boinc/doc/affinity.html

<title>Compute Model: handling very long computations</title>

<h2>Compute model: handling very long computations</h2>

<p>
Some applications have very long (multi-month) computations
that generate large amounts of local state,
and therefore are most efficiently done on a single host.

<p>
To accommodate such applications, BOINC allows workunits and results
to be linked together into <b>sequences</b>.
Each workunit and result may have a predecessor and a successor.

<p>
If a result A has a predecessor B,
the scheduling server will dispatch A only
to the host that has handled (or is currently handling) B.
If result A times out,
then the scheduling server will dispatch B to any eligible host.