On 24 Oct 2006, at 11:19, Trausti Thor Johannsson wrote:
I know. But I had a dream about some stack like that, just dump objects into it, and run, and it would run on all cpu's :)
It is an example of data-flow programming, where you have a sequencing engine which distributes the processing of an operation on a set of data across a set of processors. Graphics programming is a good example of this, where the GPU is applying the same pixel shader operations to multiple pixels in parallel.
Some types of data (like images) are well suited to this because there is often no data interdependencies. And even when there are dependencies, you are often creating a complete new set of data, rather than modifying existing data. This allows all your operations to run in parallel without interfering with each others' input data, ensuring you don't run into subtle bugs caused by changes in the order in which operations are applied to your data.
More complex data structures, with all their independencies, can be harder to pick apart and process in such a way.
It's not a bad idea, though, and would be quite a cool way to introduce multi-processor programming into something like RB without bogging people down with all the house-keeping that comes with writing multi-threaded code. Just write one multi-processing scheduler class, and provide some rules about how you write the operations so that they can be parallelised.
-- Kind regards, James Milne _______________________________________________ Unsubscribe or switch delivery mode: <http://www.realsoftware.com/support/listmanager/> Search the archives of this list here: <http://support.realsoftware.com/listarchives/lists.html>
