Skip to content

GitLab

Commit f4a8b977 authored by iker_martin's avatar iker_martin
Browse files

Cambios sobre recopiladores para obtener tiempos de creacion de procesos -... 

Cambios sobre recopiladores para obtener tiempos de creacion de procesos - Analizador (ipynb) tiene graficas automatizadas
parent 880a6260
Show whitespace changes
Inline Side-by-side
Analysis/MallTimes.py
View file @ f4a8b977
...@@ -60,31 +60,38 @@ def read_file(f, dataA, dataB, it): ...@@ -60,31 +60,38 @@ def read_file(f, dataA, dataB, it):
if lineS[0] == "Config": # CONFIG LINE if lineS[0] == "Config": # CONFIG LINE
recording = True recording = True
it += 1 it += 1
dataA.append([None]*9) dataA.append([None]*13)
dataB.append([None]*13) dataB.append([None]*15)
resizes = int(lineS[2].split('=')[1].split(',')[0]) #resizes = int(lineS[2].split('=')[1].split(',')[0])
resizes = 2
compute_tam = int(lineS[3].split('=')[1].split(',')[0]) compute_tam = int(lineS[3].split('=')[1].split(',')[0])
comm_tam = int(lineS[4].split('=')[1].split(',')[0]) comm_tam = int(lineS[4].split('=')[1].split(',')[0])
sdr = int(lineS[5].split('=')[1].split(',')[0]) sdr = int(lineS[5].split('=')[1].split(',')[0])
adr = int(lineS[6].split('=')[1].split(',')[0]) #TODO Que lo tome como porcentaje adr = int(lineS[6].split('=')[1].split(',')[0]) #TODO Que lo tome como porcentaje
css = int(lineS[8].split('=')[1].split(',')[0])
cst = int(lineS[9].split('=')[1].split(',')[0])
# TODO Que obtenga Aib # TODO Que obtenga Aib
time = float(lineS[8].split('=')[1]) time = float(lineS[10].split('=')[1])
dataB[it][0] = sdr dataB[it][0] = sdr
dataB[it][1] = adr dataB[it][1] = adr
dataB[it][4] = "" dataB[it][4] = ""
dataB[it][5] = compute_tam dataB[it][5] = compute_tam
dataB[it][6] = comm_tam dataB[it][6] = comm_tam
dataB[it][7] = time dataB[it][7] = cst
dataB[it][8] = "" dataB[it][8] = css
dataB[it][9] = time
dataB[it][10] = ""
dataA[it][0] = sdr dataA[it][0] = sdr
dataA[it][1] = adr dataA[it][1] = adr
dataA[it][3] = "" dataA[it][5] = ""
dataA[it][4] = compute_tam dataA[it][6] = compute_tam
dataA[it][5] = comm_tam dataA[it][7] = comm_tam
dataA[it][6] = time dataA[it][8] = cst
dataA[it][7] = "" dataA[it][9] = css
dataA[it][10] = time
dataA[it][11] = ""
elif recording and resizes != 0: # RESIZE LINE elif recording and resizes != 0: # RESIZE LINE
iters = int(lineS[2].split('=')[1].split(',')[0]) iters = int(lineS[2].split('=')[1].split(',')[0])
...@@ -95,44 +102,49 @@ def read_file(f, dataA, dataB, it): ...@@ -95,44 +102,49 @@ def read_file(f, dataA, dataB, it):
if resizes == 0: if resizes == 0:
dataB[it][3] = npr dataB[it][3] = npr
dataB[it][4] += dist dataB[it][4] += dist
dataB[it][8] += str(iters) dataB[it][10] += str(iters)
dataA[it][4] = npr #FIXME No sera correcta si hay mas de una reconfig
dataA[it][2] = str(previousNP) + "," + str(npr) dataA[it][2] = str(previousNP) + "," + str(npr)
dataA[it][3] += dist dataA[it][5] += dist
dataA[it][7] += str(iters) dataA[it][11] += str(iters)
timer = 4 timer = 4
else: else:
dataB[it][2] = npr dataB[it][2] = npr
dataB[it][4] += dist + "," dataB[it][4] += dist + ","
dataB[it][8] += str(iters) + "," dataB[it][10] += str(iters) + ","
dataA[it][3] += dist + "," dataA[it][3] = npr
dataA[it][7] += str(iters) + "," dataA[it][5] += dist + ","
dataA[it][11] += str(iters) + ","
previousNP = npr previousNP = npr
else: # SAVE TIMES else: # SAVE TIMES
if timer == 4: if timer == 4:
dataB[it][9] = float(lineS[1]) dataB[it][11] = float(lineS[1])
elif timer == 3: elif timer == 3:
dataB[it][10] = float(lineS[1]) dataB[it][12] = float(lineS[1])
elif timer == 2: elif timer == 2:
dataB[it][11] = float(lineS[1]) dataB[it][13] = float(lineS[1])
elif timer == 1: elif timer == 1:
dataB[it][12] = float(lineS[1]) dataB[it][14] = float(lineS[1])
else: else:
dataA[it][8] = float(lineS[1]) dataA[it][12] = float(lineS[1])
timer = timer - 1 timer = timer - 1
return it return it
#columnsA1 = ["N", "%Async", "Groups", "Dist", "Matrix", "CommTam", "Time", "Iters", "TE"] #8 #columnsA1 = ["N", "%Async", "Groups", "Dist", "Matrix", "CommTam", "Cst", "Css", "Time", "Iters", "TE"] #8
#columnsB1 = ["N", "%Async", "NP", "NS", "Dist", "Matrix", "CommTam", "Time", "Iters", "TC", "TS", "TA"] #12 #columnsB1 = ["N", "%Async", "NP", "NS", "Dist", "Matrix", "CommTam", "Cst", "Css", "Time", "Iters", "TC", "TS", "TA"] #12
#Config loaded: resizes=2, matrix=1000, sdr=1000000000, adr=0, aib=0, time=2.000000 || grp=1 #Config loaded: resizes=2, matrix=1000, sdr=1000000000, adr=0, aib=0, time=2.000000 || grp=1
#Resize 0: Iters=100, Procs=2, Factors=1.000000, Phy=2 #Resize 0: Iters=100, Procs=2, Factors=1.000000, Phy=2
#Resize 1: Iters=100, Procs=4, Factors=0.500000, Phy=2 #Resize 1: Iters=100, Procs=4, Factors=0.500000, Phy=2
#Tspawn: 0.249393 #Tspawn: 0.249393
#Tthread: 0
#Tsync: 0.330391 #Tsync: 0.330391
#Tasync: 0 #Tasync: 0
#Tex: 301.428615 #Tex: 301.428615
#Config loaded: resizes=1, matrix=0, comm_tam=0, sdr=0, adr=0, aib=0, cst=3, css=1, time=1 || grp=1
#----------------------------------------------- #-----------------------------------------------
if len(sys.argv) < 2: if len(sys.argv) < 2:
...@@ -158,8 +170,8 @@ print("Number of files found: "+ str(len(lista))); ...@@ -158,8 +170,8 @@ print("Number of files found: "+ str(len(lista)));
it = -1 it = -1
dataA = [] dataA = []
dataB = [] dataB = []
columnsA = ["N", "%Async", "Groups", "Dist", "Matrix", "CommTam", "Time", "Iters", "TE"] #9 columnsA = ["N", "%Async", "Groups", "NP", "NS", "Dist", "Matrix", "CommTam", "Cst", "Css", "Time", "Iters", "TE"] #13
columnsB = ["N", "%Async", "NP", "NS", "Dist", "Matrix", "CommTam", "Time", "Iters", "TC", "TH", "TS", "TA"] #13 columnsB = ["N", "%Async", "NP", "NS", "Dist", "Matrix", "CommTam", "Cst", "Css", "Time", "Iters", "TC", "TH", "TS", "TA"] #15
for elem in lista: for elem in lista:
f = open(elem, "r") f = open(elem, "r")
......
Analysis/analyser.ipynb
View file @ f4a8b977
This source diff could not be displayed because it is too large. You can view the blob instead.
Analysis/iterTimes.py
View file @ f4a8b977
import sys import sys
import glob import glob
import numpy as np import numpy as numpy
import pandas as pd import pandas as pd
#----------------------------------------------- #-----------------------------------------------
def read_file(f, data, it): def read_file(f, dataA, dataB, itA, itB):
compute_tam = 0 compute_tam = 0
comm_tam = 0 comm_tam = 0
sdr = 0 sdr = 0
adr = 0 adr = 0
dist = 0 dist = 0
css = 0
cst = 0
time = 0 time = 0
recording = False recording = False
it_line = 0 it_line = 0
aux_it = 0 aux_itA = 0
aux_itB = 0
iters = 0 iters = 0
np = 0 np = 0
np_par = 0 np_par = 0
ns = 0 ns = 0
array = []
columnas = ['Titer','Ttype','Top'] columnas = ['Titer','Ttype','Top']
#print(f) #print(f)
...@@ -26,45 +30,72 @@ def read_file(f, data, it): ...@@ -26,45 +30,72 @@ def read_file(f, data, it):
if len(lineS) > 1: if len(lineS) > 1:
if recording and lineS[0].split(':')[0] in columnas: #Record data if recording and lineS[0].split(':')[0] in columnas: #Record data
aux_it = 0 aux_itA = 0
lineS.pop(0) lineS.pop(0)
if it_line==0: if it_line==0:
for observation in lineS: for observation in lineS:
data.append([None]*13) dataA.append([None]*15)
data[it+aux_it][0] = sdr dataA[itA+aux_itA][0] = sdr
data[it+aux_it][1] = adr dataA[itA+aux_itA][1] = adr
data[it+aux_it][2] = np dataA[itA+aux_itA][2] = np
data[it+aux_it][3] = np_par dataA[itA+aux_itA][3] = np_par
data[it+aux_it][4] = ns dataA[itA+aux_itA][4] = ns
data[it+aux_it][5] = dist dataA[itA+aux_itA][5] = dist
data[it+aux_it][6] = compute_tam dataA[itA+aux_itA][6] = compute_tam
data[it+aux_it][7] = comm_tam dataA[itA+aux_itA][7] = comm_tam
data[it+aux_it][8] = time dataA[itA+aux_itA][8] = cst
data[it+aux_it][9] = iters dataA[itA+aux_itA][9] = css
data[it+aux_it][10] = float(observation) dataA[itA+aux_itA][10] = time
aux_it+=1 dataA[itA+aux_itA][11] = iters
dataA[itA+aux_itA][12] = float(observation)
array.append(float(observation))
aux_itA+=1
elif it_line==1: elif it_line==1:
deleted = 0
for observation in lineS: for observation in lineS:
data[it+aux_it][11] = float(observation) dataA[itA+aux_itA][13] = float(observation)
aux_it+=1 if float(observation) == 0:
array.pop(aux_itA - deleted)
deleted+=1
aux_itA+=1
else: else:
for observation in lineS: for observation in lineS:
data[it+aux_it][12] = float(observation) dataA[itA+aux_itA][14] = float(observation)
aux_it+=1 aux_itA+=1
it_line += 1 it_line += 1
if(it_line % 3 == 0): # Comprobar si se ha terminado de mirar esta ejecucion if(it_line % 3 == 0): # Comprobar si se ha terminado de mirar esta ejecucion
recording = False recording = False
it_line = 0 it_line = 0
it = it + aux_it itA = itA + aux_itA
if ns != 0: # Solo obtener datos de grupos con hijos
dataB.append([None]*14)
dataB[itB][0] = sdr
dataB[itB][1] = adr
dataB[itB][2] = np
dataB[itB][3] = np_par
dataB[itB][4] = ns
dataB[itB][5] = dist
dataB[itB][6] = compute_tam
dataB[itB][7] = comm_tam
dataB[itB][8] = cst
dataB[itB][9] = css
dataB[itB][10] = time
dataB[itB][11] = iters
dataB[itB][12] = tuple(array)
dataB[itB][13] = numpy.sum(array)
itB+=1
array = []
if lineS[0] == "Config:": if lineS[0] == "Config:":
compute_tam = int(lineS[1].split('=')[1].split(',')[0]) compute_tam = int(lineS[1].split('=')[1].split(',')[0])
comm_tam = int(lineS[2].split('=')[1].split(',')[0]) comm_tam = int(lineS[2].split('=')[1].split(',')[0])
sdr = int(lineS[3].split('=')[1].split(',')[0]) sdr = int(lineS[3].split('=')[1].split(',')[0])
adr = int(lineS[4].split('=')[1].split(',')[0]) adr = int(lineS[4].split('=')[1].split(',')[0])
time = float(lineS[6].split('=')[1]) css = int(lineS[6].split('=')[1].split(',')[0])
cst = int(lineS[7].split('=')[1].split(',')[0])
time = float(lineS[8].split('=')[1])
elif lineS[0] == "Config": elif lineS[0] == "Config":
recording = True recording = True
iters = int(lineS[2].split('=')[1].split(',')[0]) iters = int(lineS[2].split('=')[1].split(',')[0])
...@@ -73,7 +104,7 @@ def read_file(f, data, it): ...@@ -73,7 +104,7 @@ def read_file(f, data, it):
np_par = int(lineS[6].split('=')[1].split(',')[0]) np_par = int(lineS[6].split('=')[1].split(',')[0])
ns = int(float(lineS[7].split('=')[1])) ns = int(float(lineS[7].split('=')[1]))
return it return itA,itB
#----------------------------------------------- #-----------------------------------------------
#Config: matrix=1000, sdr=1000000000, adr=0, aib=0 time=2.000000 #Config: matrix=1000, sdr=1000000000, adr=0, aib=0 time=2.000000
#Config Group: iters=100, factor=1.000000, phy=2, procs=2, parents=0, sons=4 #Config Group: iters=100, factor=1.000000, phy=2, procs=2, parents=0, sons=4
...@@ -90,7 +121,7 @@ else: ...@@ -90,7 +121,7 @@ else:
BaseDir = sys.argv[2] BaseDir = sys.argv[2]
if len(sys.argv) >= 4: if len(sys.argv) >= 4:
print("Csv name will be: " + sys.argv[3] + ".csv") print("Csv name will be: " + sys.argv[3] + ".csv and "+ sys.argv[3] + "_Total.csv")
name = sys.argv[3] name = sys.argv[3]
else: else:
name = "data" name = "data"
...@@ -99,18 +130,25 @@ insideDir = "Run" ...@@ -99,18 +130,25 @@ insideDir = "Run"
lista = glob.glob("./" + BaseDir + insideDir + "*/" + sys.argv[1]+ "*ID*.o*") lista = glob.glob("./" + BaseDir + insideDir + "*/" + sys.argv[1]+ "*ID*.o*")
print("Number of files found: "+ str(len(lista))); print("Number of files found: "+ str(len(lista)));
it = 0 itA = itB = 0
data = [] #0 #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 dataA = []
columns = ["N", "%Async", "NP", "N_par", "NS", "Dist", "Compute_tam", "Comm_tam", "Time", "Iters", "Ti", "Tt", "To"] #13 dataB = [] #0 #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14
columnsA = ["N", "%Async", "NP", "N_par", "NS", "Dist", "Compute_tam", "Comm_tam", "Cst", "Css","Time", "Iters", "Ti", "Tt", "To"] #15
columnsB = ["N", "%Async", "NP", "N_par", "NS", "Dist", "Compute_tam", "Comm_tam", "Cst", "Css","Time", "Iters", "Ti", "Sum"] #14
for elem in lista: for elem in lista:
f = open(elem, "r") f = open(elem, "r")
it = read_file(f, data, it) itA,itB = read_file(f, dataA, dataB, itA, itB)
f.close() f.close()
#print(data) #print(data)
df = pd.DataFrame(data, columns=columns) dfA = pd.DataFrame(dataA, columns=columnsA)
dfB = pd.DataFrame(dataB, columns=columnsB)
df['N'] += df['%Async'] dfA['N'] += dfA['%Async']
df['%Async'] = (df['%Async'] / df['N']) * 100 dfA['%Async'] = (dfA['%Async'] / dfA['N']) * 100
df.to_csv(name + '.csv') dfA.to_csv(name + '.csv')
dfB['N'] += dfB['%Async']
dfB['%Async'] = (dfB['%Async'] / dfB['N']) * 100
dfB.to_csv(name + '_Total.csv')
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment