Differences

This shows you the differences between two versions of the page.

--- start [2016/03/10 11:51] – ecalore
+++ start [2024/12/23 15:45] (current) – external edit 127.0.0.1
@@ Line 1: / Line 1: @@
-====== The COKA Cluster ======
+====== Decommissioning Notice (December 2024) ======
-intro...
+Dear Users,
+the 4 compute nodes funded by The Univeristy of Ferrara with the grant "Grandi Attrezzature 2015", integrated in the COKA cluster hosted by INFN-Ferrara, were installed in January 2016 and have been operating since February 2016.
+More than **200 users** have run on them, completing more than **2 millions jobs**, corresponding to approximately a total of **2M CPU core-hours** and **0.8 M GPU core-hours**.
+The cluster has supported the activities of several research groups for: simulations in fundamental physics, astrophysics, engineering, automatic speech recognition, artificial intelligence, machine learning, and many more; as well as the teaching activities of many courses of the University of Ferrara.
+{{ :img_usage.png?direct&800 }}
+After almost 10 years of continuous operation, i.e. approximately 3 times longer than what initially planned, due to aging hardware and lacking of updated software support, the effort to keep it operational is becoming too high.
+For these reasons, the COKA partition funded by The University of Ferrara will be **decommissioned on January 31st, 2025**, and user accounts will not be able to submit jobs on the associated compute nodes anymore.
+User accounts will still be able to **login on the front-end until the February 27th, 2025**, thus users are kindly requested to back up their data before this date.
+----
+Other pages and sections of this website are kept on-line for historical reasons only and are not valid, neither updated anymore.
+----
+====== The COKA Cluster (February 2016) ======
 {{ :img_20160212_163533.jpg?direct&300|}}
+The "Computing On Kepler Architectures" (COKA) is a computing cluster funded by the University of Ferrara with the support of INFN.
+The cluster has four computing nodes, each powered by **2x 8-core Intel Xeon CPUs** and **8x dual-GPU K80 NVIDIA** accelerator cards.
+Each node is tightly interconnected with two interfaces to an Infiniband switched network. The peak computing performance of the cluster is in the order of **100Tflops**.
+This system is intended to support applications development, optimization and test in the fields of computational physics, parallel algorithms, and numerical simulations carried out by research groups of University of Ferrara and INFN.