<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>Probably your best bet is to use QoS's to accomplish this. Be
advised that suspending jobs still leaves them in memory space.</p>
<p>-Paul Edmon-<br>
</p>
<div class="moz-cite-prefix">On 9/18/19 9:16 PM, Benjamin Wong
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:CA+-Ei1q+1Qp2vt3WmPi2n5n50hXzNm8-GjKEh3_EMUu+Hdaz+w@mail.gmail.com">
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
<div dir="ltr">
<div>Hello, <br>
</div>
<div><br>
</div>
<div>I plan to purchase a GPU machine with 8 GPUs which will be
shared between group A and group B. Group A is an existing
group with SLURM nodes. Group B has no SLURM nodes but will
have access to half of the resources on one SLURM node. I'm
trying to figure out how to get SLURM to implement the
policies I want below: <br>
</div>
<div>
<ul>
<li>If both groups are using the machine evenly, then I want
the resources to be split evenly. <br>
</li>
<li>If only group A is using the resources, then they will
consume all the resources and vice versa. <br>
</li>
<li>If group A is using all resources but group B begins
requesting resources, then group A will suspend half of
its work for group B to use resources. Vice versa
applies. </li>
</ul>
<div>What's the best way to implement this? Should I have two
halves of a machine in two different partitions? <br>
</div>
<div><br>
</div>
<div>Looking forward to hints,</div>
<div>Ben Wong<br>
</div>
</div>
</div>
</blockquote>
</body>
</html>