Which is quicker?

Question

HI all.im writing a program that is to search an xml file for the different components that can be present, CDATA, elements...etc
There are two ways im thinking of approaching this.
The first is creating several threads, each one is responsible for retrieving a different type of node.For example, one thread is responsible for retreving elements, another thread is responsible for retrieving comments and so on.
The other approach is to have one the one exectuing thread that performs a switch statement on every single component it encounters in the xml file.
My question is, will the multi threaded approach be quicker?Im not sure because in the multi-threaded approach, each thread will have to travers teh xml file exactly once, and in the other approach, using the one thread and a switch statement, this will have to traverse the file structure once also.
Many thanks!

Answer 1

Don't fall for the myth that two threads are quicker than one on a single-CPU machine.i think it's depends on how th ethread is used. For application that perform
read and writing to/from disk..then, thread would most likely make the
application faster. Your computer (for single processor..doesn't matter if i's "Hyper-Thread" or not) only run one thread at a time. However, when you make an IO call..the latency time is much greater than if you were reading and writing using RAM. during this time..you thread is block or waiting (forgot which..but i think it's block). if you multi-thread the applicatioin..while you are block, you can perform other task..this in sense, speed up the application.
a good example is when you ask for an HTML page.. the server may takes 5
minutes to send you the information (given that your socket timeout is not
smaller than this time perdiod). while waiting for th eserver response, you could
be doing something else.
As for working with DOM....I don't think multi-thread will help you here...since it's
likely that the DOM is in memory (RAM). Actually..inproper use of thread may
slow down you application (swapping data to register at context switch if lots of
thrashing occurs).
As mention in previous reply. you may wants to consider using SAX parser for
speed performance. SAX will parse th edocument once, while it's parsing, it will perform a "call back" to the startElement, endsElement, characters, etc..This is where you provide the code to listener for the nodename, values, etc.. Dom will parse the document and store it in memory before the DOM can be used. However, you should not worries about speed performances until you profile you application to determine where the bottle neck is. What type of parser to use should depends on how you use the XML data. worries about this first.