Cpu Overload Behavior - Siemens SIMATIC ET 200SP System Manual

7.2

CPU overload behavior

Requirement
For the event scenarios considered in the following section, we assume that you have
assigned an OB to each event source and that these OBs have the same priority. The
second condition, in particular, is only for the sake of a simplified representation.
Principle of CPU overload behavior
An occurring event triggers the execution of the associated OB. Depending on the OB
priority and the current processor load, a time delay may occur before the OB is executed
when there is an overload. The same event can therefore occur once or several times before
the user program processes the OB belonging to the preceding event. The CPU treats such
a situation as follows: The operating system positions the events in the order of their
occurrence into the queue for their priority level.
To control temporary overload situations, you can limit the number of pending events that
are linked from the same source. The next event is discarded as soon as the maximum
number of pending start events of a specific cyclic interrupt OB, for example, is reached.
An overload occurs when events which originate from the same source occur faster than
they can be processed by the CPU.
More detailed information is available in the following sections.
Discarding similar events or handling them later
Below, the term "similar events" refers to events from a single source, such as triggers for a
specific cyclic interrupt OB.
The OB parameter "Events to be queued" is used to specify how many similar events the
operating system places in the associated queue and therefore post-processes. If this
parameter has the value 1, for example, exactly one event is stored temporarily.
Note
The post-processing of cyclic events is often undesirable, as this can lead to an overload
with OBs of the same or lower priority. Therefore, it is generally advantageous to discard
similar events and to react to the overload situation during the next scheduled OB
processing. If the value of the "Events to be queued" parameter is low, this ensures that an
overload situation is defused rather than aggravated.
If the maximum number of start events is reached in the queue for a cyclic interrupt OB
(Cyclic interrupt), for example, each additional start event is only counted and subsequently
discarded. During the next scheduled execution of the OB, the CPU provides the number of
discarded start events in the "Event_Count" input parameter (in the start information). You
can then react appropriately to the overload situation. The CPU then sets the counter for lost
events to zero.
If the CPU discards a start event for a cyclic interrupt OB the first time, its further behavior
depends on the OB parameter "Report event overflow into diagnostics buffer": If the check
box is selected, the CPU enters the event DW#16#0002:3507 once in the diagnostics buffer
for the overload situation at this event source. The CPU suppresses additional diagnostics
buffer entries of the event DW#16#0002:3507 that refer to this event source until all events
from this source have been post-processed.
Distributed I/O system
System Manual, 12/2015, A5E03576849-AF
Basics of program execution
7.2 CPU overload behavior
123