[Matplotlib-users] Memory leak when using pyplot.ion() ?
Hi,
I am having problems with a script. It runs a number of iterations and plots
and saves a number of plots on each iteration. After the plots have been saved
I issue the pyplot.close('all') command so despite many plots being created
only 4 should be open at any given time which should not cause any memory
problems. When I run the script however I see the RAM usage gradually growing
without bound and eventually causing the script to crash. Interestingly I have
found if I comment out the pyplot.ion() and pyplot.ioff() the problem
vanishes. So I do have a workaround but it would still be good to have this
fixed in case I forget about it in future and loose another weekend's work.
My OS is Windows XP Service Pack 3
Python 2.6
Matplotlib 1.0.1
The code below is a stripped down version of my script which still exhibits the
problem.
Oisín.
# -*- coding: utf-8 -*-
import sys
import time
import numpy as np
from matplotlib import pyplot
import os
# Main script body
try:
for gain in range(1,20,2):
for PortToTest in range(8):
dirname = '.\crash'
f = open(dirname + '\\results.m','w')
runname = '\P' + str(PortToTest) + str(gain) + \
'_' + time.strftime('d%dh%Hm%Ms%S')
dirname = dirname + runname
os.mkdir(dirname)
os.system('copy ' + sys.argv[0] + ' ' + dirname )
nIts = 50
# Decimate data for plotting if many iterations are run
if(nIts>10):
echoPlotDec = 10
else:
echoPlotDec = 1
ResidN = np.zeros((4,2*nIts))
MaxSl= np.zeros((4,2*nIts))
MaxOld = np.zeros((4,2*nIts))
MaxNew = np.zeros((4,2*nIts))
EchoA= np.zeros((2*nIts,160))
for kk in range(2*nIts):
ResidN[0,kk] = np.random.rand(1,1)
ResidN[1,kk] = np.random.rand(1,1)
ResidN[2,kk] = np.random.rand(1,1)
ResidN[3,kk] = np.random.rand(1,1)
MaxSl[0,kk] = np.random.rand(1,1)
MaxSl[1,kk] = np.random.rand(1,1)
MaxSl[2,kk] = np.random.rand(1,1)
MaxSl[3,kk] = np.random.rand(1,1)
MaxOld[0,kk] = np.random.rand(1,1)
MaxOld[1,kk] = np.random.rand(1,1)
MaxOld[2,kk] = np.random.rand(1,1)
MaxOld[3,kk] = np.random.rand(1,1)
MaxNew[0,kk] = np.random.rand(1,1)
MaxNew[1,kk] = np.random.rand(1,1)
MaxNew[2,kk] = np.random.rand(1,1)
MaxNew[3,kk] = np.random.rand(1,1)
EchoA[kk,:] = np.random.rand(1,160)
f.close()
pyplot.ion()
pyplot.figure()
pyplot.hold(True)
LegendTexts = ("A","B","C","D")
pyplot.title("R (" + runname +")")
pyplot.xlabel("Index")
pyplot.ylabel("Noise (dB)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(ResidN),'.-')
pyplot.legend(LegendTexts,loc=1)
pyplot.axis([0, 2*nIts, -33, -25])
pyplot.savefig(dirname + '\\results.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("Coefs")
pyplot.xlabel("Coef Index")
pyplot.ylabel("Coef Value")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(EchoA[0:nIts-1:echoPlotDec,:]),'.-')
pyplot.plot(np.transpose(EchoA[nIts:2*nIts-1:echoPlotDec,:]),'*-')
pyplot.axis([0, 160, -0.5, 2])
pyplot.savefig(dirname + '\\CoefsA.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("MaxAbs, Old = '.', New = '*' ")
pyplot.xlabel("Iteration")
pyplot.ylabel("o/p (LSBs)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(MaxOld),'.-')
pyplot.plot(np.transpose(MaxNew),'*-')
pyplot.axis([0, 2*nIts, 32, 128])
pyplot.savefig(dirname + '\\MaxAbsA.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("MaxAbs")
pyplot.xlabel("Iteration")
pyplot.ylabel("(LSBs)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(MaxSl),'.-')
pyplot.axis([0, 2*nIts, 0, 64])
pyplot.savefig(dirname + '\\MaxAbsSl.emf',format='emf')
pyplot.close('all')
except RuntimeError, msg:
print 'Exception occurred in main script body'
print >>sys.stderr, msg
raise
finally:
print "Test done"
# Display plots
pyplot.ioff()
--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk___
Matplotlib-users mailing l
Re: [Matplotlib-users] Memory leak when using pyplot.ion() ?
I haven't had a chance to look into where the memory is actually
leaking, ion/ioff are intended for interactive use, and here you are
saving a large number of plots to files. Why do you need ion at all?
Mike
On 10/14/2013 08:51 AM, OCuanachain, Oisin (Oisin) wrote:
Hi,
I am having problems with a script. It runs a number of iterations and
plots and saves a number of plots on each iteration. After the plots
have been saved I issue the pyplot.close('all') command so despite
many plots being created only 4 should be open at any given time which
should not cause any memory problems. When I run the script however I
see the RAM usage gradually growing without bound and eventually
causing the script to crash. Interestingly I have found if I comment
out the pyplot.ion() and pyplot.ioff() the problem vanishes. So I do
have a workaround but it would still be good to have this fixed in
case I forget about it in future and loose another weekend's work.
My OS is Windows XP Service Pack 3
Python 2.6
Matplotlib 1.0.1
The code below is a stripped down version of my script which still
exhibits the problem.
Oisín.
# -*- coding: utf-8 -*-
import sys
import time
import numpy as np
from matplotlib import pyplot
import os
# Main script body
try:
for gain in range(1,20,2):
for PortToTest in range(8):
dirname = '.\crash'
f = open(dirname + '\\results.m','w')
runname = '\P' + str(PortToTest) + str(gain) + \
'_' + time.strftime('d%dh%Hm%Ms%S')
dirname = dirname + runname
os.mkdir(dirname)
os.system('copy ' + sys.argv[0] + ' ' + dirname )
nIts = 50
# Decimate data for plotting if many iterations are run
if(nIts>10):
echoPlotDec = 10
else:
echoPlotDec = 1
ResidN = np.zeros((4,2*nIts))
MaxSl= np.zeros((4,2*nIts))
MaxOld = np.zeros((4,2*nIts))
MaxNew = np.zeros((4,2*nIts))
EchoA= np.zeros((2*nIts,160))
for kk in range(2*nIts):
ResidN[0,kk] = np.random.rand(1,1)
ResidN[1,kk] = np.random.rand(1,1)
ResidN[2,kk] = np.random.rand(1,1)
ResidN[3,kk] = np.random.rand(1,1)
MaxSl[0,kk] = np.random.rand(1,1)
MaxSl[1,kk] = np.random.rand(1,1)
MaxSl[2,kk] = np.random.rand(1,1)
MaxSl[3,kk] = np.random.rand(1,1)
MaxOld[0,kk] = np.random.rand(1,1)
MaxOld[1,kk] = np.random.rand(1,1)
MaxOld[2,kk] = np.random.rand(1,1)
MaxOld[3,kk] = np.random.rand(1,1)
MaxNew[0,kk] = np.random.rand(1,1)
MaxNew[1,kk] = np.random.rand(1,1)
MaxNew[2,kk] = np.random.rand(1,1)
MaxNew[3,kk] = np.random.rand(1,1)
EchoA[kk,:] = np.random.rand(1,160)
f.close()
pyplot.ion()
pyplot.figure()
pyplot.hold(True)
LegendTexts = ("A","B","C","D")
pyplot.title("R (" + runname +")")
pyplot.xlabel("Index")
pyplot.ylabel("Noise (dB)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(ResidN),'.-')
pyplot.legend(LegendTexts,loc=1)
pyplot.axis([0, 2*nIts, -33, -25])
pyplot.savefig(dirname + '\\results.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("Coefs")
pyplot.xlabel("Coef Index")
pyplot.ylabel("Coef Value")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(EchoA[0:nIts-1:echoPlotDec,:]),'.-')
pyplot.plot(np.transpose(EchoA[nIts:2*nIts-1:echoPlotDec,:]),'*-')
pyplot.axis([0, 160, -0.5, 2])
pyplot.savefig(dirname + '\\CoefsA.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("MaxAbs, Old = '.', New = '*' ")
pyplot.xlabel("Iteration")
pyplot.ylabel("o/p (LSBs)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(MaxOld),'.-')
pyplot.plot(np.transpose(MaxNew),'*-')
pyplot.axis([0, 2*nIts, 32, 128])
pyplot.savefig(dirname + '\\MaxAbsA.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("MaxAbs")
pyplot.xlabel("Iteration")
pyplot.ylabel("(LSBs)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(MaxSl),'.-')
pyplot.axis([0, 2*nIts, 0, 64])
pyplot.savefig(dirname + '\\MaxAbsSl.emf',format='emf')
pyplot.close('all')
except RuntimeError, msg:
print 'Exception occurred in main script body'
print >>sys.stderr, msg
raise
finally:
print "Test done"
# Display plots
pyplot.ioff()
--
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel process
Re: [Matplotlib-users] Matplotlib eating memory
Sorry to repeat myself, but please reduce this to a short, self contained example, that is absolutely minimal to demonstrate the problem. http://sscce.org/ should help better explain what I'm after. I don't want to find the needle in the haystack here -- there is code in your example that doesn't even run, for example. That said, are you really after creating a legend entry for each of the dots? (See below). That just isn't going to work, and I'm not surprised it eats up excessive amounts of memory. I think you want (and can) reduce this to a single scatter call. _series = [_ax1.scatter(_x, _y, color=_c, s=objsize, label=_l, hatch='.') for _x, _y, _c, _l in izip(mydata_x, mydata_y, colors, legends)] # returns PathCollection object Mike On 10/12/2013 12:57 PM, Martin MOKREJŠ wrote: Hi, so here is some quick but working example. I added there are 2-3 functions (unused) as a bonus, you can easily call them from the main function using same API (except the piechart). I hope this shows what I lack in matplotlib - a general API so that I could easily switch form scatter plot to piechart or barchart without altering much the function arguments. Messing with return objects line2D, PathCollection, Rectangle is awkward and I would like to stay away from matplotlib's internals. ;) Some can be sliced, so not, you will see in the code. This eatmem.py will take easily all your memory. Drawing 30 dots is not feasible with 16GB of RAM. While the example is for sure inefficient in many places generating the data in python does not eat RAM. That happens afterwards. I would really like to hear whether matplotlib could be adjusted instead. ;) I already mentioned in this thread that it is awkward to pre-create colors before passing all data to a drawing function. I think we could all save a lot if matplotlib could dynamically fetch colors on the fly from user-created generator, same for legends descriptions. I think my example code shows the inefficient approach here. Would I have more time I would randomize a bit more the sublist of each series so that the numbers in legends would be more variable but that is a cosmetic issue. Probably due to my ignorance you will see that figures with legends have different font sizes, axes are rescaled and the figure. Of course I wanted to have the drawing same via both approaches but failed badly. The files/figures with legends should be just accompanied by the legend "table" underneath but the drawing itself should be same. Maybe an issue with DPI settings but not only. I placed some comments in the code, please don't take them in person. ;) Of course I am glad for the existing work and am happy to contribute my crap. I am fine if you rewamp this ugly code into matplotlib testsuite, provide similar function (the API mentioned above) so that I could use your code directly. That would be great. I just tried to show multiple issues at once, notably that is why I included those unused functions. You will for sure find a way to use them. Regarding the "unnecessary" del() calls etc., I think I have to use keep some, Ben, because the function is not always left soon enough. I could drop some, you are right, but for some I don't think so. Matplotlib cannot recycle the memory until me (upstream) deletes the reference so ... go and test this lousy code. Now you have a testcase. ;) Same with the gc.collect() calls. Actually, the main loop with 10 iteration is there just to show why I always want to clear a figure when entering a function and while leaving it as well. It happened too many times that I drawed over an old figure, and this was posted also few times on this list by others. That is a weird behavior in my opinion. We, users, are just forced to use too low-level functions. So, have fun eating your memory! :)) Martin -- _ |\/|o _|_ _. _ | | \.__ __|__|_|_ _ _ ._ _ | ||(_| |(_|(/_| |_/|(_)(/_|_ |_|_)(_)(_)| | | http://www.droettboom.com -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
[Matplotlib-users] Maidir le: Memory leak when using pyplot.ion() ?
Hi Mike,
ion(), ioff() are useful to get immediate feedback when developing a script,
when it is fully debugged I then increase the number of iterations and leave it
running over the weekend. At that point I could obviously also have removed
ion(), ioff() but given that I had no idea that this was necessary my sim
crashed and I lost a weekend's worth of sim time. Anyway, whether or not
ion(),ioff() are needed in this particular script is really besides the point.
If the script, however unusual, is revealing a bug in matplotlib it should be
logged so that it can hopefully be fixed.
Oisín
From: Michael Droettboom [mailto:md...@stsci.edu]
Sent: 14 October 2013 18:13
To: matplotlib-users@lists.sourceforge.net
Subject: Re: [Matplotlib-users] Memory leak when using pyplot.ion() ?
I haven't had a chance to look into where the memory is actually leaking,
ion/ioff are intended for interactive use, and here you are saving a large
number of plots to files. Why do you need ion at all?
Mike
On 10/14/2013 08:51 AM, OCuanachain, Oisin (Oisin) wrote:
Hi,
I am having problems with a script. It runs a number of iterations and plots
and saves a number of plots on each iteration. After the plots have been saved
I issue the pyplot.close('all') command so despite many plots being created
only 4 should be open at any given time which should not cause any memory
problems. When I run the script however I see the RAM usage gradually growing
without bound and eventually causing the script to crash. Interestingly I have
found if I comment out the pyplot.ion() and pyplot.ioff() the problem
vanishes. So I do have a workaround but it would still be good to have this
fixed in case I forget about it in future and loose another weekend's work.
My OS is Windows XP Service Pack 3
Python 2.6
Matplotlib 1.0.1
The code below is a stripped down version of my script which still exhibits the
problem.
Oisín.
# -*- coding: utf-8 -*-
import sys
import time
import numpy as np
from matplotlib import pyplot
import os
# Main script body
try:
for gain in range(1,20,2):
for PortToTest in range(8):
dirname = '.\crash'
f = open(dirname + '\\results.m','w')
runname = '\P' + str(PortToTest) + str(gain) + \
'_' + time.strftime('d%dh%Hm%Ms%S')
dirname = dirname + runname
os.mkdir(dirname)
os.system('copy ' + sys.argv[0] + ' ' + dirname )
nIts = 50
# Decimate data for plotting if many iterations are run
if(nIts>10):
echoPlotDec = 10
else:
echoPlotDec = 1
ResidN = np.zeros((4,2*nIts))
MaxSl= np.zeros((4,2*nIts))
MaxOld = np.zeros((4,2*nIts))
MaxNew = np.zeros((4,2*nIts))
EchoA= np.zeros((2*nIts,160))
for kk in range(2*nIts):
ResidN[0,kk] = np.random.rand(1,1)
ResidN[1,kk] = np.random.rand(1,1)
ResidN[2,kk] = np.random.rand(1,1)
ResidN[3,kk] = np.random.rand(1,1)
MaxSl[0,kk] = np.random.rand(1,1)
MaxSl[1,kk] = np.random.rand(1,1)
MaxSl[2,kk] = np.random.rand(1,1)
MaxSl[3,kk] = np.random.rand(1,1)
MaxOld[0,kk] = np.random.rand(1,1)
MaxOld[1,kk] = np.random.rand(1,1)
MaxOld[2,kk] = np.random.rand(1,1)
MaxOld[3,kk] = np.random.rand(1,1)
MaxNew[0,kk] = np.random.rand(1,1)
MaxNew[1,kk] = np.random.rand(1,1)
MaxNew[2,kk] = np.random.rand(1,1)
MaxNew[3,kk] = np.random.rand(1,1)
EchoA[kk,:] = np.random.rand(1,160)
f.close()
pyplot.ion()
pyplot.figure()
pyplot.hold(True)
LegendTexts = ("A","B","C","D")
pyplot.title("R (" + runname +")")
pyplot.xlabel("Index")
pyplot.ylabel("Noise (dB)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(ResidN),'.-')
pyplot.legend(LegendTexts,loc=1)
pyplot.axis([0, 2*nIts, -33, -25])
pyplot.savefig(dirname + '\\results.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("Coefs")
pyplot.xlabel("Coef Index")
pyplot.ylabel("Coef Value")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(EchoA[0:nIts-1:echoPlotDec,:]),'.-')
pyplot.plot(np.transpose(EchoA[nIts:2*nIts-1:echoPlotDec,:]),'*-')
pyplot.axis([0, 160, -0.5, 2])
pyplot.savefig(dirname + '\\CoefsA.emf',format='emf')
pyplot.figure()
pyplot.hold(True)
pyplot.title("MaxAbs, Old = '.', New = '*' ")
pyplot.xlabel("Iteration")
pyplot.ylabel("o/p (LSBs)")
pyplot.grid(True)
pyplot.hold(True)
pyplot.plot(np.transpose(MaxOld),'.-')
pyplot.plot(np.transpose(MaxNew),'*-')
pyplot.axis([0, 2*nIts, 32, 128])
pyplot.savefig(dirname + '\\MaxAbs
Re: [Matplotlib-users] Memory leak when using pyplot.ion() ?
On 14/10/2013 13:51, OCuanachain, Oisin (Oisin) wrote: > Hi, > > I am having problems with a script. It runs a number of iterations and > plots and saves a number of plots on each iteration. After the plots > have been saved I issue the pyplot.close(‘all’) command so despite many > plots being created only 4 should be open at any given time which should > not cause any memory problems. When I run the script however I see the > RAM usage gradually growing without bound and eventually causing the > script to crash. Interestingly I have found if I comment out the > pyplot.ion() and pyplot.ioff() the problem vanishes. So I do have a > workaround but it would still be good to have this fixed in case I > forget about it in future and loose another weekend’s work. > > My OS is Windows XP Service Pack 3 > Python 2.6 > Matplotlib 1.0.1 > Is this actually a matplotlib problem or could it be a Windows problem as discussed here http://bugs.python.org/issue19246 ? -- Roses are red, Violets are blue, Most poems rhyme, But this one doesn't. Mark Lawrence -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Maidir le: Memory leak when using pyplot.ion() ?
I see you are using matplotlib 1.0.1. There have been several memory leak bugs fixed since then, so I would suggest upgrading. I also notice you are using the "emf" backend for saving figures. If I remember correctly, that backend has been deprecated (or maybe even removed) in the latest release (v1.3.x). So, i would suggest trying v1.2.1. It does contain many bugfixes and should still be compatible with your existing code. I should also warn that the behavior with pyplot.ion() was "odd" back in the 1.0.1 days and prior. If you are just simply running the script as-is with a more recent matplotlib, or with a different backend in v1.0.1, the script might behave a little differently than you'd expect. We would welcome feedback on your usage of pyplot.ion(). Cheers! Ben Root -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Maidir le: Memory leak when using pyplot.ion() ?
On 2013/10/14 7:48 AM, OCuanachain, Oisin (Oisin) wrote: > Hi Mike, > > ion(), ioff() are useful to get immediate feedback when developing a > script, when it is fully debugged I then increase the number of > iterations and leave it running over the weekend. At that point I could > obviously also have removed ion(), ioff() but given that I had no idea > that this was necessary my sim crashed and I lost a weekend’s worth of > sim time. Anyway, whether or not ion(),ioff() are needed in this > particular script is really besides the point. If the script, however > unusual, is revealing a bug in matplotlib it should be logged so that it > can hopefully be fixed. > > Oisín Oisín, Certainly we want to find and fix bugs, with memory leaks being high priority. (I don't think we have seen a genuine mpl memory leak for quite a while; I am not aware of any at present.) We are not trying to maintain old mpl versions such as 1.0.1, however. You are using the emf backend, which has been removed. Therefore, unless you can reproduce the problem with mpl 1.3.x or 1.4.x in a SSCCE (http://sscce.org/), it is unlikely that your report will lead to a bug fix. Perhaps it will lead to some useful insight, however. I see someone has suggested that the problem might be in Windows. Another possibility is that it is in TkAgg, which I suspect is your default interactive backend. I dimly recall that there was a time when TkAgg could leak memory, but I don't remember whether that was fixed by 1.0.1 or not. Eric -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
[Matplotlib-users] Maidir le: Maidir le: Memory leak when using pyplot.ion() ?
-Bunteachtaireacht- From: Eric Firing [mailto:efir...@hawaii.edu] Sent: 14 October 2013 19:09 To: matplotlib-users@lists.sourceforge.net Subject: Re: [Matplotlib-users] Maidir le: Memory leak when using pyplot.ion() ? On 2013/10/14 7:48 AM, OCuanachain, Oisin (Oisin) wrote: > Hi Mike, > > ion(), ioff() are useful to get immediate feedback when developing a > script, when it is fully debugged I then increase the number of > iterations and leave it running over the weekend. At that point I > could obviously also have removed ion(), ioff() but given that I had > no idea that this was necessary my sim crashed and I lost a weekend's > worth of sim time. Anyway, whether or not ion(),ioff() are needed in > this particular script is really besides the point. If the script, > however unusual, is revealing a bug in matplotlib it should be logged > so that it can hopefully be fixed. > > Oisín Oisín, Certainly we want to find and fix bugs, with memory leaks being high priority. (I don't think we have seen a genuine mpl memory leak for quite a while; I am not aware of any at present.) We are not trying to maintain old mpl versions such as 1.0.1, however. You are using the emf backend, which has been removed. Therefore, unless you can reproduce the problem with mpl 1.3.x or 1.4.x in a SSCCE (http://sscce.org/), it is unlikely that your report will lead to a bug fix. Perhaps it will lead to some useful insight, however. I see someone has suggested that the problem might be in Windows. Another possibility is that it is in TkAgg, which I suspect is your default interactive backend. I dimly recall that there was a time when TkAgg could leak memory, but I don't remember whether that was fixed by 1.0.1 or not. Eric Hi Eric, If .emf is no longer supported in current versions of matplotlib is there an alternative SVG-type format I can use ? I use .emf because I find that it tends to produce the clearest plots independent of how I re-size them when imported into documents. If memory serves I used to have a lot of problems with legibility when using raster-type formats like .png. If I get a chance I will try installing an up-to-date version of matplotlib and see if I can re-produce the behaviour. Oisin. -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Maidir le: Maidir le: Memory leak when using pyplot.ion() ?
On 2013/10/14 8:26 AM, OCuanachain, Oisin (Oisin) wrote: > > > Hi Eric, > > If .emf is no longer supported in current versions of matplotlib is > there an alternative SVG-type format I can use ? I use .emf because I > find that it tends to produce the clearest plots independent of how I > re-size them when imported into documents. If memory serves I used to > have a lot of problems with legibility when using raster-type formats > like .png. If I get a chance I will try installing an up-to-date > version of matplotlib and see if I can re-produce the behaviour. > > Oisin. > Oisin, The ps, pdf, and svg vector formats are fully supported. (Postscript inherently lacks support for transparency, though, so pdf and svg are recommended.) Eric -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Memory leak when using pyplot.ion() ?
2013/10/14 Mark Lawrence : > On 14/10/2013 13:51, OCuanachain, Oisin (Oisin) wrote: >> Hi, >> >> I am having problems with a script. It runs a number of iterations and >> plots and saves a number of plots on each iteration. After the plots >> have been saved I issue the pyplot.close(‘all’) command so despite many >> plots being created only 4 should be open at any given time which should >> not cause any memory problems. When I run the script however I see the >> RAM usage gradually growing without bound and eventually causing the >> script to crash. Interestingly I have found if I comment out the >> pyplot.ion() and pyplot.ioff() the problem vanishes. So I do have a >> workaround but it would still be good to have this fixed in case I >> forget about it in future and loose another weekend’s work. >> >> My OS is Windows XP Service Pack 3 >> Python 2.6 >> Matplotlib 1.0.1 >> > > Is this actually a matplotlib problem or could it be a Windows problem > as discussed here http://bugs.python.org/issue19246 ? I think this is different. That bug report is not about RAM usage growing without bound but memory allocation failing with plenty of RAM available. Goyo -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Re: [Matplotlib-users] Matplotlib eating memory
Michael Droettboom wrote: > Sorry to repeat myself, but please reduce this to a short, self contained > example, that is absolutely minimal to demonstrate the problem. > http://sscce.org/ should help better explain what I'm after. I don't want to > find the needle in the haystack here -- there is code in your example that > doesn't even run, for example. > > That said, are you really after creating a legend entry for each of the dots? > (See below). That just isn't going to work, and I'm not surprised it eats > up excessive amounts of memory. I think you want (and can) reduce this to a > single scatter call. > > _series = [_ax1.scatter(_x, _y, color=_c, s=objsize, label=_l, hatch='.') for > _x, _y, _c, _l in izip(mydata_x, mydata_y, colors, legends)] # returns > PathCollection object Are you sure? I think it was concluded on this list that scatter cannot (or does not) take nested lists of lists with series like histogram and piechart do. I cannot find the thread but maybe you are more lucky. I even think that I already opened a bugreport/feature requested in the past for this. But maybe not. Martin > > Mike > > On 10/12/2013 12:57 PM, Martin MOKREJŠ wrote: >> Hi, >> so here is some quick but working example. I added there are 2-3 functions >> (unused) >> as a bonus, you can easily call them from the main function using same API >> (except the piechart). I hope this shows what I lack in matplotlib - a >> general API >> so that I could easily switch form scatter plot to piechart or barchart >> without altering >> much the function arguments. Messing with return objects line2D, >> PathCollection, Rectangle >> is awkward and I would like to stay away from matplotlib's internals. ;) >> Some can be sliced, >> so not, you will see in the code. >> >> This eatmem.py will take easily all your memory. Drawing 30 dots is >> not feasible >> with 16GB of RAM. While the example is for sure inefficient in many places >> generating the data >> in python does not eat RAM. That happens afterwards. >> >> I would really like to hear whether matplotlib could be adjusted instead. ;) >> I already mentioned >> in this thread that it is awkward to pre-create colors before passing all >> data to a drawing >> function. I think we could all save a lot if matplotlib could dynamically >> fetch colors >> on the fly from user-created generator, same for legends descriptions. I >> think my example >> code shows the inefficient approach here. Would I have more time I would >> randomize a bit >> more the sublist of each series so that the numbers in legends would be more >> variable >> but that is a cosmetic issue. >> Probably due to my ignorance you will see that figures with legends have >> different font >> sizes, axes are rescaled and the figure. Of course I wanted to have the >> drawing same via both >> approaches but failed badly. The files/figures with legends should be just >> accompanied by the >> legend "table" underneath but the drawing itself should be same. Maybe an >> issue with DPI settings >> but not only. >> >> I placed some comments in the code, please don't take them in person. ;) >> Of course >> I am glad for the existing work and am happy to contribute my crap. I am >> fine if you rewamp >> this ugly code into matplotlib testsuite, provide similar function (the API >> mentioned above) >> so that I could use your code directly. That would be great. I just tried to >> show multiple >> issues at once, notably that is why I included those unused functions. You >> will for sure find >> a way to use them. >> >> Regarding the "unnecessary" del() calls etc., I think I have to use keep >> some, Ben, because >> the function is not always left soon enough. I could drop some, you are >> right, but for some >> I don't think so. Matplotlib cannot recycle the memory until me (upstream) >> deletes the reference >> so ... go and test this lousy code. Now you have a testcase. ;) Same with >> the gc.collect() calls. >> Actually, the main loop with 10 iteration is there just to show why I always >> want to clear >> a figure when entering a function and while leaving it as well. It happened >> too many times that >> I drawed over an old figure, and this was posted also few times on this list >> by others. That is >> a weird behavior in my opinion. We, users, are just forced to use too >> low-level functions. >> >> So, have fun eating your memory! :)) >> Martin > > > -- >_ > |\/|o _|_ _. _ | | \.__ __|__|_|_ _ _ ._ _ > | ||(_| |(_|(/_| |_/|(_)(/_|_ |_|_)(_)(_)| | | > > http://www.droettboom.com > -- Martin Mokrejs, Ph.D. Bioinformatics Donovalska 1658 149 00 Prague Czech Republic http://www.iresite.org http://www.iresite.org/~mmokrejs -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application perform
Re: [Matplotlib-users] Matplotlib eating memory
On 10/10/2013 15:05, Martin MOKREJŠ wrote: > Hi, > rendering some of my charts takes almost 50GB of RAM. I believe below is a > stracktrace > of one such situation when it already took 15GB. Would somebody comments on > what is > matplotlib doing at the very moment? Why the recursion? > > The charts had to have 262422 data points in a 2D scatter plot, each point > has assigned > its own color. They are in batches so that there are 153 distinct colors but > nevertheless, > I assigned to each data point a color value. There are 153 legend items also > (one color > won't be used). Hello Martin, can I ask what is the meaning of plotting a scatter plot with 200 thousands points in it? Either you visualize it on a screen much larger than mine, or you are not going to be able to distinguish the single data points. Maybe you should rethink the visualization tool you are using. Nevertheless, I'm perfectly able to plot a scatter plot with 262422 data points each with its own color just fine, and the python process consumes a few hundred Mb of ram (having quite a few other datasets loaded in memory):: import numpy as np import matplotlib.pyplot as plt n = 262422 x = np.random.rand(n) y = np.random.rand(n) c = np.random.rand(n) f = plt.figure() a = f.add_subplot(111) a.scatter(x, y, c=c, s=50) plt.show() and a possible solution using exactly 153 different colors, but again, I don't see how you can distinguish between hundreds different shades of colors:: n = 262422 #22 ncolors = 153 x = np.random.rand(n) y = np.random.rand(n) c = np.random.rand(ncolors) f = plt.figure() a = f.add_subplot(111) for i in xrange(n // ncolors): a.scatter(x[i*ncolors:(i+1)*ncolors], y[i*ncolors:(i+1)*ncolors], c=c, s=50) plt.show() Unfortunately the code you provide is too contrived to be useful to understand the root cause of your problem. Cheers, Daniele -- October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk ___ Matplotlib-users mailing list Matplotlib-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-users
