We want to increase our throughput performance issues from/to a Pi Server in our application that utilizes the PISDK. We ran a series of tests in our efforts to understanding throughput and performance issues using the PISDK.
PI SDK version --> 22.214.171.1248 PiSever -- > 3.4.385.59
- Our application reads and writes internal data to and from Pi Servers over ethernet. Our application is notified of changes to the point values of Pi Server data using the 'eventpipe' system via a 'pointlist' collection structure. This works as expected, ie. we write a new value and are notified of the change through events.
-In a controlled environment , using an even mix of digital, float16, float32, float64, int16, int32 and Strings totaling 7000 points, having changing values; we can write about 70k points in 65 seconds. However, The system takes about 4 minutes to finish the process of notifying us of all the changed values that were written to the pi server via the event processing methods, this is the largest bottle neck for us.
We still need to find ways to increase writing and retrieval rates and are looking at some other options.
Could implementing the Pi Buffering system help us?
Any other suggestions for increasing bandwidth to/from Pi?