1 / 5

Future Direction with NAMD

Future Direction with NAMD. David Hardy http://www.ks.uiuc.edu/Research /~ dhardy / NAIS: State-of-the-Art Algorithms for Molecular Dynamics. Short-term Outlook. Important for our software to support GPUs GPU acceleration is being incorporated into new supercomputers

berne
Download Presentation

Future Direction with NAMD

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Future Direction with NAMD David Hardy http://www.ks.uiuc.edu/Research/~dhardy/ NAIS: State-of-the-Art Algorithms for Molecular Dynamics

  2. Short-term Outlook • Important for our software to support GPUs • GPU acceleration is being incorporated into new supercomputers • GPU-accelerated desktop workstations to replace cluster computing • Easier maintenance • Improved power consumption • Both NAMD and GROMACS are in good shape for using GPU computing

  3. Trends in Computing Hardware • Computing hardware will continue to get “wider” • Moore’s Law is still in effect • Core clock speeds have plateaued • Memory hierarchies likely to get “deeper” • Memory bandwidth not increasing at the same rate as compute cores and FLOP/s • High performance software increasingly more difficult to develop

  4. NAMD and Performance • Could benefit from single core level performance improvements • SSE intrinsicswithin key computational kernels • Could make better use of GPUs • Asynchronous message-driven design is advantageous for large scale parallelization • Dynamic load balancing helps with hardware issues (e.g. system noise, recovery from failed nodes)

  5. Molecular Dynamics Challenges • Extending timescales of simulations • NAMD has high performance replica exchange, basis for other enhanced sampling methods • Improving force fields • NAMD supports leading polarizable force field efforts (Drude, FlucQ) • Plans to support AMOEBA polarizable force field • Better scaling methods • Multilevel summation method has promise

More Related