New queue resource allocation policy for Gridpoint system


We are so glad to inform you after a prolonged working the replacement air conditioning system in the gridpoint computer room has been installed and been officially delivered this morning. Accordingly, all the nodes previously shut down will resume their service immediately. We are very sorry for the delay, as there has been too much construction works in the campus before the academic term commenced. The constructors kept on delaying their schedule in handling other works all at the same time. This is entirely beyond our control.

At the same time, we have reviewed the usage statistics to this special research oriented system, and would revise settings to some of the batch queues.

Queue Resource Allocation Policy (per program job)

(A) Queues available to all users

Queue name Processing time (hours) Max. no. of node(s) Max. no. of core(s) per node Memory per node
serial max. 168 1 1 32GB
parallel max. 10 4 8
fourday min. 10, max. 96 8 8
oneday-6c max. 24 2 12 48GB
ib-parallel max. 10 2 8 16GB

(B) Queues that require special application

Queue name Processing time (hours) Max. no. of nodes Max. no. of cores per node Memory per node Remarks
twoweek min. 96, max. 336 2 8 32GB Gb Ethernet
ib-oneweek max. 168 4 8 32GB InfiniBand
ib-special min.168, max. 2160 12 8 16GB

* As the queue resource allocation policy may be updated, you can get the most update details at Gridpoint system user guide

We will monitor closely on the new queue settings, and will fine-tune the settings when necessary.




HKU HPC/Grid team
September 24, 2012