| 3 Apr 2021 |
 Iztok Jeras | I think I got all the components I needed:
additional components from OPENLANE not part of the PDK
.INCLUDE ../../../pdks/open_pdks/sky130/custom/models/short.spice
.INCLUDE ../../../pdks/open_pdks/sky130/custom/models/diode.spice
library of primitives from https://github.com/google/skywater-pdk.git
.LIB ../../../../skywater-pdk/libraries/sky130_fd_pr/latest/models/sky130.lib.spice tt
* include Skywater 130nm cell libraries
.INCLUDE ../../../pdks/sky130A/libs.ref/sky130_fd_sc_hd/spice/sky130_fd_sc_hd.spice | 10:33:10 |
|  shbo changed their display name from Shubho to shbo. | 15:15:10 |
| 4 Apr 2021 |
| Iztok Jeras | Is there a way to get a Standard Delay Format (SDF) file for the synthesis or a later stage verilog netlist? | 11:42:38 |
| 5 Apr 2021 |
 Matt Liberty | Iztok Jeras opensta has a write_sdf command you could try from openroad | 03:04:54 |
|  Matthew Guthaus joined the room. | 16:59:24 |
| Matthew Guthaus | If I set STD_CELL_LIBRARY and re-run "make pdk", it seems that flow.tcl still doesn't get the new library. I'm doing this like:
export STD_CELL_LIBRARY="sky130_fd_sc_ls"
make pdk
make mount
echo $STD_CELL_LIBRARY (not set, so I set it again)
export STD_CELL_LIBRARY="sky130_fd_sc_ls"
./flow.tcl -design spm
...
INFO]: Extracting the number of available metal layers from /home/mrg/openlane/pdks/sky130A/libs.ref/sky130_fd_sc_hd/techlef/sky130_fd_sc_hd.tlef
[ERROR]: during executing: "python3 /openLANE_flow/scripts/extract_metal_layers.py -t /home/mrg/openlane/pdks/sky130A/libs.ref/sky130_fd_sc_hd/techlef/sky130_fd_sc_hd.tlef -o /openLANE_flow/designs/spm/runs/05-04_16-57/tmp/met_layers_list.txt"
[ERROR]: Exit code: 1
[ERROR]: Last 10 lines:
Traceback (most recent call last):
File "/openLANE_flow/scripts/extract_metal_layers.py", line 37, in module
with open(techlef_name, "r") as f:
FileNotFoundError: [Errno 2] No such file or directory: '/home/mrg/openlane/pdks/sky130A/libs.ref/sky130_fd_sc_hd/techlef/sky130_fd_sc_hd.tlef' | 17:01:20 |
| Matthew Guthaus | (make test also fails) | 17:02:06 |
| Matthew Guthaus | It looks like this configuration configuration/general.tcl overwrites the library no matter what:
set ::env(STD_CELL_LIBRARY) "sky130_fd_sc_hd" | 18:26:54 |
 _slack_skywater-pdk_U016ELL4X4M | [efabless/openlane] Issue opened by mguthaus | 18:33:48 |
| Matthew Guthaus | In reply to@_slack_skywater-pdk_U016ELL4X4M:matrix.org
[efabless/openlane] Issue opened by mguthaus I submitted a PR with the fix | 20:39:53 |
| 6 Apr 2021 |
|  Yingwen Tan joined the room. | 00:39:36 |
|  Ernesto Conde joined the room. | 14:15:31 |
| Iztok Jeras | Matt Liberty I was looking at the write_sdf function from OpenSTA, and I can only partially use it to create a SDF file. | 15:49:24 |
| Matt Liberty | Iztok Jeras what do you mean by "partially"? | 15:49:53 |
| Iztok Jeras | I edited the Openlane script [sta.tcl](https://github.com/efabless/openlane/blob/master/scripts/sta.tcl).
At the end of this script I added:
write_sdf $::env(opensta_report_file_tag).sdf
I haven't checked the details, but it appears to populate the min/max part of triples, but typ is left unpopulated.
I looked at the source code for write_sdf and it appears there is no code to populate the typ part of the triple. | 15:51:37 |
| Iztok Jeras | This is the code for writing triples.
https://github.com/The-OpenROAD-Project/OpenSTA/blob/0d73b5b65a185329ea77478d4670d8a103b0b710/sdf/SdfWriter.cc#L412-L449 | 15:56:34 |
| Matthew Guthaus | In reply to@_slack_skywater-pdk_U01EYE54L5N:matrix.org
This is the code for writing triples.
https://github.com/The-OpenROAD-Project/OpenSTA/blob/0d73b5b65a185329ea77478d4670d8a103b0b710/sdf/SdfWriter.cc#L412-L449 It seems that would be an OpenRoad issue then? | 15:57:36 |
| Iztok Jeras | The only way I see to get typical timing into a SDF file is to load the tt liberty file for either min or max or both and redo the export to sdf. | 15:58:32 |
| Iztok Jeras | In reply to@_slack_skywater-pdk_U0175T39732:matrix.org
It seems that would be an OpenRoad issue then? not obviously, since the STA tool itself focuses on min/max timing checks, they might not wish to consume memory for unused typical timing | 15:59:40 |
| Iztok Jeras | In reply to@_slack_skywater-pdk_U01EYE54L5N:matrix.org
not obviously, since the STA tool itself focuses on min/max timing checks, they might not wish to consume memory for unused typical timing but again, there is report_power, which is often observed at typical timing | 16:02:14 |
| Iztok Jeras | In reply to@_slack_skywater-pdk_U01EYE54L5N:matrix.org
but again, there is report_power, which is often observed at typical timing I tried a few ways to get the typical timing out, but I soon get conflicts with previously executed steps. So I decided to try to get some support, before I waste too much time. | 16:09:14 |
|  Harrison joined the room. | 17:09:57 |
| _slack_skywater-pdk_U016ELL4X4M | [efabless/openlane] Issue opened by shafnaskm | 17:50:52 |
| Matthew Guthaus | In reply to@_slack_skywater-pdk_U01EYE54L5N:matrix.org
I tried a few ways to get the typical timing out, but I soon get conflicts with previously executed steps. So I decided to try to get some support, before I waste too much time. It seems that just averaging min/max might be the way to go. This is for a given corner, right? So I'm not sure what typical would be in the context of STA which finds max and min paths... | 19:46:40 |
| 14 Apr 2021 |
|  @trapmunewsdon1971:matrix.org joined the room. | 15:07:59 |
| @trapmunewsdon1971:matrix.org left the room. | 15:09:07 |
| 4 May 2021 |
|  The Wanderer joined the room. | 20:29:28 |
| 13 May 2021 |
|  rencugamob1981 joined the room. | 18:43:48 |
| 20 Jun 2021 |
|  @nagasabah:matrix.org left the room. | 10:12:10 |
| 6 Jul 2021 |
|  Loester F. B. joined the room. | 23:05:13 |
