basicanalysis/modelfactors.py at master · bsc-performance-tools/basicanalysis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
#!/usr/bin/env python3

"""modelfactors.py Generates performance metrics from a set of Paraver traces."""

from __future__ import print_function, division
import subprocess
import plots
from utils import parse_arguments, check_installation
from tracemetadata import get_traces_from_args
from rawdata import gather_raw_data, print_raw_data_table, print_raw_data_csv
from simplemetrics import compute_model_factors, print_mod_factors_csv, print_efficiency_table, \
    print_mod_factors_table, read_mod_factors_csv, print_other_metrics_table, \
    print_other_metrics_csv, plots_efficiency_table_matplot, plots_modelfactors_matplot, plots_speedup_matplot

import hybridmetrics
import os

# error import variables
error_import_pandas = False
error_import_seaborn = False
error_import_matplotlib = False
error_import_scipy = False
error_import_numpy = False

try:
    import pandas as pd
except ImportError:
    error_import_pandas = True

try:
    import seaborn as sns
except ImportError:
    error_import_seaborn = True

try:
    import matplotlib.pyplot as plt
except ImportError:
    error_import_matplotlib = True

try:
    import scipy.optimize
except ImportError:
    error_import_scipy = True


try:
    import numpy
except ImportError:
    error_import_numpy = True


if __name__ == "__main__":
    """Main control flow.
    Currently the script only accepts one parameter, which is a list of traces
    that are processed. This can be a regex with wild cards and only valid trace
    files are kept at the end.
    """
    # Parse command line arguments
    cmdl_args = parse_arguments()

    # Check if paramedir and Dimemas are in the path
    check_installation(cmdl_args)

    # Process the trace list
    trace_list, trace_processes, trace_tasks, trace_threads, trace_task_per_node, trace_mode = \
        get_traces_from_args(cmdl_args)

    # To validate the metric type
    trace_metrics = 0
    # print(trace_mode)
    for trace in trace_list:
        if trace_mode[trace][0:len("Detailed+MPI+")] == "Detailed+MPI+":
            trace_metrics += 1
    # Analyze the traces and gather the raw input data
    raw_data, list_mpi_procs_count = gather_raw_data(trace_list, trace_processes, trace_task_per_node,
                                                     trace_mode,trace_tasks, trace_threads, cmdl_args)
    print_raw_data_csv(raw_data, trace_list, trace_processes)

    # Compute the model factors and print them

    if cmdl_args.metrics == 'hybrid' and trace_metrics > 0:
        mod_factors, mod_factors_scale_plus_io, hybrid_factors, other_metrics, device_factors, host_factors = \
                hybridmetrics.compute_model_factors(raw_data, trace_list, trace_processes,
                                                    trace_mode, list_mpi_procs_count, cmdl_args)
        hybridmetrics.print_other_metrics_table(other_metrics, trace_list, trace_processes, trace_tasks,
                                                trace_threads, trace_mode)
        hybridmetrics.print_other_metrics_csv(other_metrics, trace_list, trace_processes)


        if (cmdl_args.pop_model_to_apply == 'talp') and (trace_mode[trace_list[0]] == "Detailed+MPI+CUDA"):
            hybridmetrics.print_talp_metrics_csv(device_factors,host_factors, trace_list, trace_processes,raw_data)
            hybridmetrics.print_mod_factors_table_talp(mod_factors, other_metrics, mod_factors_scale_plus_io, hybrid_factors, device_factors, host_factors,
                                              trace_list, trace_processes, trace_tasks, trace_threads, trace_mode, raw_data)
            hybridmetrics.plots_talp_efficiency_table_matplot(trace_list, trace_processes, trace_tasks, trace_threads, trace_mode,raw_data, cmdl_args)
        else:
            hybridmetrics.print_mod_factors_table(mod_factors, other_metrics, mod_factors_scale_plus_io, hybrid_factors, device_factors,
                                              trace_list, trace_processes, trace_tasks, trace_threads, trace_mode, raw_data)

        hybridmetrics.print_mod_factors_csv(mod_factors, hybrid_factors, trace_list, trace_processes)
        hybridmetrics.print_efficiency_table(mod_factors, hybrid_factors, trace_list, trace_processes, trace_tasks,
                                             trace_threads,trace_mode)
        # Plotting efficiency table with matplotlib
        error_plot_table = False
        if error_import_numpy or error_import_pandas or error_import_matplotlib or error_import_seaborn:
            print('Numpy/Pandas/Matplotlib/Seaborn modules not available. '
                  'Skipping efficiency table plotting with python.')
            if len(trace_list) > 1:
                out_ver_gnuplot = subprocess.check_output(["gnuplot", "--version"])
                if 'gnuplot 5.' not in str(out_ver_gnuplot):
                    print('It requires gnuplot version 5.0 or higher. '
                          'Skipping efficiency table and lineal plotting with gnuplot.')
                else:
                    try:
                        output_gnuplot_g = subprocess.check_output(["gnuplot", "efficiency_table_global.gp"])
                        output_gnuplot_h = subprocess.check_output(["gnuplot", "efficiency_table_hybrid.gp"])
                    except:
                        print(output_gnuplot_g)
                        print(output_gnuplot_h)

            error_plot_table = True

        if not error_plot_table:
            hybridmetrics.plots_efficiency_table_matplot(trace_list, trace_processes, trace_tasks,
                                                         trace_threads, trace_mode, cmdl_args)

            if len(trace_list) > 1:
                hybridmetrics.plots_modelfactors_matplot(trace_list, trace_processes, trace_tasks,
                                                         trace_threads, trace_mode, cmdl_args)
                hybridmetrics.plots_speedup_matplot(trace_list, trace_processes, trace_tasks,
                                                    trace_threads, trace_mode, cmdl_args)

        # Plotting if SciPy and NumPy are installed.
        error_plot_lineal = False
        if error_import_numpy or error_import_scipy:
            print('Scipy/NumPy module not available. Skipping lineal plotting.')
            error_plot_lineal = True

        if not error_plot_lineal:
            if len(trace_list) > 1:
                plots.plot_hybrid_metrics(mod_factors, hybrid_factors, trace_list, trace_processes,
                                          trace_tasks, trace_threads, trace_mode, cmdl_args)

        if len(trace_list) == 1:
            subprocess.check_output(["rm", "efficiency_table_global.gp"])
            subprocess.check_output(["rm", "efficiency_table_hybrid.gp"])

    elif cmdl_args.metrics == 'simple' or trace_metrics == 0:
        mod_factors, mod_factors_scale_plus_io, other_metrics = compute_model_factors(raw_data, trace_list,
                                                                                      trace_processes, trace_mode,
                                                                                      list_mpi_procs_count, cmdl_args)
        print_other_metrics_table(other_metrics, trace_list, trace_processes)
        print_other_metrics_csv(other_metrics, trace_list, trace_processes)
        print_mod_factors_table(mod_factors, other_metrics, mod_factors_scale_plus_io, trace_list, trace_processes, trace_mode)
        print_mod_factors_csv(mod_factors, trace_list, trace_processes)
        print_efficiency_table(mod_factors, trace_list, trace_processes)

        # Plotting efficiency table with matplotlib
        error_plot_table = False
        if error_import_numpy or error_import_pandas or error_import_matplotlib or error_import_seaborn:
            print('Numpy/Pandas/Matplotlib/Seaborn modules not available. '
                  'Skipping efficiency table plotting with python.')
            if len(trace_list) > 1:
                out_ver_gnuplot = subprocess.check_output(["gnuplot", "--version"])
                if 'gnuplot 5.' not in str(out_ver_gnuplot):
                    print('It requires gnuplot version 5.0 or higher. '
                          'Skipping efficiency table and lineal plotting with gnuplot.')
                else:
                    try:
                        output_gnuplot = subprocess.check_output(["gnuplot", "efficiency_table.gp"])
                    except:
                        print(output_gnuplot)
            error_plot_table = True

        if not error_plot_table:
            plots_efficiency_table_matplot(trace_list, trace_processes, trace_tasks, trace_threads, cmdl_args)
            if len(trace_list) > 1:
                plots_modelfactors_matplot(trace_list, trace_mode, trace_processes, trace_tasks, trace_threads,
                                           cmdl_args)

        # Plotting if SciPy and NumPy are installed.
        error_plot_lineal = False
        if error_import_numpy or error_import_scipy:
            print('Scipy/NumPy module not available. Skipping lineal plotting.')
            error_plot_lineal = True
        # sys.exit(1)
        if not error_plot_lineal:
            if len(trace_list) > 1:
                plots.plot_simple_metrics(mod_factors, trace_list, trace_processes, trace_mode, cmdl_args)
                plots_speedup_matplot(trace_list, trace_processes, trace_tasks, trace_threads, cmdl_args)
        if len(trace_list) == 1:
            subprocess.check_output(["rm", "efficiency_table.gp"])