hMetis reported: "Wrong cut! 33 32" ???

Hi, I recently use hMetis (version 1.5.3) recursively bi-cut circuit, but it reported this information: "Wrong cut! 33 32".
I cannot find the explaination from the software package instruction. Anybody has any clue about this?

Any help would be appreaciated!


RE: I've seen this before but

I've seen this before but can not remember the context. Check that you do not have any
size one hyperedges, or hyperedges with zero weight, or hyperedges with the same cell being there multiple times.

Can you check if the 2.0 version reports the same problem?

RE: Thanks for your reply. 1) I

Thanks for your reply.

1) I do not have any hyperedges with only one vertex.

2) I set the weight array for hyperedge to be NULL.

3) Actually, the circuit I am using is "adaptec1" from ISPD 05 Placemnt Benchmarks. In this suit of circuits, there are many multi-pin cells. When I applied hMetis, it is quite possible to generate some hyperedges containing the same set of cells. So as you said, maybe this is the reason. However my concern is that, will this error make the whole result crushed/untrust or just locally minor problem? It seems that the hMetis successfully finished except that error.

I was wondering is the library interface for 2.0 version ready? Otherwise I cannot integrate or even try it into my code.