Full documentation on using/building/running and documentation of the software of the CTF library is available on the following website http://ctf.cs.berkeley.edu as well as on the github wiki. Brief build and execution instructions are given below: To build CTF with a standard setup simply run make in the top directory. To specialize the build configuration (for instance set a custom BLAS library or specify MPI compiler), one must set up a config.mk file. Examples and premade config.mk files for a few architectures are located in the mkfiles/ subdirectory. On running make without a config.mk file in the top folder, mkfiles/config.mk.linux will be copied to config.mk in the top folder by default, unless the hostname is recognized. OpenMP is currently disabled for Apple architectures. The config.mk file will not be overwritten after the first call to 'make'. The first call to 'make' will create the file src/make/make.in, which records the top level directory. So, if you desire to move the built source code, change or delete and recreate this file by running make again. The CTF library is built by default and will be located in lib/. The main interface to the library is include/ctf.hpp. Run 'make examples' to build example programs, see examples/ dir for full list Run 'make test' to run a short test suite, which should confirm the functionality of CTF using a single MPI process. Running 'make test#n' with '#n' in {1,2,3,4,5,6,7,8} with test functionality on #n processors. The number of threads should be set via OMP_NUM_THREADS. All examples can be built individually by the command 'make <example_name>'. Any executables that are built will appear in bin/ The BSD software license for this library is in the file license.txt. Please contact Edgar Solomonik (solomon@cs.berkeley.edu), for any questions about usage or technical implementation of the CTF library.
Edgar Solomonik
authored