================================================================================ CULA Sparse S5 (CUDA 5.0) Release Notes EM Photonics, Inc. ================================================================================ -------------------------------------------------------------------------------- Installation Instructions -------------------------------------------------------------------------------- For installation instructions, please consult the CULAProgrammersGuide.pdf file included in the 'doc' folder of your CULA Sparse distribution. -------------------------------------------------------------------------------- System Requirements -------------------------------------------------------------------------------- CULA requires that your system be equipped with a NVIDIA CUDA-compatible device in order to run CULA-enabled programs. The NVIDIA drivers must be version 306.94 (or greater) for Windows systems and 304.54 for Linux systems. Mac OS X systems must have 5.0.36 or newer. -------------------------------------------------------------------------------- Supported Operating Systems -------------------------------------------------------------------------------- All systems feature 32-bit and 64-bit support. * Windows XP / Vista / 7 * Ubuntu Linux 10.04 (and newer) * Red Hat Enterprise Linux 5.7 (and newer) * Fedora 16 * Mac OSX 10.6 Snow Leopard / 10.7 Lion -------------------------------------------------------------------------------- Revision History -------------------------------------------------------------------------------- CULA Sparse S5 (May 8, 2013) CULA Sparse S4 (October 16, 2012) CULA Sparse S3 (August 14, 2012) CULA Sparse S2 (January 30, 2012) CULA Sparse S1 (November 2, 2012) CULA Sparse S1 Beta 2 (September 27, 2011) CULA Sparse S1 Beta 1 (August 24, 2011) -------------------------------------------------------------------------------- Changelog -------------------------------------------------------------------------------- Release S5 CUDA 5.0 (May 8, 2012) --------------------------------- * Feature: New plan based interface * Feature: Support for user defined matrix-free solvers * Feature: Support user defined preconditioners * Feature: Ability to cache preconditioner generation * Feature: Ability to cache CUDA data preperation * Feature: High performance multi-threaded CPU support * Feature: CUDA device support for memory allocated by cudaMalloc * Feature: Added sparse approximate inverse preconditioner (fainv) * Feature: Added factorized sparse approximate inverse preconditioner (ainv) * Feature: Many new options available * Feature: Fortran90 module interface added for legacy interface * Feature: Null parameter for options will now results in default options * Improved: Descriptive error messages via culaSparseGetLastStatusString * Changed: API overhaul; see documentation for all changes * Changed: Residual configuration option changed to relativeResidual * Changed: A number of options are more specific in name to prevent conflicts * Changed: culaSparse now prefixes every function and control structure * Changed: Dependency of cula_core library removed * Changed: Explicit handle management now required * Changed: Functionality of cula_sparse.h moved to cula_sparse_legacy.h * Changed: Runtime dependency on Intel OpenMP 5 redistributable (libiomp5) added * Changed: Indexing and reordering option moved to data options structures * Removed: Fortran compatibility libraries; please use module files Release S4 CUDA 5.0 (October 16, 2012) -------------------------------------- * Feature: CUDA runtime upgraded to 5.0 * Feature: K20 support Release S3 CUDA 4.2 (August 14, 2012) -------------------------------------- * Announcement: All packages are now "universal" and contain both 32-bit and 64-bit binaries * Feature: CUDA runtime upgraded to 4.2 * Feature: Kepler support * Changed: Fortran module is now located in "include" Release S2 CUDA 4.1 (January 30, 2012) -------------------------------------- * Feature: CUDA runtime upgraded to version 4.1 * Improved: Stability of COO and CSC interfaces * Fixed: Now shipping all dependencies required by OSX systems Release S1 CUDA 4.0 (November 2, 2011) -------------------------------------- * Feature: Improved speeds for all solvers * Feature: Matrix reordering option; can lead to large perf gains for ILU * Feature: MINRES solver * Feature: Fully compatible with CULA R13 and above * Feature: Option to disable stagnation checking for more speed * Feature: Added iterativeBenchmark example for evaluating the performance of different solvers and options * Improved: Result printout will show if useBestAnswer was invoked * Changed: Header renamed to cula_sparse.h; transitional header available * Notice: Integrated LGPL COLAMD package; see src folder and license Release S1 Beta 2 CUDA 4.0 (September 27, 2011) ----------------------------------------------- * Feature: BiCGSTAB solver * Feature: BiCGSTAB(L) solver * Feature: Complex (Z) data types available for all solvers * Feature: Fortran module added * Feature: Configuration parameter to return best experienced solution * Feature: Maximum runtime configuration parameter * Feature: New example for Fortran interface * Feature: New example for MatrixMarket data * Changed: Must link two libraries now (cula_sparse and cula_core) Release S1 Beta 1 CUDA 4.0 (August 24, 2011) -------------------------------------------- * Feature: Cg, BiCg, and GMRES solvers * Feature: CSC, CSR, COO storage formats * Feature: Jacobi, Block Jacobi, ILU0 preconditioners * Feature: Double precision only * Feature: Support for all standard CUDA platforms; Linux 32/64, Win 32/64, OSX -------------------------------------------------------------------------------- More Information -------------------------------------------------------------------------------- For more information on the CULAtools family of products, please visit our webpage at http://www.culatools.com To provide feedback, please visit http://www.culatools.com/forums and post in the appropriate forum topic.