• Home

• Overview

• Publications

• Research

• Software

• NAMD

• VMD

• GPU Computing

• Lattice Microbes

• Atomic Resolution Brownian Dynamics

• MDFF

• QwikMD

• VND (Neuronal)

• Other

• Outreach

NAMD 3.0.2 Announcement

August 27, 2025

The Theoretical and Computational Biophysics Group at the University of Illinois is proud to announce the public release of a new version of NAMD, a parallel, object-oriented molecular dynamics code designed for high-performance simulation of large biomolecular systems. NAMD is distributed free of charge and includes source code. NAMD development is supported by the NIH National Institute of General Medical Sciences through grant numbers NIH P41-GM104601 and NIH R24-GM145965.

NAMD 3.0.2 is a point release of 3.0 including the following improvements:

- Fix recently discovered errors in the calculation of applied forces of the following collective variables (CVs) in Colvars: "eigenvector" with rotational fitting enabled, "distanceZ" with dynamic axis enabled, "orientationAngle", "orientation", and "coordNum" or "selfCoordNum" with pairlist. Previously, restraints along those CVs using Colvars may be ineffective, i.e., producing a trajectory inconsistent with the restraint, or causing otherwise unexpected results. Below are details:

   - "eigenvector" with rotational fitting: The gradients with respect to the fitting group (fit gradients), which are normally enabled for all CVs, were disabled by default for eigenvector. However, numerical tests show that these fit gradients cannot be ignored (see https://github.com/Colvars/colvars/pull/828). This issue does not affect calculations of eigenvector when the keyword rotateToReference is disabled, or of other CVs irrespective of rotational fitting.

   - "orientation" and "orientationAngle": Because the two quaternions q and -q represent identical orientations, Colvars uses the convention that the first component q0 >= 0, flipping the sign if needed from the calculated quaternion to simplify analysis of trajectories. However, the same correction was not applied to the gradients (https://github.com/Colvars/colvars/issues/827), resulting in inconsistently applied forces over a simulation. This issue *does not* affect the other rotational CVs (orientationProj, tilt, spinAngle, eulerPhi, eulerTheta, eulerPsi).

   - "coordNum" and "selfCoordNum" with pairlist: When using these two types of coordination number with pairlist enabled, a factor was missing when calculating the gradients of the switching function with respect to the atomic coordinates. Note that the relative errors between the wrong and correct gradients are linearly correlated with the pairlist tolerance. For example, a pairlist tolerance of 0.01 could make the gradients 1% smaller or larger than the correct results. This issue has been fixed (https://github.com/Colvars/colvars/pull/831) and is specific to the pairlist, i.e., does not occur when the parameter "tolerance" is 0 (default).

   - "distanceZ" with dynamic axis: When defining an optional "ref2" group, the reference axis is dynamically defined as the distance vector between "ref" and "ref2". The gradients with respect to the dynamic axis were incorrectly calculated and have now been fixed (https://github.com/Colvars/colvars/pull/838). The issue *does not* affect calculations when the axis is explicitly given using the keyword "axis", or when it is omitted to use the default axis (0,0,1).

The above bugs may affect your simulations only if you are using Colvars to apply biasing forces along the above CVs with the specified options. For that purpose, you should update to the latest bugfix release of NAMD (e.g. NAMD 3.0.2), which contains a version of Colvars updated for those bugfixes. Note: because VMD does not apply forces to atoms, calculations of the above variables from a pre-existing trajectory are still reliable even without the bugfix.

- Fix crash when using volumetric mapped-based CVs in Colvars.

- Fix bug affecting GPU-resident SMD whenever the selection size is greater than or equal to 1024 atoms.

- Fix bug affecting the use of GPUAtomMigration with MonteCarloPressure, in which the atom map was not being correctly updated.

- Fix builds for AMD HIP ROCm.

NAMD 3.0.1 is a point release of 3.0 including the following improvements:

- Fix bug affecting pressure tensor calculation when running GPU-resident mode with multiple time stepping enabled. Although the resulting error introduced was small, it was sufficient to change the area per lipid in membrane simulations.

- Fix crashes from builds using CUDA 12.5 and higher when running in GPU-resident mode.

- Fix bug in supporting PME with multiple GPU devices, triggered when running GPU-offload mode across more than 7 devices.

- Update to Colvars, fixing some bugs and behaviors for specialized Colvars features.

- Fix colinear lonepair calculation and contribution to pressure tensor.

- Fix crashes when trying to use GPUAtomMigration for multi-GPU simulation for HIP builds (e.g. OLCF Frontier).

NAMD 3.0 has many advantages over NAMD 2.14, including:

- GPU-resident mode providing very fast dynamics:

   - Achieves 2x or more speedup on single GPU versus GPU-offload mode, and 7x or more speedup for multi-GPU scaling on NVIDIA DGX-A100 versus GPU-offload mode

   - Supports single GPU and single-node multi-GPU scaling for tightly coupled GPUs (e.g. NVLink-enabled architectures like DGX)

   - Improved performance for small systems (e.g., 23.6k atom DHFR benchmark using hydrogen mass repartitioning with AMBER force field parameters can achieve over 1 microsecond/day simulation rates on NVIDIA A100)

   - Supports essential dynamics simulation protocols: constant energy, thermostats, barostats, rigid bond constraints, multiple time stepping, PME

   - Supports standard TIP3P and advanced 4-site water models (TIP4P and OPC)

   - Supports minimize and run directives through TCL scripting

   - Supports several advanced features: replica-exchange MD, alchemical free energy methods (FEP & TI), REST2, harmonic restraints, external electric field, SMD, Colvars, TCL forces, Monte Carlo barostat, group position restraints, constant pH MD

- GPU-accelerated alchemical free energy methods (FEP & TI) for both GPU-offload and GPU-resident modes

   - Supports van der Waals force switching for alchemy on both GPU modes

- HIP kernel improvements for better performance on AMD GPUs

- CPU-vectorization mode compatible with Intel and AMD CPU models that support AVX-512 instructions

   - Achieves speedup of up to 1.8x on Intel Xeon over AVX2 builds

   - Exhibits good scaling on TACC Frontera supercomputer

- Extend REST2 force field compatibility

- Update Colvars to version 2024-06-04

- Update user guide documentation for GPU-resident mode

- Fix several long-standing issues

   - Fix CPU memory leaks

   - Fix race conditions running GPU-resident mode with Colvars or TCL forces

   - Fix GPU-resident mode running constant pressure with harmonic restraints

   - Fix race conditions and NaNs running alchemy on GPU

   - Fix using advanced features together with GPU atom migration

   - Implement moving averages for GPU-resident reduction averages

   - Update config file parameters to use "GPU" rather than "CUDA" naming

   - Improve guard against running GPU-resident with unsupported features

- Update Psfgen to version 2.0

   - Fix compatibility with constant-pH MD

- Update to TCL version 8.6.13 for GPU-resident with Colvars and TCL forces

- Update to Charm++ 8.0.0, enabling:

   - Updated implementation of atomics, locks, and fences to use C++11/C11 versions where suitable

   - Improved performance, support, and fixes for UCX

   - Improved scaling on InfiniBand via ucx network layer

   - Improved multi-copy on POWER via pamilrts network layer

   - Clean exit with non-zero status for many failures

   - Improved scaling on HPE/Cray Slingshot 11 via ofi network layer

- Support for CUDA versions 9.1-12.x on Maxwell or newer GPUs

- Support for compatible AMD GPUs via HIP

Details at http://www.ks.uiuc.edu/Research/namd/3.0/features.html

Benchmarks at https://www.ks.uiuc.edu/Research/namd/benchmarks/

NAMD is available from http://www.ks.uiuc.edu/Research/namd/

For your convenience, NAMD has been ported to and will be installed on the machines at the NSF-sponsored national supercomputing centers. If you are planning substantial simulation work of an academic nature you should apply for these resources. Benchmarks for your proposal are available at http://www.ks.uiuc.edu/Research/namd/performance.html

The Theoretical and Computational Biophysics Group encourages NAMD users to be closely involved in the development process through reporting bugs, contributing fixes, periodical surveys and via other means. Questions or comments may be directed to namd@ks.uiuc.edu.

We are eager to hear from you, and thank you for using our software!