1 | #!/usr/bin/python3 |
---|
2 | """ |
---|
3 | Python Script to plot values obtained by the rmit.py script |
---|
4 | Runs a R.I.P.L. |
---|
5 | |
---|
6 | ./plot.py |
---|
7 | -t trials |
---|
8 | -o option:values |
---|
9 | """ |
---|
10 | |
---|
11 | import argparse |
---|
12 | import itertools |
---|
13 | import json |
---|
14 | import math |
---|
15 | import numpy |
---|
16 | import re |
---|
17 | import statistics |
---|
18 | import sys |
---|
19 | |
---|
20 | import matplotlib.pyplot as plt |
---|
21 | from matplotlib.ticker import EngFormatter |
---|
22 | |
---|
23 | class Field: |
---|
24 | def __init__(self, unit, _min, _log, _name=None): |
---|
25 | self.unit = unit |
---|
26 | self.min = _min |
---|
27 | self.log = _log |
---|
28 | self.name = _name |
---|
29 | |
---|
30 | field_names = { |
---|
31 | "ns per ops" : Field('ns' , 0, False), |
---|
32 | "Number of processors" : Field('' , 1, False), |
---|
33 | "Ops per procs" : Field('Ops' , 0, False), |
---|
34 | "Ops per threads" : Field('Ops' , 0, False), |
---|
35 | "ns per ops/procs" : Field('' , 0, False, _name = "Latency (ns $/$ (Processor $\\times$ Operation))" ), |
---|
36 | "Number of threads" : Field('' , 1, False), |
---|
37 | "Total Operations(ops)" : Field('Ops' , 0, False), |
---|
38 | "Ops/sec/procs" : Field('Ops' , 0, False), |
---|
39 | "Total blocks" : Field('Blocks', 0, False), |
---|
40 | "Ops per second" : Field('' , 0, False), |
---|
41 | "Cycle size (# thrds)" : Field('thrd' , 1, False), |
---|
42 | "Duration (ms)" : Field('ms' , 0, False), |
---|
43 | "Target QPS" : Field('' , 0, False), |
---|
44 | "Actual QPS" : Field('' , 0, False), |
---|
45 | "Average Read Latency" : Field('us' , 0, True), |
---|
46 | "Median Read Latency" : Field('us' , 0, True), |
---|
47 | "Tail Read Latency" : Field('us' , 0, True), |
---|
48 | "Average Update Latency": Field('us' , 0, True), |
---|
49 | "Median Update Latency" : Field('us' , 0, True), |
---|
50 | "Tail Update Latency" : Field('us' , 0, True), |
---|
51 | "Update Ratio" : Field('\%' , 0, False), |
---|
52 | } |
---|
53 | |
---|
54 | def plot(in_data, x, y, options): |
---|
55 | fig, ax = plt.subplots() |
---|
56 | colors = itertools.cycle(['#0095e3','#006cb4','#69df00','#0aa000','#fb0300','#e30002','#fd8f00','#ff7f00','#8f00d6','#4b009a','#ffff00','#b13f00']) |
---|
57 | series = {} # scatter data for each individual data point |
---|
58 | groups = {} # data points for x value |
---|
59 | |
---|
60 | print("Preparing Data") |
---|
61 | |
---|
62 | for entry in in_data: |
---|
63 | name = entry[0] |
---|
64 | if not name in series: |
---|
65 | series[name] = {'x':[], 'y':[]} |
---|
66 | |
---|
67 | if not name in groups: |
---|
68 | groups[name] = {} |
---|
69 | |
---|
70 | if x in entry[2] and y in entry[2]: |
---|
71 | xval = entry[2][x] |
---|
72 | yval = entry[2][y] |
---|
73 | series[name]['x'].append(xval) |
---|
74 | series[name]['y'].append(yval) |
---|
75 | |
---|
76 | if not xval in groups[name]: |
---|
77 | groups[name][xval] = [] |
---|
78 | |
---|
79 | groups[name][xval].append(yval) |
---|
80 | |
---|
81 | print("Preparing Lines") |
---|
82 | |
---|
83 | lines = {} # lines from groups with min, max, median, etc. |
---|
84 | for name, data in groups.items(): |
---|
85 | if not name in lines: |
---|
86 | lines[name] = { 'x': [], 'min':[], 'max':[], 'med':[], 'avg':[] } |
---|
87 | |
---|
88 | for xkey in sorted(data): |
---|
89 | ys = data[xkey] |
---|
90 | lines[name]['x'] .append(xkey) |
---|
91 | lines[name]['min'].append(min(ys)) |
---|
92 | lines[name]['max'].append(max(ys)) |
---|
93 | lines[name]['med'].append(statistics.median(ys)) |
---|
94 | lines[name]['avg'].append(statistics.mean(ys)) |
---|
95 | |
---|
96 | print("Making Plots") |
---|
97 | |
---|
98 | for name, data in sorted(series.items()): |
---|
99 | _col = next(colors) |
---|
100 | plt.scatter(data['x'], data['y'], color=_col, label=name, marker='x') |
---|
101 | plt.plot(lines[name]['x'], lines[name]['min'], '--', color=_col) |
---|
102 | plt.plot(lines[name]['x'], lines[name]['max'], '--', color=_col) |
---|
103 | plt.plot(lines[name]['x'], lines[name]['med'], '-', color=_col) |
---|
104 | |
---|
105 | print("Calculating Extremums") |
---|
106 | |
---|
107 | mx = max([max(s['x']) for s in series.values()]) |
---|
108 | my = max([max(s['y']) for s in series.values()]) |
---|
109 | |
---|
110 | print("Finishing Plots") |
---|
111 | |
---|
112 | plt.ylabel(field_names[y].name if field_names[y].name else y) |
---|
113 | # plt.xticks(range(1, math.ceil(mx) + 1)) |
---|
114 | plt.xlabel(field_names[x].name if field_names[x].name else x) |
---|
115 | plt.grid(b = True) |
---|
116 | ax.xaxis.set_major_formatter( EngFormatter(unit=field_names[x].unit) ) |
---|
117 | if options.logx: |
---|
118 | ax.set_xscale('log') |
---|
119 | elif field_names[x].log: |
---|
120 | ax.set_xscale('log') |
---|
121 | else: |
---|
122 | plt.xlim(field_names[x].min, mx + 0.25) |
---|
123 | |
---|
124 | ax.yaxis.set_major_formatter( EngFormatter(unit=field_names[y].unit) ) |
---|
125 | if options.logy: |
---|
126 | ax.set_yscale('log') |
---|
127 | elif field_names[y].log: |
---|
128 | ax.set_yscale('log') |
---|
129 | else: |
---|
130 | plt.ylim(field_names[y].min, options.MaxY if options.MaxY else my*1.2) |
---|
131 | |
---|
132 | plt.legend(loc='upper left') |
---|
133 | |
---|
134 | print("Results Ready") |
---|
135 | if options.out: |
---|
136 | plt.savefig(options.out, bbox_inches='tight') |
---|
137 | else: |
---|
138 | plt.show() |
---|
139 | |
---|
140 | |
---|
141 | if __name__ == "__main__": |
---|
142 | # ================================================================================ |
---|
143 | # parse command line arguments |
---|
144 | parser = argparse.ArgumentParser(description='Python Script to draw R.M.I.T. results') |
---|
145 | parser.add_argument('-f', '--file', nargs='?', type=argparse.FileType('r'), default=sys.stdin, help="Input file") |
---|
146 | parser.add_argument('-o', '--out', nargs='?', type=str, default=None, help="Output file") |
---|
147 | parser.add_argument('-y', nargs='?', type=str, default="", help="Which field to use as the Y axis") |
---|
148 | parser.add_argument('-x', nargs='?', type=str, default="", help="Which field to use as the X axis") |
---|
149 | parser.add_argument('--logx', action='store_true', help="if set, makes the x-axis logscale") |
---|
150 | parser.add_argument('--logy', action='store_true', help="if set, makes the y-axis logscale") |
---|
151 | parser.add_argument('--MaxY', nargs='?', type=int, help="maximum value of the y-axis") |
---|
152 | |
---|
153 | options = parser.parse_args() |
---|
154 | |
---|
155 | # ================================================================================ |
---|
156 | # load data |
---|
157 | try : |
---|
158 | data = json.load(options.file) |
---|
159 | except : |
---|
160 | print('ERROR: could not read input', file=sys.stderr) |
---|
161 | parser.print_help(sys.stderr) |
---|
162 | sys.exit(1) |
---|
163 | |
---|
164 | # ================================================================================ |
---|
165 | # identify the keys |
---|
166 | |
---|
167 | series = set() |
---|
168 | fields = set() |
---|
169 | |
---|
170 | for entry in data: |
---|
171 | series.add(entry[0]) |
---|
172 | for label in entry[2].keys(): |
---|
173 | fields.add(label) |
---|
174 | |
---|
175 | if not options.out : |
---|
176 | print(series) |
---|
177 | print("fields: ", ' '.join(fields)) |
---|
178 | |
---|
179 | wantx = "Number of processors" |
---|
180 | wanty = "ns per ops" |
---|
181 | |
---|
182 | if options.x: |
---|
183 | if options.x in field_names.keys(): |
---|
184 | wantx = options.x |
---|
185 | else: |
---|
186 | print("Could not find X key '{}', defaulting to '{}'".format(options.x, wantx)) |
---|
187 | |
---|
188 | if options.y: |
---|
189 | if options.y in field_names.keys(): |
---|
190 | wanty = options.y |
---|
191 | else: |
---|
192 | print("Could not find Y key '{}', defaulting to '{}'".format(options.y, wanty)) |
---|
193 | |
---|
194 | |
---|
195 | plot(data, wantx, wanty, options) |
---|