This directory contains three types of scripts: (1) TCL scripts, (2) Python scripts and (3) Shell scripts, as detailed below:
- TCL Scripts:
- TCL Scripts:
-[extract_report.tcl](./extract_report.tcl): Contains procedure to extrac metrics (i.e., Core Area, Standard Cell Area, Macro Area, Total Power,Wire Length, WNS, TNS, Congestion) at different stage of P&R in Innovus shell. First source this file in the Innovus shell and then you can use the following commands:
-[extract_report.tcl](./extract_report.tcl): Contains procedure to extract metrics (i.e., Core Area, Standard Cell Area, Macro Area, Total Power, Wire Length, WNS, TNS, Congestion) at different stages of P&R in the Innovus shell. First source this file in the Innovus shell and then you can use the following commands:
-*extract_report preCTS*: Use this command to extract metric after running the *place_opt_design* command.
-*extract_report preCTS*: Use this command to extract metric after running the *place_opt_design* command.
-*extract_report postCTS*: Use this command to extract metric after running the *ccopt_design* command.
-*extract_report postCTS*: Use this command to extract metric after running the *ccopt_design* command.
-*extract_report postRoute*: Use this command to extract metric after running the *routeDesing* or *opt_desing -postRoute* command.
-*extract_report postRoute*: Use this command to extract metric after running the *routeDesing* or *opt_desing -postRoute* command.
-[gen_pb.tcl](./gen_pb.tcl): Contains procedure to write out flat netlist in protocol buffer format in Innovus shell. First source this file in the Innovus shell and then use gen_pb_netlist.
-[gen_pb.tcl](./gen_pb.tcl): Contains procedure to write out flat netlist in protocol buffer format in Innovus shell. First source this file in the Innovus shell and then use gen_pb_netlist.
-*gen_pb_netlist*: This command writes out flat netlist in protobuf format. The output file name is \<top desing\>.pb.txt.
-*gen_pb_netlist*: This command writes out the flat netlist in the protobuf format. The output file name is \<top design\>.pb.txt.
-[pdn_flow.tcl](./pdn_flow.tcl): This script generates power delivery network (PDN) for Innovus implementation. It uses the PDN configuration file available in the [*./Enablements/\**](../../Enablements/) directory.
-[pdn_flow.tcl](./pdn_flow.tcl): This script generates the power delivery network (PDN) for Innovus implementation. It uses the PDN configuration file available in the [*./Enablements/\**](../../Enablements/) directory.
-[place_pin.tcl](./place_pin.tcl): This script places all the top level design ports on the left boundary. Pins are spreaded over 65\% length around the center of the left boundary.
-[place_pin.tcl](./place_pin.tcl): This script places all the top level design ports on the left boundary. Pins are spreaded over 65\% length around the center of the left boundary.
-[write_required_def.tcl](./write_required_def.tcl): This scripts writes out the def and netlist file from innovus. We use this def and netlist are used as inputs to CodeElement to generate the clustered netlist.
-[write_required_def.tcl](./write_required_def.tcl): This script writes out the def and netlist files from the Innovus shell. We use these def and netlist files as inputs to CodeElement to generate the clustered netlist.
- Python Scripts:
- Python Scripts:
-[flow.py](./flow.py): This script runs griding, grouping and clustering to generate the clustered netlist. It requires two inputs:
-[flow.py](./flow.py): This script runs gridding, grouping and clustering to generate the clustered netlist. It requires two inputs:
-*run directory*: Provide the SP&R run directory path.
-*run directory*: Provide the SP&R run directory path.
-*output directory*: Provide the name of the output directory. In this directory the script will write out the clustered netlist.
-*output directory*: Provide the name of the output directory. In this directory the script will write out the clustered netlist.
-[gen_setup.py](./gen_setup.py): This is a helper script of [flow.py](./flow.py). This extracts the required inputs from the rundir to run the CodeElement.
-[gen_setup.py](./gen_setup.py): This is a helper script of [flow.py](./flow.py). This extracts the required inputs from the run directory to run the CodeElement.
-[genJobList.py](./genJobList.py): It create a run directory to run Flow-1 and Flow-2 for each design on all platforms and writes out the job file. You can use this job file to submit a GNU Parallel job.
-[genJobList.py](./genJobList.py): It creates a run directory to run Flow-1 and Flow-2 for each design on all platforms and writes out the job file. You can use this job file to submit a GNU Parallel job.
-**Example**: python ./util/genJobList.py (ensure you are in the [Flows](../) directory.)
-**Example**: python ./util/genJobList.py (ensure you are in the [Flows](../) directory.)
-[plc_pb_to_placement_tcl.py](./plc_pb_to_placement_tcl.py): It writes out the *.plc file from the clustreed-protobuf netlist.
-[plc_pb_to_placement_tcl.py](./plc_pb_to_placement_tcl.py): It writes out the *.plc file from the clustreed-protobuf netlist.
-[shuffle_macro.tcl](./shuffle_macro.tcl): It shuffle same type (having same reference name.) macros. First source this tcl file and then use shuffle_macros command to shuffle the macro location. This script randomly shuffle macros. If macro A moves to position of macro B then the orientation of macro A will be the initial orientation of macro B.
-[shuffle_macro.tcl](./shuffle_macro.tcl): It shuffles the same type (having the same reference name) macros. First source this tcl file and then use the *shuffle_macros* command to shuffle the macro locations. This script randomly shuffles macros. If macro A moves to the position of macro B then the orientation of macro A will be the initial orientation of macro B.
- Shell Scripts:
- Shell Scripts:
-[run_CodeFlow.sh](./run_CodeFlow.sh): This runs the [flow.py](./flow.py) in the run directory to generate clustred netlist.
-[run_CodeFlow.sh](./run_CodeFlow.sh): This runs the [flow.py](./flow.py) in the run directory to generate the clustered netlist.
-**Example**: In the run directory just use *./run_CodeFlow.sh* command to generate the clustred netlist while using Flow-4. Make sure *PHY_SYNTH* is set to 1 or some greater value.
-**Example**: In the run directory just use *./run_CodeFlow.sh* command to generate the clustered netlist while using Flow-4. Make sure *PHY_SYNTH* is set to 1.
-[run_grp.sh](./run_grp.sh): This script generates clustered netlist using [CircuitTraining grouping code](https://github.com/google-research/circuit_training/tree/main/circuit_training/grouping). It uses the following environmental variabels:
-[run_grp.sh](./run_grp.sh): This script generates clustered netlist using [CircuitTraining grouping code](https://github.com/google-research/circuit_training/tree/main/circuit_training/grouping). It uses the following environmental variables:
-*CT_PATH*: Provide the full path of Circuit Training (CT). This path is added to import the grouping module available in the CT repo.
-*CT_PATH*: Provide the full path of Circuit Training (CT). This path is added to import the grouping module available in the CT repo.
-*HMETIS_DIR*: Provide the path of the hMETIS binary directory.
-*HMETIS_DIR*: Provide the path of the hMETIS binary directory.
-*PLC_WRAPPER_MAIN*: Provide the path of the plc_wrapper_main binary.
-*PLC_WRAPPER_MAIN*: Provide the path of the plc_wrapper_main binary.
@@ -95,7 +95,7 @@ We also provide the steps to generate the fakeram models for each of the enablem
...
@@ -95,7 +95,7 @@ We also provide the steps to generate the fakeram models for each of the enablem
## **Flows**
## **Flows**
We provide multiple flows for each of the testcases and enablements. They are: (1) a logical synthesis-based SP&R flow using Cadence Genus and Innovus ([Flow-1](./Flows/figures/flow-1.PNG)), (2) a physical synthesis-based SP&R flow using Cadence Genus iSpatial and Innovus ([Flow-2](./Flows/figures/flow-2.PNG)), (3) a logical synthesis-based SP&R flow using Yosys and OpenROAD ([Flow-3](./Flows/figures/flow-3.PNG)), and (4) creation of input data for Physical synthesis-based Circuit Training using Genus iSpatial ([Flow-4](./Flows/figures/flow-4.PNG)).
We provide multiple flows for each of the testcases and enablements. They are: (1) a logical synthesis-based SP&R flow using Cadence Genus and Innovus ([Flow-1](./Flows/figures/flow-1.PNG)), (2) a physical synthesis-based SP&R flow using Cadence Genus iSpatial and Innovus ([Flow-2](./Flows/figures/flow-2.PNG)), (3) a logical synthesis-based SP&R flow using Yosys and OpenROAD ([Flow-3](./Flows/figures/flow-3.PNG)), and (4) creation of input data for Physical synthesis-based Circuit Training using Genus iSpatial ([Flow-4](./Flows/figures/flow-4.PNG)).
The details of each flow are are given in the following.
The details of each flow are given in the following.
