GROMACS 2016.4 Release Notes

These release notes document the changes that have taken place in GROMACS since version 2016.3 to fix known issues. It also incorporates all fixes made in version 5.1.4 and several since.

• Fixes where mdrun could behave incorrectly
  • Disabled PME tuning with the group scheme
  • Fixed value of Ewald shift
  • Fixed orientation restraint reference
  • Fixed bugs with setup for orientation restraints
  • Worked around missing OpenMP implementation in orientation restraints
  • Enable group-scheme SIMD kernels on recent AVX extensions
  • Fix FEP state with rerun
  • Fixed COM pull force with SD
  • Fix PBC bugs in the swap code
  • Fixed flat-bottomed position restraints with multiple ranks
  • Fixed null pointer print in DD
  • Improved the “files not present” error message
  • Fixed LJ-PME + switch grompp error
  • Fixed unused SIMD PME table kernel
  • Fixed bugs in most double-precision Simd4 implementations
  • Avoid inf in SIMD double sqrt()
  • Fix NVML error messages
  • Fixed IMD interface malfunctions
  • Fixed initial temperature reporting
  • Fix exception in SIMD LJ PME solve
• Fixes for gmx tools
  • Fixed memory access issues in gmx solvate
  • Fixed a consistency check in gmx make_edi for flooding
  • Supported quiet trajectory-handling I/O
• Improved documentation
  • Migrated much content from the wiki to the user guide
  • Corrected the PDF manual to reflect that all tools are called gmx <tool>
  • Clarified gmx editconf help text
  • Added missing .mdp file documentation for the enforced rotation module
  • Fixed parameter description for dihedral_restraints
  • Replaced instance of “group” by “coord” in pull .mdp documentation
• Portability enhancements
  • Supported CUDA 9/Volta for nonbonded kernels
  • Really enabled AVX512 in the GROMACS-managed build of FFTW
  • Fixed aspects for compiling and running on Solaris
  • Fixed AVX512F compiler flags
  • Fixed compiler flags for using MKL
  • Fixes compilation issues with ARM SIMD