# I have some trouble when running Openroad tools for foreign pdk. Tool: Initialize Floorplan Subcategory: Floorplan initialization issue ## Conversation ### OuDret Hello, I have been reading documentation and trying stuff with given pdks for about two weeks. My goal was to run OpenRoadFlowScripts but, since there is a lot of stuff, I’ve took its makefile, config files and scripts and just extracted the openroad commands that I really need, so I can exactly know what args and what files i need and create them. Also running the commands 1 by 1 alongside GUI application helps me understanding what’s going on. When calling the commands tho, I had some issues i could not solve. If you have an idea of any of the following problems I'm facing or any advice let me know. **1. I didn’t manage to initialize floorplan with custom die area. It doesn’t matter which values I use as die_are or core_area, it just doesn’t work properly.** When I run something like the command below, having stdSite size defined as 400 by 400 um it just does nothing. The verilog file provided is a simple inverter already synthesized using yosys into a single NAND cell which size is set in .lef as 200 by 200. Screenshot 2023-09-15 101107 Meanwhile, if I just let openroad to decide what size to give it just works for small core utilization value, which kinda makes sense. Screenshot 2023-09-15 101209 Do you think I’m forgetting any step? The commands I run before getting to that point are the following: ``` read_liberty /path/lib/NandCellLibrary.lib read_lef /path/lef/NandCellLibrary.tech.lef read_lef /path/lef/NandCellLibrary.macro.lef read_verilog /path//inverter.v link_design inverter read_sdc /path/constraint.sdc initialize_floorplan -die_area {0 0 10000 10000} -core_area {1000 1000 5000 5000} -site stdSite ``` **2. I have no macros stored in odb despite having them declared in lef/NandCellLibrary.macro.lef** If I just use utilization_area **floorplan_initialization** and keep running the following commands all seems to be working properly ``` make_tracks met1 -x_offset 10 -x_pitch 10 -y_offset 10 -y_pitch 10 make_tracks met2 -x_offset 10 -x_pitch 10 -y_offset 10 -y_pitch 10 make_tracks met3 -x_offset 10 -x_pitch 10 -y_offset 10 -y_pitch 10 remove_buffers place_pins -hor_layer met2 -ver_layer met3 -random global_placement -density 0.5 -pad_left 0 -pad_right 0 ``` Screenshot 2023-09-15 101442 until I try to run macro placement Screenshot 2023-09-15 101654 the error raised means that when running the following tcl procedure it has not found any macro ``` proc find_macros {} { set macros "" set db [ord::get_db] set block [[$db getChip] getBlock] foreach inst [$block getInsts] { set inst_master [$inst getMaster] # BLOCK means MACRO cells if { [string match [$inst_master getType] "BLOCK"] } { append macros " " $inst } } return $macros } ``` the thing is that I have the NAND macro and another one called FILL defined in my .lefs and .lib . Then why could it be that there is no macro in db ? **Some things to be said.** When running **read_liberty**, it prints 1 into terminal. Does it mean that there’s an error? It seems I can keep running things. If so, is there any way to properly debug it? It also happens if I run **check_status** command after reading all input files, which I didn’t even find in documentation, so I just ignored it. Again, if you had any idea or advice of what I’m doing wrong, if I skipt some important step, etc. let me know, please. If i manage to make it work i will try to post the whole flow command by command. It may help someone someday Thank you. ### maliberty What is the size of stdSite? Is this a PDK you can share or is it proprietary? Do you have any instances of the macro in your netlist? ### maliberty ``` SITE stdSite SYMMETRY Y ; CLASS CORE ; SIZE 500 BY 500 ; END stdSite ``` is a huge site (500um x 500um) - I'm guessing that is a mistake though it doesn't explain the problem. ### maliberty Your LEF is missing the units for " DATABASE MICRONS \;". There is a bug when this is unspecified that I'll fix but you should set it (the LEF default value is very old).