Amber Bug Fixes and Updates
Note: The individual bug fixes for Amber are described below in
case you want to see what they do, or apply patches by hand.
Most users are encouraged to use the update_amber script
(which is automatically run during the cmake step) to
handle patching tasks. This script automatically downloads and applies patches
on this site individually and keeps track of every patch that has been applied,
making patching and trouble-shooting easier.
Instructions for using update_amber can be found in the
Amber Reference Manual in Chapter 2, Section 4, or just go to your
amber20_src directory and type
./update_amber --help
If you wish to apply the listed Amber updates, run
cmake again.
Amber24 Updates (April, 2024)
- update.1:
Fix some cases where HREMD could hang with pmemd (parallel CPU version)
- update.2:
Adds compilation target
pmemd.decomp which
supports Thermodynamic Integration Atomic Decomposition
- update.3:
Fixes softcore potentials in
pmemd.cuda when gti_add_sc=25
Amber22 Updates (April, 2022)
- update.1:
Fix two (rare) problems, with 12-6-4 potentials and duplicate TI
- update.2:
Fix to the default values of soft-core potential parameters scalpha
and scbeta.
- update.3:
Fix the tishake2vite test outputs, so that the test cases now pass.
- update.4:
Fix a memory leak, speed up NMR calculations in
pmemd.cuda .
- update.5:
Updates scaling in the xray module of
pmemd.cuda .
Amber20 Updates (April, 2020)
- update.1:
Fix CpHMD problem in
pmemd.cuda.MPI .
- update.2:
Improve performance of
pmemd.cuda on Turing GPUs.
- update.3:
Update the way Plumed handles truncated octahedral boundary conditions.
- update.4:
Change RAMD code in
pmemd to avoid bias in
the direction selection.
- update.5:
Several fixes to
pmemd and pmemd.cuda related to TI and REMD
- update.6:
Update
pmemd to the latest version of plumed.
- update.7:
Fix MBAR printout in
pmemd.MPI
- update.8:
Fix CMAP memory allocation in
pmemd.cuda
- update.9:
Fixes a bug in LiGaMD & Pep-GaMD in
pmemd.cuda
- update.10:
Updates
pmemd to support gfortran10
- update.11:
Second update to
pmemd to support gfortran10
- update.12:
Add GPU support for the mcwat option in
pmemd
- update.13:
Fix code for igb=7,8 runs in
pmemd.cuda
built with recent GNU compilers.
Amber18 Updates (April, 2018)
- update.1:
Adds an error message to
pmemd when users run REMD
and the box sizes are not the same across all replicas.
- update.2:
Fix in
pmemd the neighbor pairs inside the pH and redox
potential dimensions on multidimensional REMD simulations.
- update.3:
This update fixes the following:
- Corrects REMD error trap for different box sizes in
pmemd to account for non-periodic systems.
- Bugfix in
pmemd to properly restart MultiD-REMD
simulations with two or more pH, redox potential or temperature
dimensions.
- update.4: Major update
to
pmemd.cuda , fixing various bugs. See the
GPU Patches page for a full description.
- update.5:
Adds support for MC and Berendsen barostats in order to run the benchmark
suite in the midpoint code. Additionally improves dynamic memory allocation.
- update.6:
Fix rare non-reproducibility issue in
pmemd.cuda .
- update.7:
Fix in
pmemd.cuda Fixes a memory usage issue when
using center of mass restraints.
- update.8:
Fix and extend the NFE capabilities in
pmemd .
- update.9:
Re-allow the functionality of backbone restraints/water restraints
when running GTI simulations in
pmemd.cuda .
- update.10:
Adds a new speed optimized mcwat code that allows for the simulation of
water exchange in buried binding sites in
pmemd.cuda and pmemd .
- update.11:
Minor tweak to update.10 to allow compilation using Intel
compilers.
- update.12:
Performance updates for
pmemd.cuda running on Volta or
Turing GPUs and using CUDA10.
- update.13:
Fixes for center of mass restraints in
pmemd.cuda . Also minor
variable declaration updates for reproduction of mcwat results.
- update.14:
Put a copy of AmberNetcdf.F90 into the
pmemd
source code tree, allowing it to be different than the one used in
AmberTools.
- update.15:
Update NEB capability in
pmemd.cuda.MPI .
- update.16:
Modifies the way softcore free energy calculations are done in
pmemd ; more
info.
- update.17:
Add files needed to implement the ff19SB protein force field.
Amber16 Updates (April, 2016)
- update.1:
Adds GPU support for including external electric fields.
- update.2:
Completes GPU support for including external electric fields.
- update.3:
Adds support for Intel Xeon and Intel Xeon Phi Optimizations.
- update.4:
New free energy routines in
pmemd ; see
nfe.pdf for full documentation.
- update.5:
Support Titan-XP in AMBER 16 pmemd.cuda.
- update.6:
Provides a 10 to 15 percent performance boost to GP102,104 and 107 based
gpus, which include GTX-1060,1070,1080, Titan-XP, Quadro P5000 and P6000.
Also enables execution on P100 based GPUs.
- update.7:
Updates printouts in the code (re SM60) to match those on the gpu page
of the web page.
- update.8:
Fix NetCDF restart writing for
pmemd when internal
coordinate axes are flipped.
- update.9:
Fixes and extensions to the "nfe" options in
pmemd .
- update.10:
Fix vulnerability that allowed Lennard-Jones 12-6-4 potentials to be
applied with incompatible features in
pmemd .
- update.11:
Fixes overwriting output in gamd.log when continuing a simulation.
pmemd .
- update.12:
Fixes dt and nstep passing in gpu force calculations (primarily used in
oscillating electric fields).
pmemd,sander .
- update.13:
Fix for
pmemd.cuda to trap simulations
with small periodic cells that can lead to incorrect pair lists.
- update.14:
Adds an error message to
pmemd when users run REMD
and the box sizes are not the same across all replicas.
- update.15:
Corrects REMD error trap for different box sizes in
pmemd
to account for non-periodic systems.
|
