v0.7 Released 2017-03-28 --------------------- * Tarball supports higher AM (7) * Compilation to lower AM is supported (SIMINT_MAXAM variable) * Fix bad logic in primitive screening code * Some cmake tweaks * SIMINT_STANDALONE - as self-contained library as possible * Standardized install paths * More proper simintConfig.cmake file * Better handling of restrict keyword v0.6 Released 2017-01-24 --------------------- * Fixed major bug in SSE and AVX512 contraction code * Implemented vectorized Boys function * Fixes some issues with Clang v0.5 Released 2017-01-19 --------------------- * Interface changes: * simint_compute_eri_sharedwork has been changed to simint_compute_eri and the old simint_compute_eri has been removed. * Separate microarchitecture packages reduced to one * Compiler fixes for GCC and Intel * Experimental support for Intel KNL * Better handling of permutational symmetry for higher AM v0.4 Released 2016-11-08 --------------------- * Compilation works with GCC * Implemented primitive screening (Schwarz screening) * Interface changes * Add screening arguments to some functions * Convenience header "simint/simint.h" * Fix miscellaneous rough edges in the build * Now CMAKE_BUILD_TYPE and arguments such as "-xavx" do not need to be passed explicitly to cmake * OpenMP is not needed to build the library itself * SSE version is re-enabled * Static linking of Intel + MKL v0.3 Released 2016-09-23 --------------------- * Interface changes: Consistent function names and arguments * Workaround for bug in Intel 2017 compiler * Removed benchmark/testing interface to libint/liberd * Added permutations of angular momentum (ie, ( s p | s s ), etc) * Benchmark code is now threaded