HELP: received error when trying to run two functions in parallel.

I am trying to run two functions in parallel so that they will plot a real-time graph at the same time. My code looks like:

tf = 60;

parfor i = 1:2
    if i == 1 
        disp('OpenBCI start');
        [EEGfilt, EMGfiltNormal, t] = OpenBCIfunction(tf);
    else 
        disp('Flex Sensor start');
        [R_f0, R_f1, R_f2, R_f4, R_f5, t] = Glovefunction(tf);
    end
end 

When I run the code, I get the command window to display "OpenBCI start" and "Flex Sensor start". However, I never get a graph to show up and right after "Flex Sensor start" I recieve this message in the command window.

Caught "std::exception" Exception message is:

Unknown exception

Error using OpenBCI_Glove_Parallel (line 9)

Unknown exception

How do I fix this so I can get these two functions to run in parallel and output individual live graphs? (The graphs are being plotted within the functions.)

Answer 1

[Raymond Norris]

It's unclear to me what could be causing the exception. Might need to troubleshoot this with Technical Support ([e-mail]).

If the functions OpenBCIfunction and Glovefunction generate the graphs, then they're going to generate them via the headless workers -- i.e. no graphs. However, if the initial graphs are created a prior, then the functions can send data back to the main MATLAB, which can intern update the graphs. Take a look at parallel.pool.DataQueue for an example that might work for you.

https://www.mathworks.com/help/parallel-computing/parallel.pool.dataqueue.html

[Madison Bates]

I see a lot of examples for updating a surface plot with a parfor loop. However, I am trying to plot multiple signals in 2 different 2D plots. Is there an easy way to do this? At the beginning of my code, I wrote this:

parpool('local',2); %creates and returns a pool with a specified number of workers
q = parallel.pool.DataQueue;
afterEach(q,@(data) updatePlot(surface,data));

I also created the two figures before the parfor loop. My parfor loop looks like this now:

parfor ii = 1:2
    if ii == 1 
        disp('Flex Sensor start');
        [R_f0, R_f1, R_f2, R_f4, R_f5, t] = Glovefunction(tf);
    else 
        disp('OpenBCI start');
        [EEGfilt, EMGfiltNormal, t] = OpenBCIfunction(tf);
    end
    send(q, ii); 
end 

Thank you.

[Raymond Norris]

I would suggest two data queues. Here's a sketch of what you could do.

parpool('local',2);

glove_h = figure();
glove_q = parallel.pool.DataQueue;
afterEach(glove_q,@(data)updateGlovePlot(glove_h,data));

bci_h = figure();
bci_q = parallel.pool.DataQueue;
afterEach(bci_q,@(data)updateBciPlot(bci_h,data);

parfor ii = 1:2
    if ii == 1 
        disp('Flex Sensor start');
        [R_f0, R_f1, R_f2, R_f4, R_f5, t] = Glovefunction(tf);
        send(glove_q,[ data(s) to be passed to update the plot ])
    else 
        disp('OpenBCI start');
        [EEGfilt, EMGfiltNormal, t] = OpenBCIfunction(tf);
        send(bci_q,[ data(s) to be passed to update the plot ]);
    end
end 

The "data(s) to be passed to update the plot" is probably based on the variables passed back form calling the Glovefunction / OpenBCIfunction.