• Previous message: R. Timothy Edwards: "Magic and GDS2"
• In reply to: R. Timothy Edwards: "Magic and GDS2"
• Next in thread: Philippe O. Pouliquen: "Re: Magic and GDS2"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Tim, the Cadence online reference manual has an appendix on gds
format (run 'openbook' and search for 'calma' or i can send you a PS
version).  also i can give you a sun sparc binary (no source)
(unsupported from cadence) of a program that converts to and from
gds/text.

gds files are big because they represent boxes with 5 coord's (10
reals, counting the duplicated first point) whether using the
'BOUNDARY' or 'BOX' record type (the 2 names are essentially syntactic
sugar, i think).  you could use a PATH element (with a WIDTH) to
represent simple boxes but the space savings is marginal.

i'm confuse about CIF box center points -- must they fall on a .01
grid?  why can' they just fall whereever, for example, on a .005 grid.
it's only the unit size that's defined as .01 um, right?

/jeff

R. Timothy Edwards writes:
 > Dear Andrew,
 > 
 >    I've been working on the magic GDS2 code recently in conjunction
 > with the non-manhattan geometry, and have already fixed a couple of
 > bugs in the GDS2 read/write routines.  However, I agree that there
 > are limitations to the way magic deals with GDS2 labels, and would be
 > happy to take a shot at improving it.  Actually, there's more I'd
 > like to do:  I know that magic produces huge GDS files compared to
 > its CIF output.  I always took this as a sign that GDS was an
 > inferior file format, but on closer inspection, I think magic is
 > writing GDS in about the most non-optimal way possible.  Anyway, if
 > you know of a good reference for GDS syntax (book or online), please
 > let me know.
 > 
 >    The main thing I don't like about GDS is that it's binary, so I
 > can't tweak an output file by hand like I can do with CIF.  CIF has
 > a very well thought-out syntax except for the extreme ugliness of
 > specifying rectangles by center and not by corner.  I think the magic
 > CIF output routine should be re-written to check for rectangle centers
 > falling on fractional values (which is easy enough to do) and opt to
 > write a polygon statement to the output file if so.
 > 
 > 						---Tim

• Next message: Philippe O. Pouliquen: "Re: Magic and GDS2"
• Previous message: R. Timothy Edwards: "Magic and GDS2"
• In reply to: R. Timothy Edwards: "Magic and GDS2"
• Next in thread: Philippe O. Pouliquen: "Re: Magic and GDS2"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]