Scheduling Adaptive Server tasks to engines in the SMP environment is similar to scheduling tasks in the single-CPU environment, as described in “Scheduling tasks to the engine”. The difference is that in the SMP environment:
Each engine has a run queue. Tasks have soft affinities to engines. When a task runs on an engine, it creates an affinity to the engine. If a task yields the engine and then is queued again, it tends to be queued on the same engine’s run queue.
Any engine can process the tasks in the global run queue (unless logical process management has been used to assign the task to a particular engine or set of engines).