visualize.py

from collections import defaultdict
from matplotlib import pyplot as plt
from matplotlib.ticker import FixedLocator
import matplotlib.cm as cm
from numpy import std
import os

colors = ["#e41a1c", "#377eb8", "#4daf4a", "#984ea3", 
          "#ff7f00", "#ffff33", "#a65628", "#f781bf", 
          "#999999", "#FFFFFF"]
 
languages = {"pypy3":   "Pypy 2.4",
             "hhvm":    "HipHop VM 3.13",
             "java":    "Java OpenJDK 1.8",
             "lua":     "Lua 5.3",
             "luajit":  "Luajit 2.0",
             "php":     "Php 7.0",
             "python3": "Python 3.5",
             "ruby":    "Ruby 2.3",
             "js":      "NodeJS 6.2"}

benchmarks = {"binarytrees": "Binary tree",
              "fannkuchredux": "Fannkuchredux",
              "fasta": "Fasta",
              "nbody": "N-body",
              "spectralnorm": "Spectralnorm",
              "bubble": "Bubble sort",
              "fib": "Fibonacci"}

def decorate_plot(axis, xlabel, ylabel, title, xticks):
    plt.title(benchmarks[title])
    
    # set legend position outside to right
    box = ax.get_position()
    ax.set_position([box.x0, box.y0, box.width * 0.8, box.height])
    ax.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize='small')

    plt.ylabel(ylabel)
    plt.xlabel(xlabel)
    axis.xaxis.set_major_locator(FixedLocator(xticks))
    axis.grid(True)


data_file = "script/summary/all_measurements.csv"
imgs_folder = "imgs/"
if not os.path.isdir(imgs_folder):
    os.mkdir(imgs_folder)
    
try:
    with open(data_file, "r") as fp:
        dat = defaultdict(lambda: defaultdict(list))
        fp.readline()
        for line in fp:
            split_line = fp.readline().split(",")
            if len(split_line) < 9:
                continue
            
            bench, lang, _, n, _, cpu, mem, status, *load, elapsed = split_line
            dat[bench][lang].append((int(n), float(cpu), float(mem), float(elapsed)))

        for meas, mdata in dat.items():
            fig = plt.figure()
            ax = fig.add_subplot(111)   

            y_min = 1e10
            y_max = 0
            x_min = 1e10
            x_max = 0
            
            XY = []
            for idx, lang in enumerate(sorted(mdata.keys())):
                lst = mdata[lang]
                D = defaultdict(list)
                for tpl in lst:
                    D[tpl[0]].append(tpl[3])

                x = list(sorted(D.keys()))
                y = [sum(D[i]) / len(D[i]) for i in x]
                y_err = [std(D[i]) for i in x]

                x_min = min(x) if min(x) < x_min else x_min
                x_max = max(x) if max(x) > x_max else x_max
                y_min = min(y) if min(y) < y_min else y_min
                y_max = max(y) if max(y) > y_max else y_max

                XY.append((x, y, lang))
                ax.errorbar(x, y, yerr=y_err, color=colors[idx], label=languages[lang])

            x_min, x_max = x_min - (x_max - x_min) / 20, x_max + (x_max - x_min) / 20
            y_min, y_max = y_min - (y_max - y_min) / 20, y_max + (y_max - y_min) / 20
            
            ax.axis((x_min, x_max, y_min, y_max))
            decorate_plot(ax, "Problem size", "Execution time in seconds", meas, XY[0][0])
            plt.savefig(imgs_folder + meas + ".svg")
            
            plt.clf()
            ax = fig.add_subplot(111) 
            for i, (x, y, lang) in enumerate(XY):
                ax.semilogy(x, y, color=colors[i], label=languages[lang])
            
            decorate_plot(ax, "Problem size", "Execution time in seconds", meas, XY[0][0])
            plt.savefig(imgs_folder + meas + "_log.svg")
                

except EnvironmentError:
    print(data_file + " not found! Have you run benchmarks yet?")