TryRestarting
Member
I am working on a project that requires an additional remote rack (say about 150 i/o points) on a L73 redundancy system with about 900 i/o points. I come from a plantpax/IAB world where using only periodic tasks and observing the Null time, helps determine how much spare CPU.
For this non-plantpax system however, there is a continuous task with 50% System OH timeslice and 7 periodic tasks at 250ms and 500ms periods.
I see a few ways to check:
-ROUGH estimate solely based on memory free (51%)
-IAB as PlantPAx, for the system. This I have noticed requires a lot of fine tuning for redundancy. At one point the CPU loading was x3 that of simplex. Further the current logic and aois are much simpler than PlantPAx.
-Convert continuous task to a periodic task, dump on L73testbench, observe NULL time. The accuracy seems low (MSGs, # of connections HMI/SCADA, no redundancy)
-Use Studio Task Monitor for running system. I see user tasks at 84% with Comms, msg, redundancy, comms IO taking up the rest (16%). Further within the user task section the continuous task takes up about 60% "CPU", which I am assuming is not CPU but rather the % of User Task.
What is a quantitative way to approach this? The quick way I see, is adding roughly 15% i/o and logic adds roughly 15% to the continuous scan time. Which at this time is approximately 60ms and bumps it up to about 70ms, maybe a hair more with cross-loading. Is this reasonable?
For this non-plantpax system however, there is a continuous task with 50% System OH timeslice and 7 periodic tasks at 250ms and 500ms periods.
I see a few ways to check:
-ROUGH estimate solely based on memory free (51%)
-IAB as PlantPAx, for the system. This I have noticed requires a lot of fine tuning for redundancy. At one point the CPU loading was x3 that of simplex. Further the current logic and aois are much simpler than PlantPAx.
-Convert continuous task to a periodic task, dump on L73testbench, observe NULL time. The accuracy seems low (MSGs, # of connections HMI/SCADA, no redundancy)
-Use Studio Task Monitor for running system. I see user tasks at 84% with Comms, msg, redundancy, comms IO taking up the rest (16%). Further within the user task section the continuous task takes up about 60% "CPU", which I am assuming is not CPU but rather the % of User Task.
What is a quantitative way to approach this? The quick way I see, is adding roughly 15% i/o and logic adds roughly 15% to the continuous scan time. Which at this time is approximately 60ms and bumps it up to about 70ms, maybe a hair more with cross-loading. Is this reasonable?