From: JC Gumbart (gumbart_at_physics.gatech.edu)
Date: Tue Jan 05 2016 - 16:16:03 CST
Are there any tricks or caveats to the improved GPU performance? I just compared 2.10 and 2.11 and actually get slightly worse performance with 2.11! I’m running a 40k-atom system on a Tesla K20c card. Could it be a driver issue?
Thanks!
JC
> On Dec 22, 2015, at 11:38 AM, Jim Phillips <jim_at_ks.uiuc.edu> wrote:
>
> Dear NAMD users,
>
> NAMD 2.11 is officially released and available for download.
>
> This release fixes a bug that resulted in incorrect VDW tail correction values for systems larger than about 11,000 atoms due to an integer overflow. This feature is off by default, and the magnitude of the incorrect energy and pressure correction factor was always smaller than the correct value and declined with increasing system size. Again, this bug only affected simulations with LJcorrection explicitly enabled.
>
> Thank you all for your efforts and support!
>
> Jim
>
>
> +--------------------------------------------------------------------+
> | |
> | NAMD 2.11 Release Announcement |
> | |
> +--------------------------------------------------------------------+
>
> December 22, 2015
>
> 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.
>
> NAMD 2.11 has many advantages over NAMD 2.10, among these are:
>
> - GPU-accelerated simulations up to twice as fast as NAMD 2.10
>
> - Enhanced Tcl scripting of collective variables and "cv" command
>
> - Collective variables module improvements including to histogram bias
>
> - TclForces query total net forces for atom groups
>
> - Replica-exchange multiplexing (fewer partitions than replicas)
>
> - Tcl scripting multiple in-memory checkpoints
>
> - Improved Tcl scripting multi-copy interface documentation
>
> - Tcl scripting asynchronous multi-copy remote in-memory checkpoints
>
> - Tcl scripting asynchronous multi-copy remote trajectory output
>
> - Tcl scripting asynchronous multi-copy remote script evaluation
>
> - Tcl scripting asynchronous multi-copy central work queue support
>
> - Tcl scripting asynchronous multi-copy workflow-style programming
>
> - Improved minimization for Drude force field and rigid bonds
>
> - Fix long-range LJ correction for systems larger than 11,000 atoms
>
> - Improved long-range LJ correction with VDW force switching
>
> - Improved alchemical calculations with VDW force switching
>
> - Pressure calculation with fixed atoms on GPU works as on CPU
>
> - Improved scaling for GPU-accelerated particle-mesh Ewald calculation
>
> - Improved scaling for GPU-accelerated simulations
>
> - Improved scaling for multi-threaded "smp" builds
>
> - Prevent running smp builds with one thread per process
>
> - Support trajectory files larger than 2GB on Windows
>
> - NVIDIA CUDA GPU-acceleration binaries for Mac OS X
>
> - Improved Intel Xeon Phi coprocessor support
>
> - Update to Charm++ 6.7.0 with improved multi-copy support
>
> - Ignore ioformat statement in CHARMM topology and parameter files
>
> - Psfgen improvements including long resids and insertion codes
>
> - Psfgen package available in NAMD Tcl interpreter
>
>
> Details at http://www.ks.uiuc.edu/Research/namd/2.11/features.html
>
> 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_at_ks.uiuc.edu.
>
> We are eager to hear from you, and thank you for using our software!
>
This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:21:42 CST