EDASkel icon indicating copy to clipboard operation
EDASkel copied to clipboard

A skeleton EDA App in C++, featuring design data parsers (using Boost.Spirit), a basic GUI with Qt, a Tcl shell (with non-polling integration with the Qt event loop), a CMake build system, and a testi...

This is EDASkel, a collection of code intended to represent the "skeleton" of an EDA application. This consists of basic example implementations of a number of things you normally need in such an app, such as:

  1. A Tcl shell (with your custom commands installed)
  2. A GUI to display design data
  3. Infrastructure like a build system, parsers for various file formats, interfaces to databases, unit tests, user-controllable logging etc.
  4. Implementations of various well-known algorithms built on top of quality open-source libraries (e.g., Boost Graph Library for graphs, Boost.Polygon or CGAL for computational geometry, possibly some numerics or matrix libraries)

It's still a work in progress but I hope that by sharing what I've learned with others, and learning from them, we can have better software, faster. Please share your feedback with me so I can improve this code.

Thanks and Regards, Jeff Trull [email protected]

Building EDASkel

You need a recent copy of Boost (at least 1.53 - download from www.boost.org) and of Qt (4.6.2). You need to have at least 'program_options' library compiled. The steps are as follows:

(cd to this directory) mkdir build cd build cmake -DCMAKE_PREFIX_PATH=/path/to/BoostConfig.cmake -DEigen3_DIR=/path/to/Eigen3Config.cmake .. make

If the build succeeded you should be able to do:

make test

to run all the unit tests, or try one of the sample apps, for example the design viewer:

apps/sv --lef /path/to/file.lef --def /path/to/file.def

You may find that compiling the LEF/DEF parsers is very slow. It seems to be limited by physical memory size - I recommend at least 4GB, unless you can run the build overnight. They are in a separate module, so once built, they won't change as you experiment with the code, and incremental compiles will be much faster.