- Undocumented Matlab - https://undocumentedmatlab.com -

Performance: scatter vs. line

Posted By Yair Altman On In Handle graphics,Low risk of breaking in future versions,Memory,Stock Matlab function | 4 Comments

Following my previous article on the undocumented behavior of the scatter function [1], one of my readers, Benoit Charles, reported a discovery that in many circumstances the line function generates identical plots much faster than scatter.
Unlike scatter, line does not enable specific data-point marker customization, although the colors could be modified. On the other hand, line only uses a single handle object, saving memory and system resources compared to scatter keeping a separate handle for each data point. So, if you just need to quickly plot a bunch of scattered points then line could be a better choice than scatter.
Here is a simple code snippet, which generates identical plots and shows the performance difference:

>> x=rand(1000,1); y=rand(1000,1);
>> tic, for idx=1:100, cla; h=scatter(x,y); end; toc
Elapsed time is 2.521322 seconds.
>> props = {'LineStyle','none','Marker','o','MarkerEdge','b','MarkerSize',6};
>> tic, for idx=1:100, cla; h=line([x,x],[y,y],props{:}); end; toc
Elapsed time is 0.333369 seconds.

In the past, I have posted about other undocumented performance aspects, comparing the documented ismember function with the undocumented ismembc [2] and about cellfun‘s undocumented options [3]. If you are aware of other similar functions having identical outputs and a significant performance difference, please let me know.

Related posts:

  1. Plot performance [4] – Undocumented inner plot mechanisms can significantly improve plotting performance ...
  2. datestr performance [5] – Caching is a simple and very effective means to improve code performance, as demonstrated for the datestr function....
  3. Undocumented scatter plot behavior [6] – The scatter plot function has an undocumented behavior when plotting more than 100 points: it returns a single unified patch object handle, rather than a patch handle for each specific point as it returns with 100 or less points....
  4. Undocumented scatter plot jitter [7] – Matlab's scatter plot can automatically jitter data to enable better visualization of distribution density. ...
  5. Plot line transparency and color gradient [8] – Static and interpolated (gradient) colors and transparency can be set for plot lines in HG2. ...
  6. Performance: accessing handle properties [9] – Handle object property access (get/set) performance can be significantly improved using dot-notation. ...
4 Comments (Open | Close)

4 Comments To "Performance: scatter vs. line"

#1 Comment By Naor On October 14, 2009 @ 13:05

Now this is really interesting. Now I’m curious about generating 1000 line objects with one data point/marker each, and saving into a handle vector? Is this what scatter is doing under the hood?

#2 Comment By Benoit On October 15, 2009 @ 02:54

In this case, lines are faster than scatter only when colors are used and different for each point (???). Observe this script:

%% Scatter creation
T = 0;
cla
for i=1:10
	x = rand(1,1000);
	y = rand(1,1000);
	cla
	t = cputime;
	h = scatter( x,y );
	drawnow
	T = T + (cputime-t);
end
fprintf( 'Scatter creations: %.2f\n', T )

%% Lines creation
T = 0;
cla
for i=1:10
	x = rand(1,1000);
	y = rand(1,1000);
	cla
	t = cputime;
	h = line( [x;x], [y;y], props{:} );
	drawnow
	T = T + (cputime-t);
end
fprintf( 'Line creations: %.2f\n', T )

%% Scatter modification
T = 0;
cla
x = rand(1,1000);
y = rand(1,1000);
h = scatter(x,y);
for i=1:10
	y = rand(1,1000);
	t = cputime;
	set( h, 'YData', y );
	drawnow
	T = T + (cputime-t);
end
fprintf( '\nScatter modifications: %.2f\n', T )

%% Lines modification
T = 0;
cla
x = rand(1,1000);
y = rand(1,1000);
h = line( [x;x], [y;y], props{:} );
for i=1:10
	y = rand(1,1000);
	t = cputime;
	set( h, {'YData'}, num2cell([y;y]',2) );
	drawnow
	T = T + (cputime-t);
end
fprintf( 'Line modifications: %.2f\n', T )

%% Scatter color creation
T = 0;
cla
for i=1:10
	x = rand(1,1000);
	y = rand(1,1000);
	c = rand(1000,3);
	cla
	t = cputime;
	h = scatter( x,y,50,c );
	drawnow
	T = T + (cputime-t);
end
fprintf( '\nScatter color creation: %.2f\n', T )

%% Lines color creation
T = 0;
cla
for i=1:10
	x = rand(1,1000);
	y = rand(1,1000);
	c = rand(1000,3);
	cla
	t = cputime;
	h = line( [x;x], [y;y], props{:} );
	set( h, {'MarkerEdgeColor'}, num2cell(c,2) );
	drawnow
	T = T + (cputime-t);
end
fprintf( 'Line color creation: %.2f\n', T )

%% Scatter color modification
T = 0;
cla
x = rand(1,1000);
y = rand(1,1000);
c = rand(1000,3);
h = scatter(x,y,50,c);
for i=1:10
	y = rand(1,1000);
	t = cputime;
	set( h, 'YData', y );
	drawnow
	T = T + (cputime-t);
end
fprintf( '\nScatter color modification: %.2f\n', T )

%% Lines modification
T = 0;
cla
x = rand(1,1000);
y = rand(1,1000);
c = rand(1000,3);
h = line( [x;x], [y;y], props{:} );
set( h, {'MarkerEdgeColor'}, num2cell(c,2) );
for i=1:10
	y = rand(1,1000);
	t = cputime;
	set( h, {'YData'}, num2cell([y;y]',2) );
	drawnow
	T = T + (cputime-t);
end
fprintf( 'Line color modification: %.2f\n', T )

When you execute this script, you obtain:

Scatter creations: 0.76
Line creations: 9.53

Scatter modifications: 0.61
Line modifications: 1.51

Scatter color creation: 4.02
Line color creation: 17.07

Scatter color modification: 4.34
Line color modification: 1.53

Magic, isn’t it?
Sometimes, I wonder if Matlab was not developed at Hogwarts…

#3 Comment By Naor On October 15, 2009 @ 15:41

very strange indeed.

#4 Comment By damayi On May 22, 2012 @ 19:59

In my application, I will use line as much as possible.
What’s more, line is more simple and easy to control.