Hex Logo

Hex Server

General Information

This is a free service with no login or registration requirement. However, we reserve the right to reduce or block jobs from users or sites who appear to be exploiting this service unreasonably. If you give an e-mail address on the Docking Definition page, it will be used only to notify you of the status of your job, and possibly to inform you of new versions of Hex and/or HexServer.

In our opinion, docking is very much a visual activity, and it is therefore best done interactively on a computer with a good graphics card. If anyone wishes to perform more than a handful of docking runs, they are advised to download Hex and run it on their own machine. Because Hex is inherently a graphical program, all docking results are immediately available for viewing in a user-friendly way.

Disclaimer

Please remember that Hex and the Hex Server service are primarily research tools and are provided "as is" without any promise of support, training or even advice. However, any comments or suggestions for improvement would still be welcome and will be acted on if possible.

References

If you wish to cite Hex in a publication, please use one of the references here. Please also mention the Hex Server main page: http://hexserver.loria.fr.

Acknowledgements

Hex Server was originally implemented by Gary Macindoe in 2006. In 2009 it was adapted to run on the LORIA system (Dave Ritchie and Lazaros Mavridis) in order to use the GPU-version of Hex (Dave Ritchie and Vishwesh Venkatraman). The LORIA version of Hex Server runs on the MBI Cluster which is co-funded by INRIA and Region Lorraine: Contrat de Plan Etat-Region (CPER 2007-2013).

Technical Information

The user interface for Hex Server is written in PHP, and all communication between the PHP scripts and the compute cluster is done via a MySQL database. The cluster currently consists of 32 compute nodes running Centos 5.2. Each node has two quad-core Intel Xeon 2.5GHz processors, and 8 of the nodes also have two Nvidia Tesla C1060 GPUs. The cluster uses the OAR batch scheduling system to allocate nodes and to schedule jobs. One of the GPU nodes has been permanently reserved for Hex Server jobs, although the other GPU nodes may also be used if necessary. By default, each Hex jobs uses two GPUs simultaneously, although the user may request that the job be run on the CPU cores instead for comparison purposes. In our experience, there is very little benefit in using the CPU and GPU cores together, hence this option is not made available in the user interface.

Contact

Contact: dave.ritchie AT loria.fr (please make your own AT sign).