|
|
Magic Mailing List |
|
From: Jeff Sondeen (sondeen AT ISI DOT EDU) Date: Fri Feb 27 2004 - 14:49:57 EST
R. Timothy Edwards writes: > Dear Albert, > > > Magic gave me this warning: > > CIF warning for array px36a_0 (instances of px36a): > > cells are so small and close together that two > > non-adjacent cells could have interacting CIF! > > In all the years that I've been doing layout, I have only seen magic > screw up once when giving this error. What it means is that the > cell width is narrower than the largest CIF grow or bloat distance, > so it is possible, in theory, for some layer to be drawn on top of a > cell two places over, where it may violate a design rule with > respect to layers in that cell. > Hi, i think the most common case that magic is 'worried about' is when a cell is so thin that auto-generated bloats of cells on it's sides can "reach thru it" and thus conflict across it, as magic only checks for conflicting bloats on adjacent cells, not on cells close enough to interact but not adjacent. eg, 3 cells, A, B, and C, abut left to right, with B being only 4 lambda wide, and A generating pwell to expand right by 6 lambda. a possible conflict against autogenerated nwell from cell C will not be checked (only the adjacent cell B is checked). in general, quite dangerous (ie. could certainly have a yield impact). /jeff
|
|
|
|