Hi Fahad,
As mentioned earlier it takes 10 seconds for 10,000 rows of output. What I think is even when we take top 10,000 the calculations would happen on the entire dataset. So I think it should take not take more than 10 secs for entire dataset as well.
Would like you to try using a reporting tool like Lumira for getting full results.
When you tried it using SQLScript how much time it took?
Does your SQLScript run along with SQL. In that case it could be even more slow.
I tried building a Graphical View for your requirement. Please see if its better in performance..
Product_1(Projection) - Select Invoice Number and Product Name from Table
Product_2(Projection) - Select Invoice Number and Product Name from Table
Join_1(Join) - Inner Join based on Invoice Number
Projection_1(Projection) - Filter based on Product_1!=Product_2 and a Calculated Measure with value '1'
Aggregation_1(Aggregation) - Std Aggregation Node with Product_1, Product_2 and Measure as 'Association'
Table:
Output of Calc View:
Getting repeats in my solution 
, will try to enhance logic to remove those.
Please let me know if this solves the performance issue.
--
Shreepad