Hello,
As discussed on IRC, I timed the inl2weight3 test against multiple
documents and queries , got the output written to a text file and then
wrote a python/matplotlib graphing script which drew a bar chart by reading
the text file. I have attached the bar chart , the python script and the
api_weight.cc file which contains the timed inl2weight3 test. The timing
was done using the RealTime::now() function.
-> I am adding comments to my proposal based on what I learnt from this
exercise and have thought of some modifications.
-> I had also sent in screenshots of profiling inl2 earlier by using
callgrind.
-> Please do let me know what you think.
-Regards
-Aarsh
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20140330/506915c4/attachment-0002.html>
-------------- next part --------------
import numpy as np
import matplotlib.pyplot as plt
x=[]
y=[]
fileInput=open("perf_inl2.txt","r")
i=0
for line in fileInput:
if i==0:
title=line
i = i + 1
continue
else:
fields=line.split(" ")
x.append(fields[0].strip())
y.append(float(fields[1].strip()))
i = i + 1
fileInput.close()
tot_elements = i - 1
x_axis = np.arange(tot_elements)
bar_width = 0.2
graph = plt.bar(x_axis,y,bar_width)
plt.xlabel("Documents")
plt.ylabel("Timings")
plt.title(title)
plt.xticks(x_axis+bar_width,x)
plt.show()
-------------- next part --------------
A non-text attachment was scrubbed...
Name: inl2_graph.png
Type: image/png
Size: 41142 bytes
Desc: not available
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20140330/506915c4/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: api_weight.cc
Type: application/octet-stream
Size: 25896 bytes
Desc: not available
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20140330/506915c4/attachment-0002.obj>