Searching for a better solution

[Helge Klug]

[Jim Lewis]

It is not real clear what you are trying to accomplish.  What generates vNodes and vDataBytesPerNode?  Is it a separate randomization or an iteration?  For each pair of vNodes and vDataBytesPerNode, how many times are you calling your procedure RandCovPoint? 

What is the significance of the coverage bins for data?  Would it be ok to treat data separately from vNodes and vDataBytesPerNode?  This would allow you to have two separate coverage models and simplify the randomization of the data.

I am working on a feature that will address finding particular items within a coverage model and then randomizing, but it is some time off yet.

Knowing more of your problem would help me give better input.

Jim

[Helge Klug]

[Jim Lewis]

Hi Helge,

Is there a reason to loop on nodeN and byteN rather than randomly select them?   Is the data sent in as a single byte or in packets that are byteN long? 

Things that take a long time are often the result of the randomization space being large.  The randomization space will be (gNodes-1)*gDataBytesPerNode*60.  If you have 10 nodes, 16 bytes per note, then you will need to run 9,600 transactions to cover all of the test cases. 

On interesting problems like this, I can do at least 4 hours of initial consulting for free and give you insight into possible approaches (if any).  You can reach me at jim at synthworks.com if you are interested.  

Best Regards,

Jim

[Helge Klug]

Hello Jim, the bytes are send in a package with a size of gNodes * gDataBytesPerNode.
To cover all test cases, I need 60 transactions. For each transaction, each register is written.  #
Thanks for your offer;
I was logged into your course in Munich (2014/02), which was sadly cancelled.
Maybe the course will be repeated and we can discuss the problem directly.
Best regards
Helge

[Author]

Posts