search

google / skywater-pdk

Open source process design kit for usage with SkyWater Technology Foundry's 130nm node.
https://skywater-pdk.rtfd.io
Apache License 2.0
2.94k stars 382 forks source link

Off-grid standard cells #351

Open d-m-bailey opened 2 years ago

d-m-bailey commented 2 years ago

Expected Behavior

All layout patterns should be on grid.

Actual Behavior

When reading gds into magic, offgrid warnings are generated.

Reading "sky130_fd_sc_hd__a2111o_1".
CIF file read warning: CIF style sky130(vendor): units rescaled by factor of 5 / 1
CIF file read warning: Input off lambda grid by 2/5; snapped to grid.

Steps to Reproduce the Problem

  1. cd $OPENLANE_ROOT
  2. git clone https://github.com/efabless/caravel.git
  3. make mount
  4. cd caravel/openlane
  5. gunzip ../gds/housekeeping.gds.gz
  6. flow.tcl -design housekeeping -lvs -gds ../gds/housekeeping.gds -net ../verilog/gl/housekeeping.v

Specifications

RTimothyEdwards commented 2 years ago

@d-m-bailey You need to be specific---is this the skywater PDK or open_pdks? Because the SkyWater a2111o_1 cell has a DRC error (undetected by Calibre---never understood why) which I fixed, so open_pdks has a wholesale replacement of the GDS for the a2111o_1 cell. If it's the open_pdks that's the issue, then the issue is my fault and it should be raised in the open_pdks repository. Possibly I created an off-grid point when I did that. I would have to take a look at the GDS I created.

RTimothyEdwards commented 2 years ago

@d-m-bailey : I realize that's not an easily answered question. I'm trying to figure it out now.

RTimothyEdwards commented 2 years ago

@d-m-bailey : Aaaaand. . . it's not a2111o_1 that was customized, it was a2111oi_0. So I retract all those statements above and will now go staring at the SkyWater cell to see what's up with it.

RTimothyEdwards commented 2 years ago 
    0x0006     # RECORD_LENGTH              Bytes of data in this record
    0x0d       # RECORD_TYPE:  LAYER        Layer specification
    0x02       # DATA_TYPE:    INTEGER_2    Two byte signed integer
    0x007a     # DATA: 122

    0x0006     # RECORD_LENGTH              Bytes of data in this record
    0x0e       # RECORD_TYPE:  DATATYPE     Datatype specification
    0x02       # DATA_TYPE:    INTEGER_2    Two byte signed integer
    0x0010     # DATA: 16

    0x002c     # RECORD_LENGTH              Bytes of data in this record
    0x10       # RECORD_TYPE:  XY           An array of XY coordinates
    0x03       # DATA_TYPE:    INTEGER_4    Four byte signed integer
    0x000007b7 # DATA: 1975
    0xffffffbd # DATA: -67
    0x00000861 # DATA: 2145
    0xffffffbd # DATA: -67
    0x00000861 # DATA: 2145
    0x00000067 # DATA: 103
    0x000007b7 # DATA: 1975
    0x00000067 # DATA: 103
    0x000007b7 # DATA: 1975
    0xffffffbd # DATA: -67

Sure enough, there's a shape on layer 122:16 (pwell pin) that's completely off-grid. I will prepare another fixed version in open PDKs. Fortunately, it's a pin label layer, not a mask layer, so it won't show up as a manufacturing error.

RTimothyEdwards commented 2 years ago

@d-m-bailey : open_pdks has now been updated with a custom corrected cell layout GDS file.