Slurm Workload Manager

The Slurm Workload Manager, formerly known as Simple Linux Utility for Resource Management (SLURM), or simply Slurm, is a <a href="/facts/Free_and_open-source/1cHtjLCB">free and open-source</a> <a href="/facts/Job_scheduler/1BPxX2Lw">job scheduler</a> for <a href="/facts/Linux/9FvCKSTq">Linux</a> and <a href="/facts/Unix-like/YAKrAxzd">Unix-like</a> <a href="/facts/Kernel_(operating_system)/uEj4U3Rz">kernels</a>, used by many of the world's <a href="/facts/Supercomputer/HT9NwGAN">supercomputers</a> and <a href="/facts/Computer_cluster/mT7S1BcX">computer clusters</a>.
It provides three key functions:

<ul><li>allocating exclusive and/or non-exclusive access to resources (computer nodes) to users for some duration of time so they can perform work,</li>
<li>providing a framework for starting, executing, and monitoring work, typically a parallel job such as <a href="/facts/Message_Passing_Interface/BEuBgDpm">Message Passing Interface</a> (MPI) on a set of allocated nodes, and</li>
<li>arbitrating contention for resources by managing a queue of pending jobs.</li></ul>
Slurm is the workload manager on about 60% of the <a href="/facts/TOP500/rz8zbyss">TOP500</a> supercomputers. 
Slurm uses a <a href="/facts/Best-fit_bin_packing/JI3prfPs">best fit algorithm</a> based on <a href="/facts/Hilbert_curve_scheduling/UAx97BLg">Hilbert curve scheduling</a> or <a href="/facts/Fat_tree/AyFyz89J">fat tree</a> network topology in order to optimize locality of task assignments on parallel computers.

Slurm Workload Manager open-in-new

Slurm Workload Manager