Atlas is the application of this new paradigm to linear algebra software, with the present emphasis on the basic linear algebra subprograms (blas), a widely used, performance critical, linear. Improving linear algebra computation on numa platforms through a uto tuned nested parallelism euro micro pdp 2012. garching (germany) 6 the software the matrix multiplication routine used: the double precision routine dgemm . the blas implementation of the intel mkl toolkit used is the version 10.2.
With the auto tuning method proposed in this work, a reduction in the execution time is achieved with respect to the matrix multiplication of the blas library. the most computationally demanding scientific and engineering problems are solved with large parallel systems. in some cases those systems are non uniform memory access multiprocessors made up of a large number of cores which share a. Improving linear algebra computation on numa platforms through auto tuned nested parallelism authors : javier cuenca , luis p. garcia , domingo gimenez authors info & claims pdp '12: proceedings of the 2012 20th euromicro international conference on parallel, distributed and network based processing. The most computationally demanding scientific and engineering problems are solved with large parallel systems. in some cases those systems are non uniform memory access multiprocessors made up of a large number of cores which share a hierarchically organized memory. basic linear algebra routines of the type of blas typically constitute the kernel of the computation for those problems, and the. Doi: 10.1109 pdp.2012.12 corpus id: 18607649; improving linear algebra computation on numa platforms through auto tuned nested parallelism @article{cuenca2012improvingla, title={improving linear algebra computation on numa platforms through auto tuned nested parallelism}, author={javier cuenca and luis pedro garc{\'i}a and domingo gim{\'e}nez}, journal={2012 20th euromicro international.
The most computationally demanding scientific and engineering problems are solved with large parallel systems. in some cases those systems are non uniform memory access multiprocessors made up of a large number of cores which share a hierarchically organized memory. basic linear algebra routines of the type of blas typically constitute the kernel of the computation for those problems, and the. Doi: 10.1109 pdp.2012.12 corpus id: 18607649; improving linear algebra computation on numa platforms through auto tuned nested parallelism @article{cuenca2012improvingla, title={improving linear algebra computation on numa platforms through auto tuned nested parallelism}, author={javier cuenca and luis pedro garc{\'i}a and domingo gim{\'e}nez}, journal={2012 20th euromicro international. Basic linear algebra routines of the type of blas typically constitute the kernel of the computation for those problems, and the efficient use of these routines in those systems would contribute. The overall principle of our profiling and simulation experiments is as follows, given a task based application to be run on a target platform: 1. platform specifics such as l3 caches, numa nodes, and architectural link bandwidths are discerned through manufacturer documentation and benchmarking.
Comments are closed.