We develop numerical computation library KMATHLIB for K computer.
EigenExa, a part of KMATHLIB, is a high performance eigen-solver.
KMATH_EIGEN_GEV, a part of KMATHLIB, is a high performance generalized eigen-solver.
KMATH_RANDOM, a part of KMATHLIB, is a high performance pseudorandom number generator.
ISC '13 poster presentation
ASPEN.K2 provides GPU/CUDA users with automatic-tuned high performance Level2 BLAS kernels.
MUBLAS-GEMV is an optimized implementation of GEMV for NVIDIA GPUs. This implementation automatically adjusts the thread-block size based on the theoretical performance model before launching kernel. This code will be integrated into ASPEN.K2 in the near future.
Development of a communication time reduction tools
We develop a communication time reduction tools reduction tool, based on RMATT of Fujitsu, for large-scale computations.
Development of a high performance FFT program
We develop a high performance FFT program tuned for K computer.