NAMD Wiki: Namd28Release

  You are encouraged to improve NamdWiki by adding content, correcting errors, or removing spam.

Issues with the 2.8 release of NAMD.

Please see the release notes at

For bugs fixed since the 2.8b3 release see Namd28b3Release and for all changes see

Coulomb switching for MARTINI force field is incorrect. Fixed 9/22/2011.

Tabulated nonbonded potential only works if PME is enabled. Fixed 3/9/2012.

The experimental memory-optimized build option with parallel I/O is documented at NamdMemoryReduction.

Compilation note: If using the tcl libraries available on the namd site, be sure to download the correct version. These have moved a bit since the 2.8 release notes were written. On OS X 10.5 I had to use the following tcl library to get compilation to work: This is tcl version 8.4. The version in the current libraries directory on the namd site is 8.5 - with this version, and following the compilation instructions in the release notes, the final make step failed.

On Lincoln at NCSA:


runs on GPU-accelerated Lincoln nodes. See the CUDA section of the release notes for details.

On the Ember Altix UV at NCSA:


On Ranger at TACC:


uses ibverbs.


can use 1way, 2way, or 4way processes per node with 15, 7, or 3 compute threads per process, and is again progressively slower because more cores are used for communication.

On Lonestar at TACC:




On Kraken at NICS:


still uses MPI but the binary is built with g++ for better performance.


will use 11 compute threads per process.

On KID (Keeneland Initial Delivery) GPU-accelerated cluster at NICS:


(Note that the KID filesystem is shared with Kraken, hence the KID_scripts directory.)