Hi, after some successful tests we have substituted the cell n1_y in tutorial addaccu by our home made inverter. Be believe that this inverter fulfills all requirements stated in 'man sclib' & 'man scr'. We have also created a CATAL file for this new cell. Cell is 18*42 lambdas & has its in & out conectors in ALU2 3 lambdas from cell edges. We've created the .al & .vbe files for this cell with lynx & yagle & verified that the order in which conectors are listed in the .al, .vbe files & LOINS function call are the same. Asimut can simulate the new core & addaccu structures right. The problem is that scr doesn't route (sometimes) the core in the right way. It misses (sometimes) some connections to the conectors of the new cell. These failures seem to depend on the order in what we created those conectors with graal when editing the cell. Furthermore, these failures also seem to depend on cell in & out conector names. This way editing a non-working cell deleting conectors & creating new ones can give a working cell & vice versa. Changing the name of the cell conectors can give the same results. In the addaccu example, the inverter drives 4 mx2_y cells, but when the n1_y cell is substituted by the new one, scr usually routes 2 or none of this conections (as seen in core.ap with graal). Thanks. ----------------------------------------------- - mcesar email: mcesar at sec dot upm dot es - PGP key available - Key Fingerprint: D6 B8 E6 31 D2 CD 55 D5 - B2 81 AB 01 92 9C 06 BC