This note describes some of the wrapper routines and libraries that are part of the Intel MKL package. In particular it shows how Intel optimized versions fo the fftw2 and fftw3 can be called and linked with your programs. For additional information on MKL and the Intel compilers see http://mindy.mines.edu/intel/.
This note discusses how you can set up dependancies in slurm jobs so a second job waits for a first to finish before automatically starting. In particular, this shows how to set it up so that if the first job fails then the second will not start.
This note discusses setting of srun, sbatch, and a number of environmental and their effect on mapping of threads to cores and thus performance for OpenMP programs. We look at the Intel, Portland Group, and gnu compilers.
This note discusses setting of a number of environmental and their effect on mapping of threads to cores and thus performance for Hybrid MPI/OpenMP programs running on our Power8 nodes. We look at the IBM versions of MPI and compilers.
Qbox is a scalable parallel implementation of first-principles molecular dynamics (FPMD) based on the plane-wave, pseudopotential formalism. This note describes running the application on our BlueGene Q in a hybrid mode. We see that the best performance is obtained when we combine MPI and threading and the nodes are oversubscribed.
Quantum Espresso (QE) is an example of an ab-initio electronic structure code based on a plane-wave, pseudopotential implementation of Density Functional Theory. Besides exploiting the parallel capabilities of the HPC platform architecture, QE carries its own levels of parallelization, a common environment in electronic structure applications. This note covers the initial phases of the optimization process, discussing choices and their consequences, and explores the extent to which true optimization can be realized.
This is an online tutorial on the Linux operating system that is common on most HPC platforms and is used on the Mines' machines Mio, BlueM, AuN and Mc2.