Parallelizing the Gold Hunt program
Looking at the previous profiler output, the
find_coins function is still the main bottleneck with ~19.5 seconds of cumulative time. Let's see how parallelization can help speed it up further.
Revisiting the gold field
Here is the gold field image from Chapter 7, Performance – Identifying Bottlenecks:
Let's quickly summarize what we already saw in Chapter 7, Performance – Identifying Bottlenecks:
find_coinsmethod is called for each of the small search circles shown in the figure. So, if there are 10 search circles,
find_coinswill be called 10 times, one after the other.
find_coinsmethod returns the coordinates ...