How to plot bar plot using the following code for the given text file?

Question

I have the following output file (output.txt) for one of my HPC applications, in the following format

Data(D) Number_of_Processors(P) Process_per_node(ppn) mode time

2048 4 1 0 0.001220
2048 4 1 1 0.000858
32768 4 1 0 0.008137
32768 4 1 1 0.032052
262144 4 1 0 0.078899
262144 4 1 1 0.103439
2048 4 8 0 0.118370
2048 4 8 1 0.064003
32768 4 8 0 0.197745
32768 4 8 1 0.116132
262144 4 8 0 0.502012
262144 4 8 1 0.717104

I have only provided 12 lines here but they are 240 in number, as u can see there are only 3 types of DATA size. I am thorough in C++ but I don't know python. For my employee, I have to make a bar plot of them, like this-->

The Y-axis is the time, and the X-axis is the number of processes that is, P*ppn, and there are 3 plots for 3 data sizes separately, one for 2048, another for 32768, and the last one for 262144.

I have been provided with an example code that I need to modify and plot a graph that looks like the above one.

Here is the example code for graph-->

#!/usr/bin/env python
# coding: utf-8

# In[47]:


import pandas as pd
import seaborn as sns
import numpy as np
import matplotlib.pyplot as plt

sns.set()


# In[48]:


demo_input_format = pd.DataFrame.from_dict({
    "D": [],
    "P": [],
    "ppn": [],
    "mode": [],  # 1 --> optimized, 0 --> standard
    "time": [],
})


# In[49]:


for execution in range(10):
    for P in [4, 16]:
        for ppn in [1, 8]:
            for D in [16, 256, 2048]:
                # Change with the actual data
                demo_input_format = demo_input_format.append({
                    "D": D, "P": P, "ppn": ppn, "mode": 1, "time": np.random.rand() / 10
                }, ignore_index=True)
                demo_input_format = demo_input_format.append({
                    "D": D, "P": P, "ppn": ppn, "mode": 0, "time": np.random.rand()
                }, ignore_index=True)

demo_input_format["(P, ppn)"] = list(map(lambda x, y: ("(" + x + ", " + y + ")"), map(str, demo_input_format["P"]), map(str, demo_input_format["ppn"])))

print(demo_input_format)

# In[50]:


sns.catplot(x="(P, ppn)", y="time", data=demo_input_format, kind="box", col="D", hue="mode")
plt.show()

# In[ ]:

How can I modify this to take input from output.txt file and plot a bar plot like the above? Thank you. Please do help. :)

user_13138364 · Answer

First create the columns P * ppn and (P, ppn):

df['P * ppn'] = df.P * df.ppn
df['(P, ppn)'] = df.apply(lambda row: f'({row.P:d}, {row.ppn:d})', axis=1)

#          D  P  ppn  mode      time  P * ppn (P, ppn)
# 0     2048  4    1     0  0.001220        4   (4, 1)
# 1     2048  4    1     1  0.000858        4   (4, 1)
# 2    32768  4    1     0  0.008137        4   (4, 1)
# ...

Then melt() the dataframe into "long" form:

melted = df.melt(id_vars=['time', 'D', '(P, ppn)', 'mode'], value_vars='P * ppn')

#         time       D (P, ppn)  mode variable  value
# 0   0.001220    2048   (4, 1)     0  P * ppn      4
# 1   0.000858    2048   (4, 1)     1  P * ppn      4
# 2   0.008137   32768   (4, 1)     0  P * ppn      4
# ...

And finally catplot() the melted dataframe with the data size D along the grid columns:

sns.catplot(
    x='(P, ppn)',
    y='time',
    col='D',
    hue='mode',
    data=melted,
    kind='bar',
)

How to plot bar plot using the following code for the given text file?

1 Answers