First assigment for Special Topics in Programming Languages

Cauã and Bernardo's repo

Idea

This assigment aims to solve the situation proposed at analysis/TELP Key Ideas.pdf. That is, optimize the detection of the nearest ball in a dynamic pool table.

We aim to do this by mixing BVH (bounding volume hierarchies) and ODEs (ordinary differential equations).

Building

Standard compilation procedure (Unix systems)

mkdir -p build
cd build
cmake ..
make -j$(nrpoc)

You will get three binaries: TELP_T1, the main executable; benchmarks/TELP_T1_benchmarks, for mesuring performance; tests/TELP_T1_tests, for mesuring accuracy;

Using GCC on macOS (recommended for std::execution::par_unseq)

Apple Clang uses libc++, which does not implement the C++17 parallel execution policies. To get std::execution::par_unseq (and parallel algorithms in general) working, use GCC from Homebrew:

brew install gcc
mkdir -p build && cd build
CXX=$(brew --prefix gcc)/bin/g++-14 cmake ..   # or g++-13, depending on installed version
make -j$(sysctl -n hw.logicalcpu)
./TELP_T1

Replace g++-14 with your installed version (e.g. g++-13). Check with ls $(brew --prefix gcc)/bin/g++*.

Using the default compiler (Clang on macOS)

With Apple Clang the project will not compile as-is because of std::execution::par_unseq. Use GCC as above, or add a libc++ fallback in src/add_vecs.h (sequential transform when _LIBCPP_VERSION is defined).