GPUGRID/en

Aus BC-Wiki
Zur Navigation springen Zur Suche springen

GPUGRID.net is a novel distributed supercomputing infrastructure made of many NVIDIA graphics cards joined together to deliver high-performance all-atom biomolecular simulations. The molecular simulations performed by our volunteers are some of the most common types performed by scientists in the field, but they are also some of the most computationally demanding and usually require a supercomputer.

Running GPUGRID on GPUs innovates volunteer computing by delivering supercomputing class applications on a cost effective infrastructure which will greatly impact the way biomedical research is performed.

Important for running the project:
Updating the NVidia graphic driver to the newest version (minimal required version is 310.90 for Windows/304.88 for Linux) - possibly even a new BOINC version is necessary when something significant was changed. Furthermore new and not too old graphic cards are useful, because the Long Runs are very cumputing intensive and probably cannot reach the deadline. All carda of the series GTX 5xx and above are capable, latest list you can read here: FAQ - Recommended GPUs for GPUGrid crunching

Badges

GPUGRID
Start 2007
End
Status
Admin GDF
Institution University of Pompeu Fabra
Country Spain
Area Biomedical
Apps
Win Molecular Dynamics for CPU and AMD GPUs 9.17 (mt)
Linux
Mac
64bit Quantum Chemistry 1.01 (mt) [linux]
Anaconda Python 3 Environment 1.06 (mt) [linux]
PS3
ATI
CUDA Short runs (2-3 hours on fastest card) 8.41/8.21 (cuda60)[win/linux]
ACEMD beta version 8.41/8.21 (cuda60) [win/linux]
Long runs (8-12 hours on fastest card) 8.41/8.21 (cuda60) [win/linux]
Intel
Android
RPi
NCI
System-Specs
VRAM SP DP
RAM 22MB
Runtime 2-11h
HDD MB
Traffic dl/ul MB
Deadline 5 days
Checkpoints