StanfordAHA/garnet

Tapeout: Overlapping DTCD cells

Opened this issue · 0 comments

Okay, yes there is a comment in gen_floorplan.tcl that says # The DTCD cells overlap
...but I was not prepared to see fourteen DTCD cells all in the exact same place:

@innovus> foreach d [ get_db insts ifid_dtcd*cc* ] { echo $d }
  inst:GarnetSOC_pad_frame/ifid_dtcd_feol_cc_42
  inst:GarnetSOC_pad_frame/ifid_dtcd_beol_cc_43
  inst:GarnetSOC_pad_frame/ifid_dtcd_beol_cc_44
  ...
  inst:GarnetSOC_pad_frame/ifid_dtcd_beol_cc_55

@innovus> foreach d $dtcds { echo [ get_db $d .bbox ] }             
  {2356.47 3960.0 2361.558 3965.088}
  {2356.47 3960.0 2361.47  3965.0}
  {2356.47 3960.0 2361.47  3965.0}
  ...
  {2356.47 3960.0 2361.47  3965.0}

Actually it's even weirder, now that I look at it; dtcd #42 is slightly larger than the other thirteen cells...?

Does anyone have insight on what's going on here?