[bb9b8f0] | 1 | import os |
---|
| 2 | import sys |
---|
| 3 | import time |
---|
[4912520] | 4 | import itertools |
---|
[bb9b8f0] | 5 | import matplotlib.pyplot as plt |
---|
| 6 | import matplotlib.ticker as ticks |
---|
| 7 | import math |
---|
| 8 | from scipy import stats as st |
---|
| 9 | import numpy as np |
---|
| 10 | from enum import Enum |
---|
| 11 | from statistics import median |
---|
| 12 | |
---|
| 13 | import matplotlib |
---|
| 14 | matplotlib.use("pgf") |
---|
| 15 | matplotlib.rcParams.update({ |
---|
| 16 | "pgf.texsystem": "pdflatex", |
---|
| 17 | 'font.family': 'serif', |
---|
| 18 | 'text.usetex': True, |
---|
| 19 | 'pgf.rcfonts': False, |
---|
[4912520] | 20 | 'font.size': 16 |
---|
[bb9b8f0] | 21 | }) |
---|
[4912520] | 22 | marker = itertools.cycle(('o', 's', 'D', 'x', 'p', '^', 'h', '*', 'v' )) |
---|
[bb9b8f0] | 23 | |
---|
| 24 | readfile = open(sys.argv[1], "r") |
---|
| 25 | |
---|
| 26 | machineName = "" |
---|
| 27 | |
---|
| 28 | if len(sys.argv) > 2: |
---|
| 29 | machineName = sys.argv[2] |
---|
| 30 | |
---|
| 31 | # first line has num times per experiment |
---|
| 32 | line = readfile.readline() |
---|
| 33 | numTimes = int(line) |
---|
| 34 | |
---|
| 35 | # second line has processor args |
---|
| 36 | line = readfile.readline() |
---|
| 37 | procs = [] |
---|
| 38 | for val in line.split(): |
---|
| 39 | procs.append(int(val)) |
---|
| 40 | |
---|
| 41 | # 3rd line has num locks args |
---|
| 42 | line = readfile.readline() |
---|
| 43 | locks = [] |
---|
| 44 | for val in line.split(): |
---|
| 45 | locks.append(int(val)) |
---|
| 46 | |
---|
| 47 | # 4th line has number of variants |
---|
| 48 | line = readfile.readline() |
---|
| 49 | names = line.split() |
---|
| 50 | numVariants = len(names) |
---|
| 51 | |
---|
| 52 | lines = (line.rstrip() for line in readfile) # All lines including the blank ones |
---|
| 53 | lines = (line for line in lines if line) # Non-blank lines |
---|
| 54 | |
---|
| 55 | nameSet = False |
---|
| 56 | currLocks = -1 # default val |
---|
| 57 | count = 0 |
---|
| 58 | procCount = 0 |
---|
| 59 | currVariant = 0 |
---|
| 60 | name = "Aggregate Lock" |
---|
| 61 | var_name = "" |
---|
[14e1053] | 62 | experiment_duration = 10.0 |
---|
[bb9b8f0] | 63 | sendData = [0.0 for j in range(numVariants)] |
---|
| 64 | data = [[0.0 for i in range(len(procs))] for j in range(numVariants)] |
---|
| 65 | bars = [[[0.0 for i in range(len(procs))],[0.0 for k in range(len(procs))]] for j in range(numVariants)] |
---|
| 66 | tempData = [0.0 for i in range(numTimes)] |
---|
| 67 | for idx, line in enumerate(lines): |
---|
| 68 | # print(line) |
---|
| 69 | |
---|
| 70 | if currLocks == -1: |
---|
| 71 | lineArr = line.split() |
---|
| 72 | currLocks = lineArr[-1] |
---|
| 73 | continue |
---|
| 74 | |
---|
| 75 | if line[0:5] == "cores": |
---|
| 76 | continue |
---|
| 77 | |
---|
| 78 | if not nameSet: |
---|
| 79 | nameSet = True |
---|
| 80 | continue |
---|
| 81 | |
---|
| 82 | lineArr = line.split() |
---|
[917e1fd] | 83 | tempData[count] = float(lineArr[-1]) / experiment_duration |
---|
[bb9b8f0] | 84 | count += 1 |
---|
| 85 | if count == numTimes: |
---|
| 86 | currMedian = median( tempData ) |
---|
| 87 | data[currVariant][procCount] = currMedian |
---|
| 88 | lower, upper = st.t.interval(0.95, numTimes - 1, loc=np.mean(tempData), scale=st.sem(tempData)) |
---|
[14e1053] | 89 | bars[currVariant][0][procCount] = max( 0, currMedian - lower ) |
---|
| 90 | bars[currVariant][1][procCount] = max( 0, upper - currMedian ) |
---|
[bb9b8f0] | 91 | count = 0 |
---|
| 92 | procCount += 1 |
---|
| 93 | |
---|
| 94 | if procCount == len(procs): |
---|
| 95 | procCount = 0 |
---|
| 96 | nameSet = False |
---|
| 97 | currVariant += 1 |
---|
| 98 | |
---|
| 99 | if currVariant == numVariants: |
---|
[5eb9327] | 100 | fig, ax = plt.subplots(layout='constrained') |
---|
[bb9b8f0] | 101 | plt.title(name + " Benchmark: " + str(currLocks) + " Locks") |
---|
[14e1053] | 102 | plt.ylabel("Throughput (critical section entries per second)") |
---|
[bb9b8f0] | 103 | plt.xlabel("Cores") |
---|
| 104 | for idx, arr in enumerate(data): |
---|
[4912520] | 105 | plt.errorbar( procs, arr, [bars[idx][0], bars[idx][1]], capsize=2, marker=next(marker) ) |
---|
[60f4919] | 106 | marker = itertools.cycle(('o', 's', 'D', 'x', 'p', '^', 'h', '*', 'v' )) |
---|
[bb9b8f0] | 107 | plt.yscale("log") |
---|
| 108 | plt.xticks(procs) |
---|
| 109 | ax.legend(names) |
---|
[4eebbcc] | 110 | # fig.savefig("plots/" + machineName + "Aggregate_Lock_" + str(currLocks) + ".png") |
---|
| 111 | plt.savefig("plots/" + machineName + "Aggregate_Lock_" + str(currLocks) + ".pgf") |
---|
[bb9b8f0] | 112 | fig.clf() |
---|
| 113 | |
---|
| 114 | # reset |
---|
| 115 | currLocks = -1 |
---|
| 116 | currVariant = 0 |
---|