You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* @brief Represents a schedule for the Bulk Synchronous Parallel (BSP) model.
41
37
*
42
-
* The `BspSchedule` class manages the assignment of nodes to processors and supersteps within the BSP
43
-
* model. It serves as a core component for scheduling algorithms, providing mechanisms to:
38
+
* The `BspSchedule` class manages the assignment of nodes to processors and supersteps within the BSP model.
39
+
* It serves as a core component for scheduling algorithms, providing mechanisms to:
44
40
* - Store and retrieve node-to-processor and node-to-superstep assignments.
45
41
* - Validate schedules against precedence, memory, and node type constraints.
46
-
* - Compute costs associated with the schedule, such as work and communication costs.
42
+
* - Compute costs associated with the schedule.
47
43
* - Manipulate the schedule, including updating assignments and merging supersteps.
48
44
*
49
-
* This class is templated on `Graph_t`, which must satisfy the `computational_dag_concept`. It interacts closely with
50
-
* `BspInstance` to access problem-specific data and constraints.
45
+
* This class is templated on `Graph_t`, which must satisfy the `computational_dag_concept`.
46
+
* Moreover, the work and communication weights of the nodes must be of the same type in order to properly compute the cost.
47
+
*
48
+
* It interacts closely with `BspInstance` to access problem-specific data and constraints. In fact, a `BspSchedule` object is tied to a `BspInstance` object.
51
49
*
52
50
* @tparam Graph_t The type of the computational DAG, which must satisfy `is_computational_dag_v`.
0 commit comments