Previous
Hadamard product with numpy functions		 Parent
Peformances tests		 Outline Next
What hapened if I use python list instead of numpy array for naive implementation ?

6.7.5.3 : Hadamard product with our intrinsics pitch implementation

Now, let's write the hadamardIntrinsicsPitchPython.py file to test our implementation :

We need also to import several packages : 
1
2
3

import sys
import astericshpc
import hadamardpython
The function to initialise tables : 
1
2
3
4
5

def allocInitTable(nbElement):
	tab = astericshpc.allocTable(nbElement)
	for i in range(0, nbElement):
		tab[i] = float(i*32%17)
	return tab
The function to evaluate performances is built the same way such as the C++ one : 
1
2
3
4
5
6
7
8
9
10
11
12
13
14

def getTimeFunctionSize(nbRepetition, nbElement):
	tabX = allocInitTable(nbElement)
	tabY = allocInitTable(nbElement)
	tabRes = astericshpc.allocTable(nbElement)
	
	timeBegin = astericshpc.rdtsc()
	for i in range(0, nbRepetition):
		hadamardpython.hadamard(tabRes, tabX, tabY)
	
	timeEnd = astericshpc.rdtsc()
	elapsedTime = float(timeEnd - timeBegin)/float(nbRepetition)
	elapsedTimePerElement = elapsedTime/float(nbElement)
	print("nbElement =",nbElement,", elapsedTimePerElement =",elapsedTimePerElement,"cy/el",", elapsedTime =",elapsedTime,"cy")
	print(str(nbElement) + "\t" + str(elapsedTimePerElement) + "\t" + str(elapsedTime),file=sys.stderr)
Then, we have a function to make all the points with a list of sizes : 
1
2
3

def makeElapsedTimeValue(listSize, nbRepetition):
	for val in listSize:
		getTimeFunctionSize(nbRepetition, val)
Finally, we call the performances tests only if this script is executed as a main file and not if it is included by an other file : 
1
2
3
4
5
6
7
8
9
10
11

if __name__ == "__main__":
	listSize = [1000,
			1600,
			2000,
			2400,
			2664,
			3000,
			4000,
			5000,
			10000]
	makeElapsedTimeValue(listSize, 10000000)
The full hadamardIntrinsicsPitchPython.py file : 
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

'''
	Auteur : Pierre Aubert
	Mail : aubertp7@gmail.com
	Licence : CeCILL-C
'''

import sys
import astericshpc
import hadamardpython

def allocInitTable(nbElement):
	tab = astericshpc.allocTable(nbElement)
	for i in range(0, nbElement):
		tab[i] = float(i*32%17)
	return tab

def getTimeFunctionSize(nbRepetition, nbElement):
	tabX = allocInitTable(nbElement)
	tabY = allocInitTable(nbElement)
	tabRes = astericshpc.allocTable(nbElement)
	
	timeBegin = astericshpc.rdtsc()
	for i in range(0, nbRepetition):
		hadamardpython.hadamard(tabRes, tabX, tabY)
	
	timeEnd = astericshpc.rdtsc()
	elapsedTime = float(timeEnd - timeBegin)/float(nbRepetition)
	elapsedTimePerElement = elapsedTime/float(nbElement)
	print("nbElement =",nbElement,", elapsedTimePerElement =",elapsedTimePerElement,"cy/el",", elapsedTime =",elapsedTime,"cy")
	print(str(nbElement) + "\t" + str(elapsedTimePerElement) + "\t" + str(elapsedTime),file=sys.stderr)

def makeElapsedTimeValue(listSize, nbRepetition):
	for val in listSize:
		getTimeFunctionSize(nbRepetition, val)

if __name__ == "__main__":
	listSize = [1000,
			1600,
			2000,
			2400,
			2664,
			3000,
			4000,
			5000,
			10000]
	makeElapsedTimeValue(listSize, 10000000)
You can download it here.
Previous
Hadamard product with numpy functions		 Parent
Peformances tests		 Outline Next
What hapened if I use python list instead of numpy array for naive implementation ?