GROMACS 2019.2 release notes

This version was released on April 16th, 2019. These release notes document the changes that have taken place in GROMACS since the previous 2019.1 version, to fix known issues. It also incorporates all fixes made in version 2018.6 and earlier, which you can find described in the Release notes.

Fixes where mdrun could behave incorrectly

Fix L-BGFS minimizer

The minimizer could fail on a number of systems.

Issue 2641

Disallow pull geometry direction-periodic with AWH

This could lead to incorrect behavior or a cryptic error message.

Issue 2923

Fixed mdrun -nsteps option

Fixed that the, deprecated, mdrun option -nsteps only allowed extension of the simulation under certain conditions.

Issue 2881

Fixes for gmx tools

gmx cluster -clndx indices now correct

The reported indices of trajectory frames in clusters were too small by one.

Issue 2926

gmx editconf -f in.pdb -o out.pdb again preserves chain IDs

This had been inadvertently broken and is now fixed.

Issue 2900

Tools again accept .tpr files as input

The pdb2gmx, solvate, and insert-molecules tools could no longer accept input configurations contained in .tpr format files. This is now fixed.

Issue 2900

Fix segmentation fault when preparing simulated annealing inputs

grompp was unable to prepare tpr files for inputs containing simulated annealing procedures. The code has been fixed to allow the generation of those files again.

Issue 2871

Fixes that affect portability

Fix error in AVX 512 detection code

The CMake detection code had a typo that could lead to wrong detection results.

Miscellaneous

Added warning with the use of GROMOS force fields

grompp now warns when a GROMOS force field is used. The GROMOS force fields have been parametrized with a physically incorrect multiple-time-stepping scheme for a twin-range cut-off. When used with a single-range cut-off, physical properties, such as the density, might be off from the intended values.

Issue 2884

Prevented internal build of FFTW with clang and AVX-512 SIMD

Prevented the internal build of FFTW with clang from attempting to configure FFTW to compile with AVX-512 support. That SIMD level is not supported by FFTW with the clang compiler, and compilation fails.

Issue 2892

Updated performance guide for recent Intel processors with AVX512 instruction support

Noted the tradeoffs between CPU frequency and SIMD throughput and advising users to prefer AVX2 over AVX512 in GPU-offload or highly parallel MPI cases.

Updated release notes for 2019.1

A fix made to GPU kernels in 2019.1 was thought to resolve Issue 2845 but further investigation suggests that the real cause is not yet known.