What is the fastest network graphics engine for large datasets?

Question

Possible duplicate:
What is the fastest network graphics engine for large datasets?

Currently, we have a dynamically updated network diagram containing about 1,500 nodes and 2,000 edges . He is constantly growing. Our current linking mechanism uses Prefuse - specifically, focusing on power orientation - and it takes about 10 minutes with a massive server to get a good, stable linking.

I looked at GraphViz sfpd algorithm a bit, but have not tested it yet ...

Are there any faster alternatives that I should look at?

• I don't care about the look of the nodes and edges — we handle it separately — we just put x, y on the nodes.
• We need to be able to manipulate the layout properties for certain parts of the graph, for example, using special, denser or weaker springs for certain nodes.
• Bonus credit for statistics / benchmarks!

Thank you in advance and ask for comment if you need more specific information to answer!