Quickstart guide#

If you’ve completed the Installation section and were able to run the ASIC Demo, you will have completed a simple containerized run through an ASIC design flow!

In the following sections, you will find more details about the design, the flow and the results of the run.

Where the compilation runs#

The same script runs either against containerized tools or against tools installed natively. One line decides which, and it is worth choosing deliberately before you start:

Docker

Native

Setup

Docker installed and running; nothing else beyond pip install siliconcompiler.

Four EDA tools: Yosys, OpenROAD, OpenSTA and KLayout. See external tools.

Where the tools come from

The sc_runner image – one container per flowgraph node, started and stopped by SiliconCompiler.

Your own machine.

Needs

A network connection the first time, to pull the image.

Nothing; runs offline once the PDK is cached.

Enable with

project.option.scheduler.set_name("docker"), or -scheduler docker.

Nothing – native is the default.

Either way your design never leaves your machine: both modes read and write your working directory and nothing else.

Docker is the quickest way to a first result, which is why the ASIC Demo uses it. Native tools are the better default once you are iterating on a design, and avoid the container startup cost on every node.

Design Details#

The simple design that was used in the demo target is a single clock cycle pulse (“heartbeat”) generated by a counter.

module heartbeat #(
    parameter N = 8
) (
    //inputs
    input      clk,     // clock
    input      nreset,  //async active low reset
    //outputs
    output reg out      //heartbeat
);

    reg [N-1:0] counter_reg;

    always @(posedge clk or negedge nreset) begin
        if (!nreset) begin
            counter_reg <= {(N) {1'b0}};
            out <= 1'b0;
        end else begin
            counter_reg <= counter_reg + 1'b1;
            out <= (counter_reg == {(N) {1'b1}});
        end
    end

endmodule

The clock constraint that goes with it is heartbeat.sdc.

Run Setup#

SiliconCompiler includes a Python API to simplify the hardware compilation flow process. The following code snippet below shows how the demo design was loaded in and run through the Python API.

heartbeat.py (remote run)#
#!/usr/bin/env python3

from siliconcompiler import ASIC, Design                   # import python package
from siliconcompiler.targets import skywater130_demo

if __name__ == "__main__":
    design = Design("heartbeat")                           # create design object
    design.set_topmodule("heartbeat", fileset="rtl")       # set top module
    design.add_file("heartbeat.v", fileset="rtl")          # add input sources
    design.add_file("heartbeat.sdc", fileset="sdc")        # add input sources
    project = ASIC(design)                                 # create project
    project.add_fileset(["rtl", "sdc"])                    # enable filesets
    skywater130_demo(project)                              # load a pre-defined target
    project.option.scheduler.set_name("docker")            # run the tools in containers
    project.run()                                          # run compilation
    project.summary()                                      # print summary
    project.show()                                         # show layout

The following sub-sections will describe each line in more detail.

Project and Design Creation#

The hardware build flow centers around two main objects: the Design, which holds design-specific information, and the ASIC, which manages project settings and execution.

from siliconcompiler import ASIC, Design

design = Design("heartbeat")
project = ASIC(design)

Defining the Design#

Once the objects are created, we specify the design’s top module and add its source files. In this case, heartbeat.v is the Verilog RTL, and heartbeat.sdc is the Synopsys Design Constraints file, which defines the clock.

design.set_topmodule("heartbeat", fileset="rtl")
design.add_file("heartbeat.v", fileset="rtl")
design.add_file("heartbeat.sdc", fileset="sdc")
project.add_fileset(["rtl", "sdc"])

rtl and sdc here are filesets – named groups of input files that a design carries and a run selects among.

Loading a Target#

Next, we load a target, which bundles a Process Design Kit (PDK), standard cell libraries, and a pre-configured compilation flow.

from siliconcompiler.targets import skywater130_demo

skywater130_demo(project)

Configuring the Run#

Project.option is used to configure various settings. This is the line that chooses between the two modes described above:

project.option.scheduler.set_name("docker")   # run the tools in containers

Delete it and the identical script runs the tools installed natively on your own machine instead:

project.option.scheduler.set_name(None)       # use native tools (the default)

Nothing else in the script changes between the two.

Executing the Flow#

Finally, we execute the flow. The Project.run() method starts the compilation, Project.summary() prints a table of results, and Project.show() opens the final layout in a viewer.

project.run()
project.summary()
project.show()

Run Flow#

Running this python script directly produces the same results as the ASIC Demo target.

python3 heartbeat.py

Remote Run Controls#

When your job starts on a remote server, it will log a job ID which you can use to query your job if you close the terminal window or otherwise interrupt the run before it completes:

| INFO    | job0  | remote     | 0  | Your job's reference ID is: 0123456789abcdeffedcba9876543210

The job ID is recorded in the job’s manifest, which is how the sc-remote CLI app identifies the job – point it at that manifest to interact with a running job:

# Check on a job's progress.
sc-remote -cfg build/<design>/<jobname>/<design>.pkg.json

# Cancel a running job.
sc-remote -cfg build/<design>/<jobname>/<design>.pkg.json -cancel

# Ask the server to delete a job from its active records.
sc-remote -cfg build/<design>/<jobname>/<design>.pkg.json -delete

# Reconnect to an active job.
sc-remote -cfg build/<design>/<jobname>/<design>.pkg.json -reconnect

The sc-remote app also accepts a -credentials input parameter which works the same way as the [option,credentials] parameter.

See also

Directory structures explains the build tree these paths refer to, and where else a run writes.

Run Results#

Your run will first show the SiliconCompiler banner/info, followed by design INFO messages.

As the run goes through each step of the flow, a message will be printed to the screen every 5 seconds.

Then, at the end of the run, a summary table will be printed similar to the one shown below. This table is generated by calling the Project.summary() function call in your python script above.

../_images/summary_table.png

All design outputs are located in build/<design>/<jobname> – see Directory structures for what is in there. On a remote run, results are downloaded node by node as each one completes, so when the job finishes your local build directory holds the full results – the tool logs, reports and output files for every step, including the final GDS – along with a screenshot of the finished design, heartbeat.png:

../_images/asic_demo_result.png

Other Ways to Run#

Beyond the two modes compared above, there are two more:

  • Remote – send the job to a server you or your organization operates, for pre-configured tool installations, elastic compute or NDA-protected PDKs.

  • Cluster schedulers – dispatch each node through Slurm, LSF or SGE.

Local Run Results#

By default, only the summary of each step is printed, in order to not clutter up the screen with tool-specific output. If you wish to see the output from each tool, you can find the log files associated with each tool in: build/<design>/<jobname>/<step>/<index>/<step>.log

If you wish to see all the tool-specific information printed onto the screen, you can turn the [option,quiet] option off.

View Design#

For viewing IC layout files (DEF, GDSII) we recommend installing the open source multi-platform Klayout viewer (available for Windows, Linux, and macOS). Installation instructions for Klayout can be found in the tools directory.

If you have Klayout installed, you can browse your completed design by calling sc-show directly from the command line as shown below:

(venv) sc-show -design heartbeat

If you want to have this window pop up automatically at the end of your script, you can add Project.show() to the end of your python script.

project.show()      # pops open a window with the layout

What Next?#

Now that you’ve quickly run a simple example, you can proceed to a larger example like building your own soc, or you can dive deeper into the SiliconCompiler build flow you ran from this quickstart (asic_demo) by looking through how the flow is constructed with the Design and Compilation Data and Compilation Process in the Fundamentals section.