As of this head in the "graph_many" branch, we have achieved two things:
Single sim runs well (run ./generate_default.sh to see) and outputs a sensical tree
Multiple trials on a given sim generate sensical average # of forks per round (run ./long_sim -rounds=100 -miners=20 -lbp=50 -trials=100 for example)
Issue 1: On running this, we have a panic occuring as blocks of different heights are generated in a single round (happens maybe once every twenty runs, line 425). P2
We are now trying to have the sim run a series of repeated sims (with multiple trials each) to graph the interplay between number of miners and lbp. Specifically for different miner counts, we want to run repeated series of sims with different lbps to compare average number of forks (TODO: move results from map structure to a csv for easier reuse). However, we run into an issue: this takes too long to run.
Issue 2: Our attempt at parallelism there is largely unsuccessful, with func runTests() and func runAndAnalyze() the places where most of the issues lie. P0
Reproducible with: ./long_sim -test
(try with small values to get rapid certainty that things don't work).
As of this head in the "graph_many" branch, we have achieved two things:
Issue 1: On running this, we have a panic occuring as blocks of different heights are generated in a single round (happens maybe once every twenty runs, line 425). P2
We are now trying to have the sim run a series of repeated sims (with multiple trials each) to graph the interplay between number of miners and lbp. Specifically for different miner counts, we want to run repeated series of sims with different lbps to compare average number of forks (TODO: move results from map structure to a csv for easier reuse). However, we run into an issue: this takes too long to run.
Issue 2: Our attempt at parallelism there is largely unsuccessful, with
func runTests()andfunc runAndAnalyze()the places where most of the issues lie. P0Reproducible with: ./long_sim -test (try with small values to get rapid certainty that things don't work).