<p>12/11/2021<br />
Dear Users,</p>
<p>as you know Galileo100 nodes provide two types of nodes, thin and fat,<br />
characterised by a maximum memory per node of 375300MB and about 3TB<br />
respectively. To ease the use of the different nodes, and fully control the<br />
job environment, two new partitions have been defined in addition to the<br />
existing g100_usr_prod, which collects all thin and fat nodes :</p>
<p> * g100_usr_bmem: this partition collects only the nodes with a memory per<br />
node of about *3 TB*.<br />
* g100_usr_smem: this partition is composed only of thin nodes characterized<br />
by a memory per node of *375300 MB*.</p>
<p>A study on the performances of the thin and fat nodes is ongoing and a change<br />
of the billing on the g100_usr_bmem partition may be introduced in the<br />
future.</p>
<p>Please refer to the online documentation on<br /><a href="https://wiki.u-gov.it/confluence/display/SCAIUS/UG3.3%3A+GALILEO100+UserGuide#UG3.3:GALILEO100UserGuide-SubmittingparallelBatchJobs">https://wiki.u-gov.it/confluence/display/SCAIUS/UG3.3%3A+GALILEO100+User...</a><br />
for additional information.</p>
<p>Best regards,</p>
<p>HPC User Support @ CINECA</p>