run_bench.py

import os
import statistics

import math

versions = ["sort", "rsqrt"]
nAgents = [8000, 16000, 32000, 64000, 128000]
initialWorldRadius = 64.0
initialQueryRadius = 8.0
nSteps = 500
reps = 1

benchmarks = {
    "const_size": [
        (a, nSteps if a != 128000 else 10, initialWorldRadius, initialQueryRadius)
        for a in nAgents
    ],
    # "const_pop": [
    #     (nAgents[0], nSteps, wRadius, initialQueryRadius / initialWorldRadius * wRadius)
    #     for wRadius in [x * (initialWorldRadius / nAgents[0]) for x in nAgents]
    # ],
    "const_density": [
        (a, nSteps if a != 128000 else 10, wRadius, initialQueryRadius)
        for (a, wRadius) in zip(
            nAgents,
            [
                math.sqrt(
                    x * (4 * initialWorldRadius * initialWorldRadius) / nAgents[0]
                )
                / 2
                for x in nAgents
            ],
        )
    ],
}

OpenSteerPath = ".\\build\\OpenSteerDemo.exe"


for version in versions:
    for benchmark, configs in benchmarks.items():
        csvFile = open(f"{version}_{benchmark}.csv", "w")
        csvFile.write("Agents;Steps;WorldRadius;QueryRadius;Time(s)\n")
        print(f"Running {benchmark} benchmark, {version} version")
        for config in configs:
            args = f"-no-gui -n {config[0]} -s {config[1]} { '-w '+str(config[2]) if config[2] > initialWorldRadius else ''} {'-rsqrt' if version == 'rsqrt' else ''} {'-sort' if version == 'sort' else ''}"
            print(f" Agents, Steps, WorldRadius, QueryRadius\n {config}")

            completed = False
            while not completed:
                data = []
                csvOut = ""
                for i in range(reps):
                    output = (
                        os.popen(OpenSteerPath + " " + args).read().splitlines()[-1]
                        + "\n"
                    )
                    data.append(float(output.split(";")[-1]))
                    csvOut += output
                    print(f"   {i+1}/{reps} Completed {data[i]}")
                avg = data[0]  # statistics.mean(data)
                median = data[0]  # statistics.median(data)
                variance = 0  # statistics.variance(data)
                print(f"Avg:{avg} Median:{median} Variance:{variance}")
                completed = True
                if not completed:
                    print("Variance > 5, repeating test")
                else:
                    csvFile.write(csvOut)  # .splitlines()[data.index(median)])

            csvFile.write("\n")


# Constant Population Benchmark

# Constant Density Benchmark