Parallel Computing Toolbox To Speed Up Multiple Webread Calls

Hello All,

I am wondering if any of you have experience using the parallel computing toolbox, specifically with webread. I am currently working on a program to download, and process google finance data. Everything is going great, however the download time for aournd 300 stock stymbols is rather long. I am interested in speeding this up, and thought webread might be a good option. For example, here is a simplified version of the code I am looking to speed up.

%General set of tickers, usually 250+ in length
tickers={'AAIO','AAPL','FB','GOOGL','NVDA','TWTR'}
Length='10';
Freq='60';
for ii = 1:length(tickers)
    wop=weboptions('Timeout',15);
    temp=webread(strcat('https://www.google.com/finance/getprices?i=',Freq,'&p=',Length,...
        'd&f=d,o,h,l,c,v&df=cpct&q=',tickers{ii},'&ignore=.csv'),wop);
    data = textscan(temp, '%s%f%f%f%f%f%f', 'delimiter', ',', ...
        'Headerlines', 7);
    min=str2double(data{1});
    min(isnan(min))=0;
    op=data{2};
    high=data{3};
    low=data{4};
    cl=data{5};
    vol=data{6};
    stocks(ii).ticker=tickers{ii};
    stocks(ii).min=min;
    stocks(ii).op=op;
    stocks(ii).high=high;
    stocks(ii).low=low;
    stocks(ii).cl=cl;
    stocks(ii).vol=vol;
end

To clarify, I do not currently own the parallel computing toolbox, I am just wondering if it may help my particular situation.

Answer 1

[Kojiro Saito]

It is not always the case that doing parallel would speed up because of the overhead time of a parallel pool, but I think Parallel Computing Toolbox will improve your webread time. For using it, you need to change from

for ii = 1:length(tickers)

to

parfor ii = 1:length(tickers)

assuming "local" cluster profile is set as a default.

You can evaluate Parallel Computing Toolbox by asking a trial from this page in order to confirm it will improve your calculation time.