Simple parallel Roguelike level generation benchmark of Rust, C, D, Go and more. Speed will be measured and compared both between implementations and between them and their non-parallel equivalents, as will standard lines of code (sloc as defined by Github).
Rust is compiled with --opt-level=3
C is compiled with -O3 -lpthread
C++ is compiled with -O3 -pthread -std=c++11
If using Clang++, C++ may need to be compiled with -D__GCC_HAVE_SYNC_COMPARE_AND_SWAP_1 -D__GCC_HAVE_SYNC_COMPARE_AND_SWAP_2 -D__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 -D__GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 , to avoid a bug in less-current versions of the LLVM.
D is compiled with -O -release -inline -noboundscheck
Nimrod is compiled with -d:release, and --threads:on if using the llvm one (the llvm doesn't support OpenMP).
Scala is run with sbt "run seed" from within the scala folder.
They must be run with the seed as a command line parameter, like ./PC 123. PGo uses the seed in the form "PGo -v=seed", rather than just "PGo seed".