A Python tool that extracts partial codes out of a large Fortran application and converts them into a standalone/verifiable/executable kernel

AUTHORS

Youngsung Kim and John Dennis

VERSION

0.7.1

COPYRIGHT

See the document entitled LICENSE.txt

Send questions and comments to KGEN Dev. Team (kgen@ucar.edu).

Documentation

https://ncar.github.io/kgendocs

[ User Interface ]

• Three mandatory arguments(clean, build and run of target application) are added in command line.
• strace, rebuild, prerun options are added

[ Major Improvements ]

• Macro definitions and include paths are automatically generated by KGen
• tprof timing routines are added in generated kernels
• Improved state file generation for intrinsic pointer variables

• KGEN extracts an arbitrary region of Fortran source code as a stand-alone executable software
• In addition, it generates input & output data for executing and verifying the generated kernel
• All KGEN-generated kernels include correctness check and timing measurement

• Linux OS
• Python (>=2.7 and < 3.0)
• C Pre-Processor(cpp)
• Make build tool(make)
• System call trace tool(strace)

The latest KGEN can be obtained from the Git repository.

git clone https://github.com/NCAR/KGen.git

Current KGEN does not require to build or to install.

1. Download from KGen Github repository.

   >> git clone https://github.com/NCAR/KGen.git

2. Read Kgen documentation in "doc" directory under the top Kgen directory.

   >> evince KGENUsersGuideVersion0.7.0.pdf

3. Try a kernel generation example in "examples/simple" directory

   >> cd examples/simple; # move to an example directory

   >> vi src/Makefile; # Modify FC if required

   >> make; # extract a kernel

   >> cd kernel; # move to a kernel directory

   >> make; # build and run a kernel