However, insert speed here is compromised a little (needs to remain in order).
Sam Holder had another good idea in the comments - keep a parallel array that is always sorted, this lets you lop numbers off the top or bottom to find new minimums and maximums easier. Then finding them can be trivial sometimes, but when the last value considered holds the min and max, the entire list needs to be scanned to establish the new limits. Instead of keeping the min/max values x_min, x_max keep instead the index of where they are located in the x array with i_min and i_max. Then when you reach an 'invalidation' scenario, you generally only need to look through the latest 100ms of data or a quick pass through the min and max of the other 9 provided a great idea to save on finding the min and max values if they are invalidated:
That is, keep the running min, max, sum and count on those 10 chunks. It should be large enough to cover the worst case scenario for message size per second.ĭepending on the usage scenario one other thing to do would be to run the algorithm above but in 10 x 100ms chunks rather than 1 x 1000ms piece.
FIND THE AVERAGE FOR MINUTES AND SECONDS EXCEL FOR MAC CODE
This code could be performed on every buffer insert also, or wherever made sense.īest structure for this kind of work is a circular buffer, to avoid memory allocations and GC getting in the way. If the values are below min or above max, you'll need to loop through the rest of the array and recalculate it.ĭo step two once a second or so also so that the buffer doesn't get too full. In this scenario, you have average, min, and max updated efficiently. If the values are below max, you can keep your max. If the values are all above min you can keep your min. When a request for average, min, or max is made, loop through from the back of the buffer and start removing values older than one second. Here is an algorithm that will somewhat work to save efficiency in certain cases:Īs events come in, buffer them completely, and calculate a running sum, count, min, max (trivial).
0 kommentar(er)
