Parallel.For maintain input list order on output list

Posted by romeozor on Stack Overflow See other posts from Stack Overflow or by romeozor
Published on 2012-06-30T10:31:39Z Indexed on 2012/06/30 21:16 UTC
Read the original article Hit count: 267

I'd like some input on keeping the order of a list during heavy-duty operations that I decided to try to do in a parallel manner to see if it boosts performance. (It did!)

I came up with a solution, but since this was my first attempt at anything parallel, I'd need someone to slap my hands if I did something very stupid.

There's a query that returns a list of card owners, sorted by name, then by date of birth. This needs to be rendered in a table on a web page (ASP.Net WebForms). The original coder decided he would construct the table cell-by-cell (TableCell), add them to rows (TableRow), then each row to the table. So no GridView, allegedly its performance is bad, but the performance was very poor regardless :).

The database query returns in no time, the most time is spent on looping through the results and adding table cells etc.

I made the following method to maintain the original order of the list:

private TableRow[] ComposeRows(List<CardHolder> queryResult)   
{
    int queryElementsCount = queryResult.Count();
    // array with the query's size
    var rowArray = new TableRow[queryElementsCount];

    Parallel.For(0, queryElementsCount, i =>
    {
        var row = new TableRow();
        var cell = new TableCell();

        // various operations, including simple ones such as:
        cell.Text = queryResult[i].Name;
        row.Cells.Add(cell);

        // here I'm adding the current item to it's original index 
        // to maintain order in the output list
        rowArray[i] = row;
    });
    return rowArray;
}

So as you can see, because I'm returning a very different type of data (List<CardHolder> -> TableRow[]), I can't just simply omit the ordering from the original query to do it after the operations.

Also, I also thought it would be a good idea to Dispose() the objects at the end of each loop, because the query can return a huge list and letting cell and row objects pile up in the heap could impact performance.(?)

How badly did I do? Does anyone have a better solution in case mine is flawed?

© Stack Overflow or respective owner

Related posts about c#

Related posts about list