User Profile

I could see that affinity between threads and cores and did wonder whether it breaking at 8 threads was somewhat coincidental to having 4 cores. As I cannot know in advance which of the days in the logs will be most demanding, I have no choice but to iterate in the safest possible way. I did a test to go through each day at a time and for each day, create 4 tasks to process 6 hours each. Once they are complete, I create the next 4 for the next day and so on until it completes. It works on this device but has no benefit. The overhead of all the task management seemed to outweigh a single threaded loop just churning its way through the sequence. So for now, I'm going to assume that the safety compromise and complexity of a fancy approach to speeding this up is not going to be worth it. It's an optimisation for the future but given your answer, I'm happy I'm not doing anything wrong and just need to live with the tradeoff which gives the safest result on all possible deployment targets.