular-value solvers) are even greater, as the performance can exceed 10X the performance of a system with 48 modern CPU cores. Architecture-specific performances and comparisons can be found through the MAGMA site.
Availability
MAGMA 1.0 is available as open software under the modified BSD license:
For more information about MAGMA and other CUDA Libraries:
- A paper of MAGMA by examples written by?Andrzej Chrzeszczyk and?Jakub Chrzeszczyk
- MAGMA home page at ICL, University of Tennesee
- CULA Tools by EM Photonics
- See other?GPU Accelerated Libraries